1 2017-12-01 Jakub Jelinek <jakub@redhat.com>
3 * function.h (struct function): Remove cilk_frame_decl,
4 is_cilk_function and calls_cilk_spawn fields.
5 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
7 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
9 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
10 * target.def: Adjust comment.
11 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
12 Don't test cilk_elemental.
14 PR tree-optimization/83233
15 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
16 bswap_stat name for the struct.
19 * tree.h (SWITCH_BREAK_LABEL_P): Define.
20 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
21 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
22 SWITCH_BREAK_LABEL_P set on the label.
23 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
24 added for default case if it was missing and not all cases covered.
25 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
26 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
28 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
29 to avoid -Wimplicit-fallthrough warning.
30 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
31 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
33 PR tree-optimization/83221
34 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
36 (init_reassoc): Formatting fix.
39 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
40 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
44 * tree-cfg.c: Include asan.h.
45 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
47 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
49 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
51 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
53 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
56 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
58 * config/i386/i386.c (standard_sse_constant_opcode): Fix
59 registers type for 128bit mode.
61 2017-11-30 Jakub Jelinek <jakub@redhat.com>
63 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
65 * gengtype-state.c (read_state_pair): Likewise.
66 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
67 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
68 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
69 * ifcvt.c (noce_try_store_flag_constants): Likewise.
70 * tree-ssa-ccp.c (ccp_finalize): Likewise.
71 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
72 * builtins.c (fold_builtin_3): Likewise.
73 * graphite-scop-detection.c
74 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
75 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
78 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
81 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
82 parameter. Return true if ignore_decl_rtl_set_p is true and other
83 conditions are satisfied.
84 * asan.h (asan_protect_global): Add new parameter.
85 * varasm.c (categorize_decl_for_section): Pass true as second parameter
86 to asan_protect_global calls.
88 2017-11-30 Jim Wilson <jimw@sifive.com>
90 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
91 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
92 -mstrict-align, add info on default value. Delete redundant lines for
93 -mabi. Add missing -mexplicit-relocs docs.
95 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
97 * config/arc/arc.md (trap): New pattern.
99 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
101 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
102 instruction to end into a delay slot.
103 * config/arc/arc.md (cond_delay_insn): Check if the instruction
104 can be placed into a delay slot against reg_note.
106 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
108 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
109 labels number of usages.
111 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
113 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
115 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
117 2017-11-30 Jakub Jelinek <jakub@redhat.com>
120 * internal-fn.c (expand_mul_overflow): Optimize unsigned
121 multiplication by power of 2 constant into two shifts + comparison.
123 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
126 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
128 2017-11-30 Richard Biener <rguenther@suse.de>
130 PR tree-optimization/83202
131 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
132 allow_peel argument and guard peeling.
133 (canonicalize_loop_induction_variables): Likewise.
134 (canonicalize_induction_variables): Pass false.
135 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
136 peeling from cunrolli.
138 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
140 * combine.c (try_combine): Print a message to dump file whenever
141 I0, I1, or I2 cannot be combined into I3.
143 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
145 PR rtl-optimization/83156
146 PR rtl-optimization/82621
147 * combine.c (try_combine): Don't split an I2 if one of the dests is
148 set again before I3. Allow unused dests.
150 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
152 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
154 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
156 PR rtl-optimization/80818
157 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
158 recursively. Use insn code for clobber.
159 (lra_set_insn_recog_data): Pass the new arg to
160 collect_non_operand_hard_regs.
161 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
163 (lra_update_insn_regno_info): Pass insn to
164 add_regs_to_insn_regno_info.
166 2017-11-29 Jim Wilson <jimw@sifive.com>
167 Andrew Waterman <andrew@sifive.com>
169 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
170 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
171 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
172 use SINGLE_SHIFT_COST.
173 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
174 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
176 2017-11-29 Julia Koval <julia.koval@intel.com>
178 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
179 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
180 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
181 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
182 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
183 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
184 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
185 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
186 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
187 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
188 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
189 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
190 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
191 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
192 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
193 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
194 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
195 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
196 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
197 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
198 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
199 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
200 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
201 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
202 __builtin_ia32_vpshldv_v2di_mask,
203 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
204 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
205 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
207 2017-11-29 Julia Koval <julia.koval@intel.com>
209 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
210 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
211 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
212 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
213 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
214 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
215 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
216 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
217 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
218 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
219 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
220 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
221 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
222 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
223 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
224 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
225 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
226 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
227 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
228 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
229 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
231 2017-11-29 Daniel Cederman <cederman@gaisler.com>
233 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
234 movsi_pic_gotdata_op instruction as a load for the UT699 errata
237 2017-11-29 Martin Aberg <maberg@gaisler.com>
239 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
240 to prevent b2bst errata sequence.
241 (sqrtdf2_fix): Likewise.
243 2017-11-29 Daniel Cederman <cederman@gaisler.com>
245 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
246 (div_sqrt_insn_p): New function.
247 (sparc_do_work_around_errata): Insert NOP instructions to
248 prevent sequences that could trigger the TN-0013 errata for
249 certain LEON3 processors.
250 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
251 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
252 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
253 (in_branch_delay): Prevent div and sqrt in delay slot if
255 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
257 2017-11-29 Daniel Cederman <cederman@gaisler.com>
259 * config/sparc/sparc.c (atomic_insn_p): New function.
260 (sparc_do_work_around_errata): Insert NOP instructions to
261 prevent sequences that could trigger the TN-0010 errata for
263 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
264 instruction referable in atomic_insns_p.
266 2017-11-29 Daniel Cederman <cederman@gaisler.com>
268 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
269 (atomic_compare_and_swap_leon3_1): Likewise.
272 2017-11-29 Daniel Cederman <cederman@gaisler.com>
274 * config/sparc/sparc.c (fpop_insn_p): New function.
275 (sparc_do_work_around_errata): Insert NOP instructions to
276 prevent sequences that could trigger the TN-0012 errata for
278 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
279 * config/sparc/sparc.md (fix_gr712rc): New attribute.
280 (in_branch_annul_delay): Prevent floating-point instructions
281 in delay slot of annulled integer branch.
283 2017-11-29 Richard Biener <rguenther@suse.de>
285 PR tree-optimization/83202
286 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
288 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
289 nodes vectorized to the same stmts multiple times.
290 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
291 (vect_analyze_slp_instance): Adjust.
292 (scalar_stmts_to_slp_tree_map_t): New typedef.
293 (vect_schedule_slp_instance): Add a map recording the SLP node
294 representing the vectorized stmts for a set of scalar stmts.
295 Avoid code-generating redundancies.
296 (vect_schedule_slp): Allocate map and pass it down.
298 2017-11-29 Nathan Sidwell <nathan@acm.org>
301 * tree.c (build_complex_type): Fix canonicalization. Only fill in
304 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
306 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
307 cases for printing LDP/STP memory addresses.
308 (aarch64_print_address_internal): Renamed from
309 aarch64_print_operand_address, added parameter, add Pmode check.
310 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
311 (aarch64_print_operand_address): Indirect to
312 aarch64_print_address_internal.
313 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
314 'y' operand output specifier.
316 2017-11-29 Jakub Jelinek <jakub@redhat.com>
319 * tree.c (build_simple_mem_ref_loc): Handle
320 get_addr_base_and_unit_offset returning a MEM_REF.
323 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
326 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
329 2017-11-28 David Malcolm <dmalcolm@redhat.com>
331 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
333 (layout::move_to_column): Likewise.
335 2017-11-28 Jakub Jelinek <jakub@redhat.com>
338 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
339 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
341 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
342 Martin Jambor <mjambor@suse.cz>
345 * tree.h (expr_type_first_operand_type_p): Declare
346 * tree.c (expr_type_first_operand_type_p): New function.
347 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
348 (ipa_value_from_jfunc): Adjust declaration.
349 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
350 Use it as result type for arithmetics, unless it is NULL in which case
351 be more conservative.
352 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
353 ipa_get_jf_pass_through_result.
354 (propagate_vals_across_pass_through): Likewise.
355 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
356 is to propagate_vals_across_pass_through.
357 (propagate_constants_across_call): Pass PARM_TYPE to
358 propagate_scalar_across_jump_function.
359 (find_more_scalar_values_for_callers_subset): Pass parameter type to
360 ipa_value_from_jfunc.
361 (cgraph_edge_brings_all_scalars_for_node): Likewise.
362 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
363 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
364 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
365 target_type, pass it to ipa_value_from_jfunc.
366 (update_indirect_edges_after_inlining): Pass parameter type to
367 try_make_edge_direct_simple_call.
369 2017-11-28 Jeff Law <law@redhat.com>
371 * gimple-ssa-evrp-analyze.c
372 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
373 refine ranges if scev_initialized_p returns true.
374 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
376 2017-11-28 Julia Koval <julia.koval@intel.com>
378 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
379 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
380 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
381 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
382 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
383 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
384 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
385 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
386 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
387 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
388 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
389 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
390 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
391 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
392 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
393 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
394 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
395 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
396 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
397 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
398 __builtin_ia32_vpshrd_v2di_mask): New builtins.
399 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
401 2017-11-28 Julia Koval <julia.koval@intel.com>
403 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
404 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
405 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
406 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
407 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
408 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
409 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
410 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
411 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
412 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
413 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
414 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
415 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
416 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
417 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
418 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
419 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
420 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
421 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
422 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
423 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
424 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
425 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
426 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
427 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
428 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
429 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
430 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
431 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
432 __builtin_ia32_vpshld_v2di_mask): New builtins.
433 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
434 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
436 2017-11-28 Richard Biener <rguenther@suse.de>
438 PR tree-optimization/80776
439 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
441 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
443 (evrp_range_analyzer::record_ranges_from_incoming_edges):
444 If the incoming edge is an effective fallthru because the other
445 edge only reaches a __builtin_unreachable () then record ranges
446 derived from the controlling condition in SSA info.
447 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
448 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
450 2017-11-28 Olivier Hainque <hainque@adacore.com>
452 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
453 on mingw build hosts.
455 2017-11-28 Jakub Jelinek <jakub@redhat.com>
458 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
460 2017-11-28 Richard Biener <rguenther@suse.de>
463 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
464 copies generated from memcpy use a character array as reference
467 2017-11-28 Julia Koval <julia.koval@intel.com>
468 Sebastian Peryt <sebastian.peryt@intel.com>
470 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
471 c-family/c-cilkplus.o, c-family/array-notation-common.o,
472 cilk-common.o, cilk.h, cilk-common.c): Remove.
474 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
475 * builtins.c (is_builtin_name): Remove cilkplus condition.
476 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
477 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
478 cilk-builtins.def, cilkplus.def): Remove.
479 * cif-code.def (CILK_SPAWN): Remove.
480 * cilk-builtins.def: Delete.
481 * cilk-common.c: Ditto.
483 * cilkplus.def: Ditto.
484 * config/darwin.h (fcilkplus): Delete.
485 * cppbuiltin.c: Ditto.
486 * doc/extend.texi: Remove cilkplus doc.
487 * doc/generic.texi: Ditto.
488 * doc/invoke.texi: Ditto.
489 * doc/passes.texi: Ditto.
490 * gcc.c (fcilkplus): Remove.
491 * gengtype.c (cilk.h): Remove.
492 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
494 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
496 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
498 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
499 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
501 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
502 inline_read_section): Ditto.
503 * ipa-inline-analysis.c (cilk.h): Remove.
504 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
505 * lto-wrapper.c (merge_and_complain, append_compiler_options,
506 append_linker_options): Remove condition for fcilkplus.
507 * lto/lto-lang.c (cilk.h): Remove.
508 (lto_init): Remove condition for fcilkplus.
509 * omp-expand.c (expand_cilk_for_call): Delete.
510 (expand_omp_taskreg, expand_omp_for_static_chunk,
511 expand_omp_for): Remove cilkplus
513 (expand_cilk_for): Delete.
514 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
515 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
516 execute_lower_omp, diagnose_sb_0): Ditto.
517 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
518 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
519 * tree-nested.c: Ditto.
520 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
521 (dump_generic_node): Ditto.
522 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
523 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
525 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
527 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
529 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
530 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
531 for vector mode and !TARGET_SIMD.
533 2017-11-28 Jakub Jelinek <jakub@redhat.com>
535 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
537 * tree.h (SWITCH_LABELS): Remove.
538 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
539 assert SWITCH_BODY is non-NULL.
540 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
542 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
544 PR tree-optimization/80788
545 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
546 has TREE_OVERFLOW set, call drop_tree_overflow.
548 2017-11-28 Richard Biener <rguenther@suse.de>
550 PR tree-optimization/83158
551 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more
554 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
557 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
558 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
560 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
562 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
563 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
565 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
568 * varasm.c (compare_constant): Take the mode of the constants into
569 account when comparing floating point constants.
571 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
573 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
574 from explicit instantiation of debug_helper.
575 * vec.h (DEFINE_DEBUG_VEC): Ditto.
577 2017-11-27 Richard Biener <rguenther@suse.de>
579 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
582 2017-11-27 Richard Biener <rguenther@suse.de>
584 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
585 (type_hash_canon): Also clear int_cst_hash_table entry for
587 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
589 2017-11-27 Tamar Christina <tamar.christina@arm.com>
591 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
592 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
593 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
595 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
597 * hash-map.h (gt_cleare_cache): Avoid UB.
599 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
601 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
602 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
603 (decide_unroll_runtime_iterations): Likewise.
604 (decide_unroll_stupid): Likewise.
606 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
609 * config/i386/i386.md: Add a loop with incssp.
611 2017-11-27 Martin Jambor <mjambor@suse.cz>
613 PR tree-optimization/81248
614 * tree-sra.c (splice_param_accesses): Remove size check.
615 (decide_one_param_reduction): Fix size check.
616 * gimple-pretty-print.c (dump_profile): Silence warning.
617 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
619 2017-11-27 Jakub Jelinek <jakub@redhat.com>
622 * dbxout.c (lastlineno): New variable.
623 (dbx_debug_hooks): Use dbxout_switch_text_section as
624 switch_text_section debug hook.
625 (dbxout_function_end): Switch to current_function_section
626 rather than function_section. If crtl->has_bb_partition,
627 output just one N_FUN, depending on in_cold_section_p.
628 (dbxout_source_line): Remember last lineno in lastlineno.
629 (dbxout_switch_text_section): New function.
630 (dbxout_function_decl): Adjust dbxout_block caller.
631 (dbx_block_with_cold_children): New function.
632 (dbxout_block): Return true if any LBRAC/RBRAC have been
633 emitted. Use dbx_block_with_cold_children at depth == 0
634 in second partition. Add PARENT_BLOCKNUM argument, pass
635 it optionally adjusted to children. Output LBRAC/RBRAC
636 around recursive call only if the block is in the current
637 partition, if not and anything was output, emit empty
639 * final.c (final_scan_insn): Compute cold_function_name
640 before calling switch_text_section debug hook. Call
641 that hook even if dwarf2out_do_frame if not emitting
645 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
648 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
650 PR rtl-optimization/82488
651 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
653 2017-11-26 Julia Koval <julia.koval@intel.com>
655 * config/i386/i386.c (processor_target_table): Add skylake_cost for
657 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
660 2017-11-26 Julia Koval <julia.koval@intel.com>
662 * config/i386/driver-i386.c (host_detect_local_cpu):
663 Detect skylake-avx512.
665 2017-11-26 Julia Koval <julia.koval@intel.com>
667 * config.gcc: Add -march=cannonlake.
668 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
669 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
670 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
671 (PTA_CANNONLAKE): New.
672 (processor_target_table): Add cannonlake.
673 (ix86_option_override_internal): Ditto.
674 (fold_builtin_cpu): Ditto.
675 (get_builtin_code_for_version): Handle cannonlake.
676 (M_INTEL_COREI7_CANNONLAKE): New.
677 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
678 * doc/invoke.texi: Add -march=cannonlake.
680 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
682 * plugin.c (add_new_plugin): Use platform-specific library extensions.
683 (try_init_one_plugin): Alternative implementation for MinGW.
684 * Makefile.in (plugin_implib): New.
685 (gengtype-lex.c): Fix broken AIX workaround.
686 * configure: Regenerate.
687 * doc/plugins.texi: Document support for MinGW.
689 2017-11-25 Jakub Jelinek <jakub@redhat.com>
691 PR rtl-optimization/81553
692 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
693 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
694 is a shift where C1 has different mode than the whole shift, use C1's
695 mode for MULT rather than the shift's mode.
698 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
699 builtins not enabled in the currently selected ISA.
701 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
703 PR tree-optimization/71026
704 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
705 (insert_reciprocals): Change to insert reciprocals before a division
706 by a square and to insert the square of a reciprocal.
707 (execute_cse_reciprocals_1): Change to consider division by a square.
708 (register_division_in): Add importance parameter.
710 2017-11-24 Richard Biener <rguenther@suse.de>
712 PR tree-optimization/82402
713 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
714 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
716 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
718 * match.pd (0-ptr): New transformation.
720 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
723 * ipa-inline.c (inline_small_functions): Set current badnes correctly
724 when skipping checking.
726 2017-11-24 Richard Biener <rguenther@suse.de>
728 PR tree-optimization/83128
729 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
730 (vn_reference_lookup_3): Likewise.
732 2017-11-24 Jakub Jelinek <jakub@redhat.com>
735 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
736 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
739 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
740 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
741 check for the latter with gimple_call_builtin_p. Do not handle
742 BUILT_IN_STPNCPY_CHK which is not a pass through call.
744 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
746 * config/arm/arm_neon.h: Fix pragma GCC push_options before
749 2017-11-23 Julia Koval <julia.koval@intel.com>
751 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
752 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
753 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
754 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
755 _mm512_maskz_expandloadu_epi16): New intrinsics.
756 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
757 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
758 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
759 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
760 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
761 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
762 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
763 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
764 _mm256_maskz_expandloadu_epi8): New intrinsics.
765 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
766 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
767 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
768 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
769 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
770 * config/i386/sse.md (VI248_VLBW): New iterator.
771 (expand<mode>_mask, expand<mode>_maskz): New patterns.
773 2017-11-23 Julia Koval <julia.koval@intel.com>
775 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
776 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
777 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
778 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
779 _mm512_mask_compressstoreu_epi16): New.
780 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
781 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
782 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
783 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
784 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
785 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
786 _mm256_mask_compressstoreu_epi8): New.
787 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
788 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
789 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
790 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
791 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
792 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
793 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
794 __builtin_ia32_compresshi128_mask,
795 __builtin_ia32_compressstoreuqi512_mask,
796 __builtin_ia32_compressstoreuhi512_mask,
797 __builtin_ia32_compressstoreuqi256_mask,
798 __builtin_ia32_compressstoreuqi128_mask,
799 __builtin_ia32_compressstoreuhi256_mask,
800 __builtin_ia32_compressstoreuhi128_mask): New builtins.
801 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
803 (ix86_expand_special_args_builtin): Handle new types.
804 (s4fma_expand): Handle new builtin array.
805 * config/i386/immintrin.h: Include new headers.
806 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
807 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
809 2017-11-23 Jakub Jelinek <jakub@redhat.com>
811 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
812 fixes. Declare temp and g variables at the top in order to avoid
813 {} in most of the cases.
815 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
817 * match.pd (ptr-0): New transformation.
819 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
821 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
822 (struct addr_mode_cost_table): New.
823 (struct tune_params): Add field addr_mode_costs.
824 * config/arm/arm.c (generic_addr_mode_costs): New.
825 (arm_slowmul_tune): Initialise addr_mode_costs field.
826 (arm_fastmul_tune): Likewise.
827 (arm_strongarm_tune): Likewise.
828 (arm_xscale_tune): Likewise.
829 (arm_9e_tune): Likewise.
830 (arm_marvell_pj4_tune): Likewise.
831 (arm_v6t2_tune): Likewise.
832 (arm_cortex_tune): Likewise.
833 (arm_cortex_a8_tune): Likewise.
834 (arm_cortex_a7_tune): Likewise.
835 (arm_cortex_a15_tune): Likewise.
836 (arm_cortex_a35_tune): Likewise.
837 (arm_cortex_a53_tune): Likewise.
838 (arm_cortex_a57_tune): Likewise.
839 (arm_exynosm1_tune): Likewise.
840 (arm_xgene1_tune): Likewise.
841 (arm_cortex_a5_tune): Likewise.
842 (arm_cortex_a9_tune): Likewise.
843 (arm_cortex_a12_tune): Likewise.
844 (arm_cortex_a73_tune): Likewise.
845 (arm_v7m_tune): Likewise.
846 (arm_cortex_m7_tune): Likewise.
847 (arm_v6m_tune): Likewise.
848 (arm_fa726te_tune): Likewise.
849 (arm_mem_costs): Use table lookup to calculate cost of addressing
852 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
854 * config/arm/arm.c (arm_mem_costs): New function.
855 (arm_rtx_costs_internal): Use arm_mem_costs.
857 2017-11-23 Mark Wielaard <mark@klomp.org>
859 * dwarf2out.c (init_sections_and_labels): Use generation to create
860 unique ranges_section_label and ranges_base_label. Return generation.
861 (output_rnglists): Add generation argument. Use generation to create
862 unique ranges labels.
863 (dwarf2out_finish): Get generation from init_sections_and_labels
864 and pass generation to output_rnglists.
866 2017-11-23 Mike Stump <mikestump@comcast.net>
867 Eric Botcazou <ebotcazou@adacore.com>
869 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
870 * cfgloop.h (struct loop): Add unroll field.
871 * function.h (struct function): Add has_unroll bitfield.
872 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
873 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
874 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
875 (pass_rtl_unroll_loops::gate): Likewise.
876 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
877 for which loop->unroll==1.
878 (decide_unroll_constant_iterations): Use note for consistency and
879 take loop->unroll into account. Return early if loop->unroll is set.
880 Fix thinko in existing test.
881 (decide_unroll_runtime_iterations): Use note for consistency and
882 take loop->unroll into account.
883 (decide_unroll_stupid): Likewise.
884 * lto-streamer-in.c (input_cfg): Read loop->unroll.
885 * lto-streamer-out.c (output_cfg): Write loop->unroll.
886 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
888 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
889 (print_loop): Print loop->unroll if set.
890 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
891 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
892 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
894 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
895 loop->unroll is set and smaller than the trip count. Otherwise bypass
896 entirely the heuristics if loop->unroll is set. Remove dead note.
897 Fix off-by-one bug in other note.
898 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
899 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
901 (tree_unroll_loops_completely): Make static.
902 (pass_complete_unroll::execute): Use correct type for variable.
903 (pass_complete_unrolli::execute): Fix formatting.
904 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
906 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
908 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
909 enable when TARGET_PREFER_AVX128 is set.
911 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
913 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
915 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
916 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
918 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
920 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
921 frequencies. Fix estimation of aggregate parameters.
923 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
925 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
928 2017-11-23 Tom de Vries <tom@codesourcery.com>
930 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
932 2017-11-23 Tom de Vries <tom@codesourcery.com>
934 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
937 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
940 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
941 sibcall_value_pcrel_fdpic): Use local variable instead of
943 (calli_tbr_rel): Add missing operand 2.
944 (call_valuei_tbr_rel): Add missing operand 3.
946 2017-11-23 Jakub Jelinek <jakub@redhat.com>
949 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
950 bitpos/bitsize covering the whole destination, use store_expr only if
951 the complex mode is the same. Otherwise, use expand_normal and if
952 it returns CONCAT, subreg each part separately instead of trying to
953 subreg the whole result.
955 2017-11-23 Richard Biener <rguenther@suse.de>
957 PR tree-optimization/23094
958 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
959 come here from walking over backedges in the first iteration.
960 (vn_reference_lookup_3): Skip clobbers that store the same value.
962 2017-11-23 Richard Biener <rguenther@suse.de>
964 PR tree-optimization/81403
965 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
966 a block we need a leader relative to.
967 (phi_translate_1): For nary processing require a leader from
968 get_representative_for given we run expression simplification
969 using match-and-simplify. Remove previous fix.
971 2017-11-22 Jeff Law <law@redhat.com>
973 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
974 Use new method allocate_value_range rather than accessing the
975 vrp_value_range_pool data member directly.
976 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
977 to use extract_range_from_stmt method to avoid need for
978 extract_range_from_assignment method.
979 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
980 method rather than setting values_propgated data member directly.
981 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
982 and values propagated data members and extract_range_from_assignment
983 method. Reorder private data members to conform to standards.
984 Add new methods set_lattice_propagation_complete and
985 allocate_value_range.
987 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
989 PR rtl-optimization/83030
990 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
991 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
992 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
993 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
994 CROSSING_JUMP_P flag is already set before setting it.
995 * cfgrtl.c (fixup_partition_crossing): Likewise.
996 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
999 2017-11-22 Jakub Jelinek <jakub@redhat.com>
1001 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
1002 Handle the case where both arguments are using gen_const_vec_series.
1004 2017-11-22 David Malcolm <dmalcolm@redhat.com>
1007 * pretty-print.c (pp_format): Move quoting implementation to
1008 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
1009 to pass address of "quote" local.
1010 (pp_begin_quote): New function.
1011 (pp_end_quote): New function.
1012 * pretty-print.h (printer_fn): Convert penultimate param from bool
1014 (pp_begin_quote): New decl.
1015 (pp_end_quote): New decl.
1016 * tree-diagnostic.c (default_tree_printer): Convert penultimate
1017 param from bool to bool *.
1018 * tree-diagnostic.h (default_tree_printer): Likewise.
1020 2017-11-22 Jeff Law <law@redhat.com>
1022 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
1023 Initialize vr_values.
1024 (evrp_range_analyzer::try_find_new_range): Call methods attached to
1025 vr_values via vr_values class instance rather than delegators.
1026 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
1027 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
1028 (evrp_range_analyzer::push_value_range): Likewise.
1029 (evrp_range_analyzer::pop_value_range): Likewise.
1030 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
1031 most delegators. Those remaining are exposed as public interfaces.
1032 Make vr_values a pointer and private.
1033 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
1035 (evrp_range_analyzer::get_vr_value): New method.
1036 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
1037 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
1038 (class evrp_dom_walker): Attach evrp_folder class, initialize
1039 it in the ctor. Remove temporary delegators.
1040 (evrp_dom_walker::before_dom_children): Call methods in attached
1041 evrp_range_analyzer class via class instance pointer. Use
1042 free value_range_constant_singleton to remove need for
1043 op_with_constant_singleton_value delegator method. Do not
1044 create a vrp_prop class instance for every call! Narrow
1045 scope of a couple variables.
1046 (evrp_dom_walker::cleanup): Call methods in attached
1047 evrp_range_analyzer class via class instance pointer.
1048 * vr-values.h (class vr_values): Privatize many methods and
1051 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1053 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
1054 * cfgexpand.c (expand_debug_expr): Remove handling for them.
1055 * expr.c (expand_expr_real_2): Likewise.
1056 * fold-const.c (const_unop): Likewise.
1057 * optabs-tree.c (optab_for_tree_code): Likewise.
1058 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
1059 * tree-inline.c (estimate_operator_cost): Likewise.
1060 * tree-pretty-print.c (dump_generic_node): Likewise.
1061 (op_code_prio): Likewise.
1062 (op_symbol_code): Likewise.
1063 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
1064 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
1065 * internal-fn.c (direct_internal_fn_optab): New function.
1066 (direct_internal_fn_array, direct_internal_fn_supported_p
1067 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
1068 * fold-const-call.c (fold_const_reduction): New function.
1069 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
1071 * tree-vect-loop.c: Include internal-fn.h.
1072 (reduction_code_for_scalar_code): Rename to...
1073 (reduction_fn_for_scalar_code): ...this and return an internal
1075 (vect_model_reduction_cost): Take an internal_fn rather than
1077 (vect_create_epilog_for_reduction): Likewise. Build calls rather
1079 (vectorizable_reduction): Use internal functions rather than tree
1080 codes for the reduction operation. Update calls to the functions
1082 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
1083 Use calls to internal functions rather than REDUC tree codes.
1084 * config/aarch64/aarch64-simd.md: Update comment accordingly.
1086 2017-11-22 Olivier Hainque <hainque@adacore.com>
1088 * config/vxworks.c (vxworks_override_options): Pick default
1089 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
1090 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
1091 DWARF_GNAT_ENCODINGS_DEFAULT.
1092 * config/vxworksae.h: Likewise.
1094 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
1096 PR tree-optimization/83104
1097 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
1100 2017-11-22 Marek Polacek <polacek@redhat.com>
1101 H.J. Lu <hongjiu.lu@intel.com>
1102 Jason Merrill <jason@redhat.com>
1107 * calls.c (initialize_argument_information): Call
1108 warn_parameter_passing_abi target hook.
1109 (store_one_arg): Use 0 for empty record size. Don't push 0 size
1110 argument onto stack.
1111 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
1112 * common.opt: Update -fabi-version description.
1113 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
1114 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
1116 (ix86_is_empty_record): New function.
1117 (ix86_warn_parameter_passing_abi): New function.
1118 (TARGET_EMPTY_RECORD_P): Redefine.
1119 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
1120 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
1121 * doc/tm.texi: Regenerated.
1122 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
1123 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
1124 * dwarf2out.c (get_ultimate_context): Move to tree.c.
1125 * explow.c (hard_function_value): Call arg_int_size_in_bytes
1126 instead of int_size_in_bytes.
1127 * expr.c (copy_blkmode_to_reg): Likewise.
1128 * function.c (aggregate_value_p): Return 0 for empty types.
1129 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi target hook.
1130 (locate_and_pad_parm): Call arg size_in_bytes instead
1132 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
1133 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
1134 * target.def (empty_record_p, warn_parameter_passing_abi): New target
1136 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
1137 (std_gimplify_va_arg_expr): Skip empty records. Call
1138 arg_size_in_bytes instead size_in_bytes.
1139 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
1140 * tree-core.h (tree_type_common): Add empty_flag.
1141 (tree_decl_common): Update comments.
1142 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
1144 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
1145 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
1147 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
1148 * tree.c (default_is_empty_type): New function.
1149 (default_is_empty_record): New function.
1150 (arg_int_size_in_bytes): New function.
1151 (arg_size_in_bytes): New function.
1152 (get_ultimate_context): New function.
1153 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
1154 TRANSLATION_UNIT_WARN_EMPTY_P.
1155 (default_is_empty_record, arg_int_size_in_bytes,
1156 arg_size_in_bytes, get_ultimate_context): Declare.
1158 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1160 * config/arm/arm.c (cmse_clear_registers): New function.
1161 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
1162 code by call to cmse_clear_registers.
1163 (cmse_nonsecure_entry_clear_before_return): Likewise.
1165 2017-11-22 Tamar Christina <tamar.christina@arm.com>
1167 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
1168 (vdot_s32, vdotq_s32): New.
1169 (vdot_lane_u32, vdotq_lane_u32): New.
1170 (vdot_lane_s32, vdotq_lane_s32): New.
1173 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1176 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
1177 for unsigned values with fewer HWIs than the precision.
1178 (test_overflow): New function.
1179 (wide_int_cc_tests): Call it.
1181 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1183 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
1184 has the same mode class as Pmode.
1186 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1188 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
1189 padding_bits_to_clear_ptr.
1190 (cmse_nonsecure_entry_clear_before_return): Likewise.
1192 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1194 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
1195 auto_sbitap instead of integer bitfield to control register needing
1198 2017-11-22 Jakub Jelinek <jakub@redhat.com>
1200 PR tree-optimization/83044
1201 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
1202 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
1203 that need the upper bound. Subtract offset from
1204 get_addr_base_and_unit_offset only if positive and subtract it
1205 before division by eltsize rather than after it.
1208 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
1209 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
1210 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
1214 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
1215 expand_binop with *mul_widen_optab, make sure at least one of the
1216 operands doesn't have VOIDmode.
1219 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
1221 PR rtl-optimization/82044
1222 PR tree-optimization/82042
1223 * dse.c (record_store): Check for overflow.
1224 (check_mem_read_rtx): Properly check for overflow if width == -1, call
1225 add_wild_read instead of clear_rhs_from_active_local_stores on
1226 overflow and log it into dump_file.
1228 2017-11-22 Richard Biener <rguenther@suse.de>
1230 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
1231 fake edges to exit when looking for a place to insert.
1232 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
1234 (insert_into_preds_of_block): Commit edge insertion immediately,
1235 assert that doesn't require new BBs.
1236 (fini_pre): Release expressions.
1237 (pass_pre::execute): Shuffle things around a bit, if the fn
1238 is too large do not compute AVAIL either as this is really the
1241 2017-11-22 Richard Biener <rguenther@suse.de>
1243 PR tree-optimization/83089
1244 * tree-if-conv.c (pass_if_conversion::execute): If anything
1245 changed reset SCEV and free the number of iteration estimates.
1247 2017-11-21 Martin Sebor <msebor@redhat.com>
1249 PR tree-optimization/82945
1250 * calls.h (warn_nonstring_bound): Remove unused function.
1252 2017-11-21 Martin Sebor <msebor@redhat.com>
1254 PR tree-optimization/82945
1255 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
1256 * calls.h (maybe_warn_nonstring_arg): Declare new function.
1257 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
1259 (initialize_argument_information): Call maybe_warn_nonstring_arg.
1260 * calls.h (get_attr_nonstring_decl): Declare new function.
1261 * doc/extend.texi (attribute nonstring): Update.
1262 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
1263 get_attr_nonstring_decl and handle it.
1264 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
1265 detection of nul-termination.
1266 (strlen_to_stridx): Change to a pointer.
1267 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
1268 (pass_strlen::execute): Same.
1270 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
1272 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
1273 for the new option -mprefer-vector-width=[none|128|256|512].
1274 * config/i386/i386.c (ix86_target_string): remove old style options
1275 -mprefer-avx256 and make -mprefer-avx128 as alias.
1276 (ix86_option_override_internal): Apply defaults for the
1277 -mprefer-vector-width=[128|256] option.
1278 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
1279 Implement macros to work with -mprefer-vector-width=.
1280 * config/i386/i386.opt: Implemented option
1281 -mprefer-vector-width=[none|128|256|512].
1282 * doc/invoke.texi: Documentation for
1283 -mprefer-vector-width=[none|128|256|512].
1285 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
1287 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
1289 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
1291 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
1293 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
1295 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
1297 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
1298 * expr.c (expand_expr_real_2): Likewise.
1299 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
1300 fold_binary_loc): Likewise.
1301 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
1302 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
1303 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
1304 MINUS_EXPR transformations.
1305 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
1306 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
1307 * tree-inline.c (estimate_operator_cost): Likewise.
1308 * tree-pretty-print.c (dump_generic_node, op_code_prio,
1309 op_symbol_code): Likewise.
1310 * tree-vect-stmts.c (vectorizable_operation): Likewise.
1311 * vr-values.c (extract_range_from_binary_expr): Likewise.
1312 * varasm.c (initializer_constant_valid_p_1): Likewise.
1313 * tree.def: New tree code POINTER_DIFF_EXPR.
1315 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
1317 * config/i386/i386.md (*bswap<mode>2_movbe): Add
1318 integer suffix to movbe mnemonic.
1319 (*bswaphi2_movbe): Ditto.
1320 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
1322 2017-11-21 Jakub Jelinek <jakub@redhat.com>
1325 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
1326 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
1327 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
1329 2017-11-21 Martin Liska <mliska@suse.cz>
1331 * tree-inline.c (expand_call_inline): Remove not needed
1334 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
1335 Jakub Jelinek <jakub@redhat.com>
1338 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
1339 Remove static keyword from f variable.
1341 2017-11-21 Jakub Jelinek <jakub@redhat.com>
1343 PR tree-optimization/83086
1344 * gimple-ssa-store-merging.c
1345 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
1346 rather than n.base_addr.
1348 2017-11-21 Martin Liska <mliska@suse.cz>
1350 PR rtl-optimization/82044
1351 PR tree-optimization/82042
1352 * dse.c (check_mem_read_rtx): Check for overflow.
1354 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
1356 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
1359 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
1361 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
1362 and its corresponding call arg location note.
1364 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
1365 Andrew Burgess <andrew.burgess@embecosm.com>
1367 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
1369 (arc_return_slot_offset): Likewise.
1370 (arc_eh_return_address_location): New declaration.
1371 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
1372 (MUST_SAVE_REGISTER): Add exception handler case.
1373 (MUST_SAVE_RETURN_ADDR): Likewise.
1374 (arc_frame_pointer_required): Likewise.
1375 (arc_frame_pointer_needed): New function.
1376 (arc_compute_frame_size): Changed.
1377 (arc_expand_prologue): Likewise.
1378 (arc_expand_epilogue): Likewise.
1379 (arc_initial_elimination_offset): Likewise.
1380 (arc_return_slot_offset): Delete.
1381 (arc_eh_return_address_location): New function.
1382 (arc_builtin_setjmp_frame_value): Likewise.
1383 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
1384 (EH_RETURN_STACKADJ_RTX): Define.
1385 (EH_RETURN_HANDLER_RTX): Likewise.
1386 * config/arc/arc.md (eh_return): Delete.
1388 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
1390 * print-tree.h (debug_vec_tree): Remove prototype.
1391 * gdbinit.in (pvt): Remove macro.
1393 2017-11-21 Jakub Jelinek <jakub@redhat.com>
1395 PR tree-optimization/83047
1396 * gimple-ssa-store-merging.c
1397 (imm_store_chain_info::output_merged_store): If the loads with the
1398 same vuse are in different basic blocks, for load_gsi pick a load
1399 location that is dominated by the other loads.
1402 * config/i386/i386.c (ix86_memmodel_check): Start
1403 -Winvalid-memory-model diagnostics with lowercase letter.
1406 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
1407 set in_cold_section_p to the partition containing loc_list->first.
1408 When seeing loc_list->last_before_switch node, update secname and
1409 perform range_across_switch second partition handling only after that.
1412 * run-rtl-passes.c: Include debug.h.
1413 (run_rtl_passes): Call debug_hooks->assembly_start.
1414 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
1418 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
1419 OPTAB_DIRECT in calls to expand_simple_binop.
1421 2017-11-20 David Malcolm <dmalcolm@redhat.com>
1424 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
1426 2017-11-20 David Malcolm <dmalcolm@redhat.com>
1429 * spellcheck.h (best_match::blithely_get_best_candidate): New
1432 2017-11-20 Jakub Jelinek <jakub@redhat.com>
1434 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
1435 with lower case letter.
1437 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
1439 * config/i386/i386.md (bswaphi2): New expander.
1440 (*bswaphi2_movbe): New insn pattern.
1441 (bswaphi -> rorhi pepehole2): New peephole pattern.
1443 2017-11-20 Jeff Law <law@redhat.com>
1445 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
1446 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
1447 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
1448 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
1449 gimple-ssa-evrp-analyze.h.
1451 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
1452 set BB_VISITED here.
1453 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
1455 2017-11-20 Bin Cheng <bin.cheng@arm.com>
1457 * tree-predcom.c: Add general comment on Store-Store chains.
1458 (split_data_refs_to_components): Postpone clearing eliminate_store_p
1460 (get_chain_last_ref_at): Rename into...
1461 (get_chain_last_write_at): ...this.
1462 (get_chain_last_write_before_load): New function.
1463 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
1464 CT_STORE_STORE when write reference is added.
1465 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
1466 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
1467 (initialize_root_vars_store_elim_1): Ditto.
1468 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
1469 definition is created.
1470 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
1471 chain by replacing it with dominant stored value.
1473 2017-11-20 Bin Cheng <bin.cheng@arm.com>
1475 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
1477 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
1479 * vr-values.c (extract_range_from_binary_expr): Use a full range
1482 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
1484 * config/arm/arm.md (R4_REGNUM): Define constant.
1485 (nonsecure_call_internal): Remove r4 clobber.
1486 (nonsecure_call_value_internal): Likewise.
1487 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
1488 clobber and resequence match_operands.
1489 (nonsecure_call_value_reg_thumb1_v5): Likewise.
1490 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
1491 (nonsecure_call_value_reg_thumb2): Likewise.
1493 2017-11-20 Jakub Jelinek <jakub@redhat.com>
1495 PR tree-optimization/78821
1496 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
1497 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
1498 to the address of the base rather than the base itself.
1499 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
1500 (find_bswap_or_nop_finalize): New function.
1501 (find_bswap_or_nop): Use it.
1502 (bswap_replace): Return a tree rather than bool, change first
1503 argument from gimple * to gimple_stmt_iterator, allow inserting
1504 into an empty sequence, allow ins_stmt to be NULL - then emit
1505 all stmts into gsi. Fix up MEM_REF address gimplification.
1506 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
1507 (struct store_immediate_info): Add N and INS_STMT non-static
1509 (store_immediate_info::store_immediate_info): Initialize them
1510 from newly added ctor args.
1511 (merged_store_group::apply_stores): Formatting fixes. Sort by
1513 (stmts_may_clobber_ref_p): For stores call also
1514 refs_anti_dependent_p.
1515 (gather_bswap_load_refs): New function.
1516 (imm_store_chain_info::try_coalesce_bswap): New method.
1517 (imm_store_chain_info::coalesce_immediate_stores): Use it.
1518 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
1519 (imm_store_chain_info::output_merged_store): Fail if number of
1520 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
1521 and NOP_EXPR rhs_code.
1522 (pass_store_merging::process_store): Compute n and ins_stmt, if
1523 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
1524 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
1526 (pass_store_merging::execute): Calculate dominators.
1528 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
1529 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
1530 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
1531 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
1532 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
1533 class pass_optimize_bswap, bswap_replace,
1534 pass_optimize_bswap::execute): Moved to ...
1535 * gimple-ssa-store-merging.c: ... this file.
1536 Include optabs-tree.h.
1537 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
1538 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
1539 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
1540 anonymous namespace, remove static keywords.
1541 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
1542 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
1544 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
1547 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
1549 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
1551 * vec.h (debug_helper): New function.
1552 (DEFINE_DEBUG_VEC): New macro.
1553 * hash-set.h (debug_helper): New function.
1554 (DEFINE_DEBUG_HASH_SET): New macro.
1555 * cfg.c (debug_slim (edge)): New function.
1556 Call DEFINE_DEBUG_VEC for edges.
1557 Call DEFINE_DEBUG_HASH_SET for edges.
1558 * cfghooks.c (debug_slim (basic_block)): New function.
1559 Call DEFINE_DEBUG_VEC for basic blocks.
1560 Call DEFINE_DEBUG_HASH_SET for basic blocks.
1561 * print-tree.c (debug_slim): New function to handle trees.
1562 Call DEFINE_DEBUG_VEC for trees.
1563 Call DEFINE_DEBUG_HASH_SET for trees.
1564 (debug (vec<tree, va_gc>) &): Remove.
1565 (debug (<vec<tree, va_gc>) *): Remove.
1566 * print-rtl.c (debug_slim): New function to handle const_rtx.
1567 Call DEFINE_DEBUG_VEC for rtx_def.
1568 Call DEFINE_DEBUG_VEC for rtx_insn.
1569 Call DEFINE_DEBUG_HASH_SET for rtx_def.
1570 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
1571 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
1572 (debug (vec<rtx_insn *> *ptr): Remove.
1573 (debug_insn_vector): Remove.
1574 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
1576 2017-11-20 Tom de Vries <tom@codesourcery.com>
1578 PR rtl-optimization/82020
1579 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
1580 IF_THEN_ELSE condition.
1582 2017-11-19 Jeff Law <law@redhat.com>
1584 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1585 of degenerates resulting from ignoring an edge.
1587 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1590 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
1592 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1595 * profile-count.c (profile_count::to_sreal_scale): Fix return value
1596 for uninitialied counts.
1598 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1601 * tree-inline.c (estimate_num_insns): Set to 1 at least.
1603 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1606 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
1608 2017-11-19 Tom de Vries <tom@codesourcery.com>
1610 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
1613 2017-11-19 Tom de Vries <tom@codesourcery.com>
1615 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
1618 2017-11-19 Tom de Vries <tom@codesourcery.com>
1620 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1623 2017-11-19 Tom de Vries <tom@codesourcery.com>
1625 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
1627 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
1629 2017-11-19 Tom de Vries <tom@codesourcery.com>
1631 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
1633 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
1634 semicolon after MCORE_EXPORT_NAME call.
1636 2017-11-19 Tom de Vries <tom@codesourcery.com>
1639 * vmsdbgout.c (vmsdbgout_early_finish): New function.
1640 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
1642 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
1644 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
1645 (cgraph_edge::clone): Cleanup updating of profile.
1646 * ipa-cp.c (update_profiling_info): Likewise.
1647 * ipa-inline-transform.c (inline_transform): Likewise.
1648 * ipa-inline.c (inline_small_functions): Add missing space to dump.
1649 * ipa-split.c (execute_split_functions): Do not split when function
1651 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
1652 * profile-count.c (profile_count::dump): Add global0.
1653 (profile_count::to_cgraph_frequency): Do not ICE when entry is
1655 (profile_count::to_sreal_scale): Likewise.
1656 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
1657 (profile_count::combine_with_ipa_count): New function.
1658 * profile-count.h (profile_guessed_global0adjusted): New.
1659 (profile_count::adjusted_zero): New.
1660 (profile_count::global0adjusted): New.
1661 (profile_count::combine_with_ipa_count): New.
1662 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
1663 correct profile of return block of split functions.
1664 (copy_cfg_body): Remove unused profile_count.
1665 (copy_body): Likewise.
1666 (expand_call_inline): Update.
1667 (tree_function_versioning): Update.
1669 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
1671 * hash-set.h (hash_set::empty): New.
1672 * tree-ssa-threadbackward.h: Delete.
1673 * tree-ssa-threadbackward.c (class thread_jumps): New.
1674 Move max_threaded_paths into class.
1675 (fsm_find_thread_path): Remove arguments that are now in class.
1676 (profitable_jump_thread_path): Rename to...
1677 (thread_jumps::profitable_jump_thread_path): ...this.
1678 (convert_and_register_jump_thread_path): Rename to...
1679 (thread_jumps::convert_and_register_current_path): ...this.
1680 (check_subpath_and_update_thread_path): Rename to...
1681 (thread_jumps::check_subpath_and_update_thread_path): ...this.
1682 (register_jump_thread_path_if_profitable): Rename to...
1683 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
1684 (handle_phi): Rename to...
1685 (thread_jumps::handle_phi): ...this.
1686 (handle_assignment): Rename to...
1687 (thread_jumps::handle_assignment): ...this.
1688 (fsm_find_control_statement_thread_paths): Rename to...
1689 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
1690 (find_jump_threads_backwards): Rename to...
1691 (thread_jumps::find_jump_threads_backwards): ...this.
1692 Initialize path local data.
1693 (pass_thread_jumps::execute): Call find_jump_threads_backwards
1694 from within thread_jumps class.
1695 (pass_early_thread_jumps::execute): Same.
1697 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1699 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
1701 2017-11-17 Steve Ellcey <sellcey@cavium.com>
1703 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
1704 to canonical location.
1706 2017-11-17 Steve Ellcey <sellcey@cavium.com>
1709 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
1711 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
1713 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
1715 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
1716 rcrt1.o%s/grcrt1.o%s for -static-pie.
1718 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1720 * i386.c (ix86_multiplication_cost, ix86_division_cost,
1721 ix86_shift_rotate_cost): Break out from ...
1722 (ix86_rtx_costs): ... here.
1723 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
1726 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1728 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
1729 when entry block was promoted unlikely.
1730 (estimate_bb_frequencies): Increase frequency scale.
1731 * profile-count.h (profile_count): Export precision info.
1733 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1735 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
1736 disturbin profile of entry block.
1738 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1740 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
1743 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1745 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
1746 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
1748 2017-11-17 Jeff Law <law@redhat.com>
1750 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
1751 from evrp_dom_walker class. Various methods moved into new class.
1752 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
1753 (evrp_range_analyzer::enter): New method.
1754 (evrp_range_analyzer::leave): New method.
1755 (evrp_dom_walker): Remove delegators no longer needed by this class.
1756 Replace vr_values data member with evrp_range_analyzer
1758 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
1759 method extracted from evrp_dom_walker::before_dom_children.
1760 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
1761 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
1763 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
1764 Add private copy constructor and move assignment operators.
1765 Privatize methods and class data where trivially possible.
1766 (evrp_dom_walker::cleanup): New function, extracted from
1767 execute_early_vrp. Simplify access to class data.
1769 * vr-values.h (get_output_for_vrp): Prototype.
1770 * vr-values.c (get_output_for_vrp): New function extracted from
1771 vrp_visit_assignment_or_call and extract_range_from_stmt.
1772 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
1774 2017-11-17 Luis Machado <luis.machado@linaro.org>
1776 * config/aarch64/aarch64.c
1777 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
1778 (qdf24xx_tunings) <autoprefetcher_model>: Set to
1779 tune_params::AUTOPREFETCHER_WEAK.
1781 2017-11-17 Tamar Christina <tamar.christina@arm.com>
1784 * config/arm/arm.c (arm_valid_target_attribute_rec):
1785 Parse "arch=" and "+<ext>".
1786 (arm_valid_target_attribute_tree): Re-init global options.
1787 (arm_option_override): Make non-static.
1788 (arm_options_perform_arch_sanity_checks): Make errors fatal.
1789 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
1790 (__ARM_FEATURE_CRC32): Support undef.
1791 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
1792 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
1794 2017-11-17 David Malcolm <dmalcolm@redhat.com>
1796 * gdbinit.in (break-on-diagnostic): New command.
1798 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1800 * config/i386/i386.c (ix86_expand_epilogue): Change simple
1801 return to indirect jump for EH return if control-flow
1802 protection is enabled. Change explicit 'false' argument in
1803 pro_epilogue_adjust_stack with a value of
1805 * config/i386/i386.md (simple_return_indirect_internal):
1806 Remove SImode restriction to support 64-bit.
1808 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
1810 * combine.c (added_notes_insn): New.
1811 (try_combine): Handle added_notes_insn like added_links_insn.
1812 Rewrite return value code.
1813 (distribute_notes): Set added_notes_insn to the earliest insn we added
1816 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
1818 PR rtl-optimization/82621
1819 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
1820 dest of one of those SETs is unused.
1822 2017-11-17 Richard Biener <rguenther@suse.de>
1825 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
1826 * tree-pretty-print.c (dump_generic_node): Handle
1827 annot_expr_parallel_kind.
1828 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
1829 * gimplify.c (gimple_boolify): Likewise.
1831 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1833 * config.gcc (extra_headers): Add cet.h for x86 targets.
1834 * config/i386/cet.h: New file.
1835 * doc/install.texi: Add --enable-cet/--disable-cet.
1837 2017-11-17 Richard Biener <rguenther@suse.de>
1839 PR tree-optimization/83017
1840 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
1841 (gen_parallel_loop): Properly count iterations.
1842 (parallelize_loops): Handle loop->can_be_parallel independent
1843 of flag_loop_parallelize_all. Make static profitability test match
1845 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
1846 * invoke.texi (parloops-min-per-thread): Document.
1848 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
1850 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
1851 upstreaming review comments.
1853 2017-11-17 Tamar Christina <tamar.christina@arm.com>
1855 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
1856 with fast unaligned access.
1857 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
1859 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
1861 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
1862 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
1864 2017-11-17 Richard Biener <rguenther@suse.de>
1866 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
1867 folding of references.
1869 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
1872 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
1873 of replacing call to strncmp with corresponding call to strcmp when
1876 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
1878 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
1879 option prefer-avx256 for skylake-avx512 configuration.
1880 * config/i386/i386.c (ix86_option_override_internal): Ditto.
1881 (get_builtin_code_for_version): Ditto.
1883 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
1884 Monk Chiang <sh.chiang04@gmail.com>
1886 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
1887 (FIXED_REGISTERS): Reserve more register numbers.
1888 (CALL_USED_REGISTERS): Likewise.
1889 (REG_ALLOC_ORDER): Likewise.
1890 (REG_CLASS_CONTENTS): Likewise.
1891 (REGISTER_NAMES): Likewise.
1893 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
1894 Kito Cheng <kito.cheng@gmail.com>
1896 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
1898 * config/nds32/iterators.md: Add vector mode iterators and attributes.
1900 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
1902 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
1903 parameter list for vec_splats.
1905 2017-11-16 Joseph Myers <joseph@codesourcery.com>
1907 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
1909 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
1911 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
1914 * doc/install.texi: Document incompatibility of Perl >=5.6.26
1915 with the required version of automake 1.11.6.
1917 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
1919 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
1921 (DU_C2_3_power9): Correct reservation combinations.
1922 (FP_DIV_power9, VEC_DIV_power9): New.
1923 (power9-alu): Split out rotate/shift...
1924 (power9-rot): ...to here, correct dispatch resource.
1925 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
1927 (power9-fp): Correct latency.
1928 (power9-sdiv): Add div/sqrt resource.
1929 (power9-ddiv): Correct latency, add div/sqrt resource.
1930 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
1931 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
1933 (power9-qpdiv, power9-qpmul): Adjust resource usage.
1935 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1937 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
1938 switch statement mapping KF built-ins to TF built-ins if we don't
1939 have the proper ISA 3.0 assembler support.
1941 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1943 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
1944 (lower_emutls_function_body): Do not compute it.
1946 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1948 * ipa-split.c (split_bb_info): Turn time to sreal.
1949 (split_point): Likewise.
1950 (dump_split_point): Likewise.
1951 (fine_split_points): Likewise.
1952 (execute_split_functions): Only zero split_bbs; turn time to sreals.
1954 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1956 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
1959 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1961 * predict.c (combine_predictions_for_bb): Preserve zero predicted
1963 (expensive_function_p): Remove useless assert.
1964 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
1966 2017-11-16 Martin Sebor <msebor@redhat.com>
1968 PR tree-optimization/82588
1969 PR tree-optimization/82583
1970 * tree-vrp.c (check_array_ref): Handle flexible array members,
1971 string literals, and inner indices.
1972 (search_for_addr_array): Add detail to diagnostics.
1974 2017-11-16 Nathan Sidwell <nathan@acm.org>
1978 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
1979 (SET_DECL_ASSEMBLER_NAME): Forward to
1980 overwrite_decl_assembler_name.
1981 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
1982 (overwrite_decl_assembler_name): Declare.
1983 * tree.c (overwrite_decl_assembler_name): New.
1984 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
1985 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
1986 (LANG_HOOKS_INITIALIZER): Add it.
1987 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
1988 * langhooks.c (lhd_set_decl_assembler_name): Use
1989 SET_DECL_ASSEMBLER_NAME.
1990 (lhd_overwrite_decl_assembler_name): Default implementation.
1992 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
1993 Jackson Woodruff <jackson.woodruff@arm.com>
1995 PR tree-optimization/71026
1996 * match.pd: Canonicalize constant multiplies in division.
1998 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
2000 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
2001 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
2002 Remove OPT_fomit_frame_pointer entry.
2003 * common/config/arc/arc-common.c: Likewise.
2004 * common/config/arm/arm-common.c: Likewise.
2005 * common/config/avr/avr-common.c: Likewise.
2006 * common/config/c6x/c6x-common.c: Likewise.
2007 * common/config/cr16/cr16-common.c: Likewise.
2008 * common/config/cris/cris-common.c: Likewise.
2009 * common/config/epiphany/epiphany-common.c: Likewise.
2010 * common/config/fr30/fr30-common.c: Likewise.
2011 * common/config/frv/frv-common.c: Likewise.
2012 * common/config/ia64/ia64-common.c: Likewise.
2013 * common/config/iq2000/iq2000-common.c: Likewise.
2014 * common/config/lm32/lm32-common.c: Likewise.
2015 * common/config/m32r/m32r-common.c: Likewise.
2016 * common/config/mcore/mcore-common.c: Likewise.
2017 * common/config/microblaze/microblaze-common.c: Likewise.
2018 * common/config/mips/mips-common.c: Likewise.
2019 * common/config/mmix/mmix-common.c: Likewise.
2020 * common/config/mn10300/mn10300-common.c: Likewise.
2021 * common/config/nios2/nios2-common.c: Likewise.
2022 * common/config/pa/pa-common.c: Likewise.
2023 * common/config/pdp11/pdp11-common.c: Likewise.
2024 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
2025 * common/config/riscv/riscv-common.c: Likewise.
2026 * common/config/rs6000/rs6000-common.c: Likewise.
2027 * common/config/rx/rx-common.c: Likewise.
2028 * common/config/s390/s390-common.c: Likewise.
2029 * common/config/sh/sh-common.c: Likewise.
2030 * common/config/sparc/sparc-common.c: Likewise.
2031 * common/config/tilegx/tilegx-common.c: Likewise.
2032 * common/config/tilepro/tilepro-common.c: Likewise.
2033 * common/config/v850/v850-common.c: Likewise.
2034 * common/config/visium/visium-common.c: Likewise.
2035 * common/config/xstormy16/xstormy16-common.c: Likewise.
2036 * common/config/xtensa/xtensa-common.c: Likewise.
2037 * invoke.texi (-fomit-frame-pointer): Update documentation.
2039 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2041 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
2043 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2045 * cfg.c (scale_bbs_frequencies_int,
2046 cale_bbs_frequencies_gcov_type): Remove.
2047 * cfg.h (scale_bbs_frequencies_int,
2048 cale_bbs_frequencies_gcov_type): Remove.
2050 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2052 * tree-ssa-loop-manip.c
2053 (scale_dominated_blocks_in_loop): Update to profile counts.
2054 (tree_transform_and_unroll_loop): Likewise.
2056 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2058 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
2059 scale_bbs_frequencies_int.
2061 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2063 * final.c (compute_alignments): Use counts rather than frequencies.
2065 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2067 * cfgloopanal.c: Include sreal.h
2068 (average_num_loop_insns): Use counts and sreal for accounting.
2070 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2072 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
2075 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2076 Kito Cheng <kito.cheng@gmail.com>
2078 * config/nds32/constraints.md: Provide more constraints.
2079 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
2080 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
2081 support constraints usage.
2083 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2085 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
2087 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2088 Kito Cheng <kito.cheng@gmail.com>
2090 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
2091 * config/nds32/nds32.opt: Refine the layout.
2092 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
2093 TARGET_EXT_STRING): Support new options.
2094 * config/nds32/nds32.h: Likewise.
2095 * config/nds32/nds32.md: Likewise.
2096 * config/nds32/nds32-predicates.c: Likewise.
2097 * config/nds32/constraints.md: Likewise.
2098 * common/config/nds32/nds32-common.c: Likewise.
2100 2017-11-16 Julia Koval <julia.koval@intel.com>
2103 * config/i386/gfniintrin.h: Add sse check.
2104 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
2106 2017-11-16 Julia Koval <julia.koval@intel.com>
2108 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
2109 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
2110 (ix86_handle_option): Handle -mavx512vbmi2.
2111 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
2112 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
2113 * config/i386/i386-c.c (__AVX512VBMI2__): New.
2114 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
2115 (ix86_valid_target_attribute_inner_p): Ditto.
2116 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
2117 * config/i386/i386.opt (mavx512vbmi2): New option.
2118 * doc/invoke.texi: Add new option.
2120 2017-11-16 Julia Koval <julia.koval@intel.com>
2122 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
2123 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
2124 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
2125 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
2126 _mm512_gf2p8mul_epi8): New intrinsics.
2127 * config/i386/i386-builtin-types.def
2128 (V64QI_FTYPE_V64QI_V64QI): New type.
2129 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
2130 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
2131 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
2132 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
2133 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
2134 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2136 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
2138 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
2139 explicitly as a stream of bytes.
2141 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2143 * config/rs6000/altivec.h (vec_xst_be): New #define.
2144 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
2145 and externalize from *altivec_vperm_<mode>_internal.
2146 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
2148 (XL_BE_V8HI): Likewise.
2149 (XL_BE_V4SI): Likewise.
2150 (XL_BE_V4SI): Likewise.
2151 (XL_BE_V2DI): Likewise.
2152 (XL_BE_V4SF): Likewise.
2153 (XL_BE_V2DF): Likewise.
2154 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
2155 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
2156 all array entries with these keys: VSX_BUILTIN_VEC_XL,
2157 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
2158 VSX_BUILTIN_VEC_XST_BE.
2159 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
2160 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
2162 (altivec_init_builtins): Replace conditional calls to def_builtin
2163 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
2164 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
2165 with unconditional calls. Remove calls to def_builtin for
2166 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
2167 __builtin_vec_xst_be.
2168 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
2169 to define_expand, and add alternate RTL generation for P8.
2170 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
2171 vsx_ld_elemrev_v8hi.
2172 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
2173 add alternate RTL generation for P8.
2174 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
2175 vsx_ld_elemrev_v16qi.
2176 (vsx_st_elemrev_v8hi): Convert define_insn
2177 to define_expand, and add alternate RTL generation for P8.
2178 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
2179 vsx_st_elemrev_v8hi.
2180 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
2181 add alternate RTL generation for P8.
2182 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
2183 vsx_st_elemrev_v16qi.
2185 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
2188 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
2189 TARGET_AVX512ER check.
2190 (ix86_option_override_internal): Set MASK_VZEROUPPER if
2191 neither -mzeroupper nor -mno-zeroupper is used and
2192 TARGET_EMIT_VZEROUPPER is set.
2193 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
2194 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
2196 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
2198 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2199 folding of vector compares.
2200 (fold_build_vec_cmp): New helper function.
2201 (fold_compare_helper): New helper function.
2202 (builtin_function_type): Add compare builtins to the list of functions
2203 having unsigned arguments. Cosmetic updates to comment indentation.
2204 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
2205 the not+eq combination.
2207 2017-11-15 Bin Cheng <bin.cheng@arm.com>
2209 PR tree-optimization/82726
2210 PR tree-optimization/70754
2211 * tree-predcom.c (order_drefs_by_pos): New function.
2212 (combine_chains): Move code setting has_max_use_after to...
2213 (try_combine_chains): ...here. New parameter. Sort combined chains
2214 according to position information.
2215 (tree_predictive_commoning_loop): Update call to above function.
2216 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
2218 2017-11-15 Bin Cheng <bin.cheng@arm.com>
2220 PR tree-optimization/82726
2222 2017-01-23 Bin Cheng <bin.cheng@arm.com>
2224 PR tree-optimization/70754
2225 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
2226 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
2227 combined stmt before it if not NULL.
2228 (combine_chains): Process refs reversely and compute dominance point
2232 2017-02-23 Bin Cheng <bin.cheng@arm.com>
2234 PR tree-optimization/79663
2235 * tree-predcom.c (combine_chains): Process refs in reverse order
2236 only for ZERO length chains, and add explaining comment.
2238 2017-11-15 Tamar Christina <tamar.christina@arm.com>
2240 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
2241 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
2242 * doc/invoke.texi (ARM Options): Add armv8.3-a.
2244 2017-11-15 Tamar Christina <tamar.christina@arm.com>
2246 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
2248 2017-11-15 Martin Liska <mliska@suse.cz>
2250 * tree-cfg.c (pass_warn_function_return::execute):
2251 Compare warn_return_type for greater than zero.
2253 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
2257 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
2258 to return true on Xeon and not on Xeon Phi.
2259 (ix86_check_avx256_register): Changed to ...
2260 (ix86_check_avx_upper_register): ... this. Add extra check for
2261 VALID_AVX512F_REG_OR_XI_MODE.
2262 (ix86_avx_u128_mode_needed): Changed
2263 ix86_check_avx256_register to ix86_check_avx_upper_register.
2264 (ix86_check_avx256_stores): Changed to ...
2265 (ix86_check_avx_upper_stores): ... this. Changed
2266 ix86_check_avx256_register to ix86_check_avx_upper_register.
2267 (ix86_avx_u128_mode_after): Changed
2268 avx_reg256_found to avx_upper_reg_found. Changed
2269 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
2270 (ix86_avx_u128_mode_entry): Changed
2271 ix86_check_avx256_register to ix86_check_avx_upper_register.
2272 (ix86_avx_u128_mode_exit): Ditto.
2273 * config/i386/i386.h: (host_detect_local_cpu): New define.
2275 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2277 * config/arm/xgene1.md (xgene1): Split into automatons
2278 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
2279 (xgene1_f_load): Adjust reservations and/or types.
2280 (xgene1_f_store): Likewise.
2281 (xgene1_load_pair): Likewise.
2282 (xgene1_store_pair): Likewise.
2283 (xgene1_fp_load1): Likewise.
2284 (xgene1_load1): Likewise.
2285 (xgene1_store1): Likewise.
2286 (xgene1_move): Likewise.
2287 (xgene1_alu): Likewise.
2288 (xgene1_simd): Likewise.
2289 (xgene1_bfm): Likewise.
2290 (xgene1_neon_load1): Likewise.
2291 (xgene1_neon_store1): Likewise.
2292 (xgene1_neon_logic): Likewise.
2293 (xgene1_neon_st1): Likewise.
2294 (xgene1_neon_ld1r): Likewise.
2295 (xgene1_alu_cond): Added.
2296 (xgene1_shift_reg): Likwise.
2297 (xgene1_bfx): Likewise.
2298 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
2300 2017-11-15 Jakub Jelinek <jakub@redhat.com>
2303 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
2305 (can_widen_mult_without_libcall): New function.
2306 (expand_mul_overflow): If only checking unsigned mul overflow,
2307 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
2308 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
2309 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
2310 (expand_DIVMOD): Formatting fix.
2311 * expmed.h (expand_mult): Add NO_LIBCALL argument.
2312 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
2313 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
2315 PR tree-optimization/82977
2316 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
2317 constructed temporary to strlen_to_stridx.put.
2319 2017-11-15 Martin Liska <mliska@suse.cz>
2321 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
2322 * configure: Regenerate.
2324 2017-11-15 Martin Liska <mliska@suse.cz>
2327 * config/sh/sh-mem.cc: Use proper probability for
2330 2017-11-14 Jeff Law <law@redhat.com>
2332 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
2333 the red zone for stack_clash_protection_final_dynamic_probe targets
2334 when the total dynamic stack size is zero bytes.
2336 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
2337 blocks is post order.
2339 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
2341 * dumpfile.h (TDF_COMPARE_DEBUG): New.
2342 * final.c (rest_of_clean_state): Set it for the
2343 -fcompare-debug dump.
2344 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
2345 class when TDF_COMPARE_DEBUG is set.
2347 * dwarf2out.c (gen_producer_string): Discard
2350 2017-11-15 Joseph Myers <joseph@codesourcery.com>
2353 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
2354 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
2355 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
2356 (__TGMATH_REAL_2_3): Remove macros.
2357 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
2358 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
2360 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
2362 (remquo): Define using __TGMATH_REAL_3.
2364 2017-11-14 Jeff Law <law@redhat.com>
2366 * vr-values.c: New file with contents extracted from tree-vrp.c.
2367 * Makefile.in (OBJS): Add vr-values.o
2368 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
2369 (set_value_range, set_and_canonicalize_value_range): Likewise.
2370 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
2371 (value_range_constant_singleton, symbolic_range_p): Likewise.
2372 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
2373 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
2374 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
2375 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
2376 (find_case_label_range, find_case_label_index): Likewise.
2377 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
2378 (range_int_cst_singleton_p, value_inside_range): Likewise.
2379 (get_single_symbol): Likewise.
2380 (switch_update): Move structure definition here.
2381 (to_remove_edges, to_update_switch_stmts): Provide externs.
2382 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
2383 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
2384 (vrp_val_is_min, set_value_range): Likewise.
2385 (set_and_canonicalize_value_range, copy_value_range): Likewise.
2386 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
2387 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
2388 (range_is_nonnull, range_int_cst_p): Likewwise.
2389 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
2390 (get_single_symbol, operand_less_p): Likewise
2391 (compare_values_warnv, compare_values): Likewise.
2392 (value_inside_range, value_range_constant_singleton): Likewise.
2393 (zero_nonzero_bitgs_from_vr): Likewise.
2394 (extract_range_from_binary_expr_1): Likewise.
2395 (overflow_comparison_p): Likewise.
2396 (to_remove_edges, to_update_switch_stmts): Likewise.
2397 (find_case_label-index, find_case_label_range): Likewise.
2398 (switch_update, set_value_range_to_nonnegative): Remove.
2399 (set_value_range_to_truthvalue): Likewise.
2400 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
2401 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
2402 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
2403 (find_case_label_ranges, test_for_singularity): Likewise.
2404 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
2405 (x_vr_values): Move to its remaining use site.
2407 2017-11-10 Jeff Law <law@redhat.com>
2409 * vr-values.h (VR_INITIALIZER): Move #define here.
2410 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
2411 * Makefile.in (OBJS): Add tree-evrp.o
2412 * tree-vrp.h (assert_info): Move structure definition here.
2413 (set_value_range_to_varying): Prototype.
2414 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
2415 (infer_value_range, register_edge_assert_for): Likewise.
2416 (stmt_interesting_for_vrp): Likewise.
2417 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
2418 (set_value_range_to_varying): No longer static.
2419 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
2420 (infer_value_range, register_edge_assert_for): Likewise.
2422 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
2424 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
2425 generate the XXBRD instruction.
2427 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
2428 (rs6000_builtin_type_compatible): Treat _Float128 and long double
2429 as being compatible if -mabi=ieeelongdouble.
2430 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
2431 to setup float128 built-ins with hardware support.
2432 (BU_FLOAT128_HW_2): Likewise.
2433 (BU_FLOAT128_HW_3): Likewise.
2434 (BU_FLOAT128_HW_VSX_1): Likewise.
2435 (BU_FLOAT128_HW_VSX_2): Likewise.
2436 (scalar_extract_expq): Change float128 built-in functions to
2437 accommodate having both KFmode and TFmode functions. Use the
2438 KFmode variant as the default.
2439 (scalar_extract_sigq): Likewise.
2440 (scalar_test_neg_qp): Likewise.
2441 (scalar_insert_exp_q): Likewise.
2442 (scalar_insert_exp_qp): Likewise.
2443 (scalar_test_data_class_qp): Likewise.
2444 (sqrtf128_round_to_odd): Delete processing the round to odd
2445 built-in functions as special built-in functions, and define them
2446 as float128 built-ins. Use the KFmode variant as the default.
2447 (truncf128_round_to_odd): Likewise.
2448 (addf128_round_to_odd): Likewise.
2449 (subf128_round_to_odd): Likewise.
2450 (mulf128_round_to_odd): Likewise.
2451 (divf128_round_to_odd): Likewise.
2452 (fmaf128_round_to_odd): Likewise.
2453 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
2454 support for KFmode and TFmode xststdcqp calls.
2455 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
2456 point, switch the built-in handlers for the get/set float128
2457 exponent, get float128 mantissa, float128 test built-ins, and the
2458 float128 round to odd built-in functions. Eliminate creating the
2459 float128 round to odd built-in functions as special built-ins.
2460 (rs6000_init_builtins): Eliminate special creation of the float128
2461 round to odd built-in functions.
2462 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
2463 function insns to support both TFmode and KFmode variants.
2464 (xsxsigqp_<mode>): Likewise.
2465 (xsiexpqpf_<mode>): Likewise.
2466 (xsiexpqp_<mode>): Likewise.
2467 (xststdcqp_<mode>): Likewise.
2468 (xststdcnegqp_<mode>): Likewise.
2469 (xststdcqp_<mode>): Likewise.
2471 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2473 * tree-ssa-threadupdate.c (compute_path_counts): Remove
2474 unused path_in_freq_ptr parameter.
2475 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
2477 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2479 * ipa-inline.c (edge_badness): Dump sreal frequency.
2480 (compute_inlined_call_time): Match natural implementaiton ...
2481 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
2482 forgotten division by CGRAPH_FREQ_BASE.
2484 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2486 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
2487 Solaris 11. Update comment.
2488 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
2490 * config/sol2.h (STARTFILE_SPEC): Likewise.
2491 * configure: Regenerate.
2493 2017-11-14 Carl Love <cel@us.ibm.com>
2495 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
2496 le_ and be_ prefixes to swap* variables. Remove
2497 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
2500 2017-11-14 Jason Merrill <jason@redhat.com>
2502 Support GTY((cache)) on hash_map.
2503 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
2504 (ggc_cache_remove): Override it instead of ggc_mx.
2505 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
2506 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
2507 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
2508 (simple_cache_map_traits): Override maybe_mx.
2509 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
2510 (hash_map): Friend gt_cleare_cache.
2511 (gt_cleare_cache): New.
2512 * tree.h (tree_cache_traits): New hash_map traits class.
2513 (tree_cache_map): New typedef.
2515 2017-11-14 Richard Biener <rguenther@suse.de>
2517 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
2518 paramter and handling.
2519 (cleanup_control_flow_bb): Likewise.
2520 (cleanup_control_flow_pre): New helper performing a DFS walk
2521 to call cleanup_control_flow_bb in PRE order.
2522 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
2523 via cleanup_control_flow_pre.
2525 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
2527 * config/aarch64/aarch64-simd.md
2528 (aarch64_simd_bsl<mode>_internal): Remove DImode.
2529 (*aarch64_simd_bsl<mode>_alt): Likewise.
2530 (aarch64_simd_bsldi_internal): New.
2531 (aarch64_simd_bsldi_alt): Likewise.
2533 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2535 * tracer.c (better_p): Do not compare frequencies.
2536 * reg-stack.c (better_edge): Likewise.
2537 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
2540 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2542 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
2543 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
2544 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
2545 * lto-streamer-in.c (input_function): Use update_max_bb_count.
2546 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
2547 * predict.c (maybe_hot_frequency_p): Inline to ...
2548 (maybe_hot_count_p): ... here; rewrite to counts.
2549 (counts_to_freqs): Rename to ...
2550 (update_max_bb_count): ... this one.
2551 (expensive_function_p): Use counts.
2552 (estimate_bb_frequencies): Update.
2553 (rebuild_frequencies): Update.
2554 * predict.h (counts_to_freqs): Rename to ...
2555 (update_max_bb_count): ... this one.
2556 * profile.c (compute_branch_probabilities): Add debug info
2557 * tree-inline.c (expand_call_inline): Update debug info.
2558 (optimize_inline_calls): Use update_max_bb_count..
2559 (tree_function_versioning): Use update_max_bb_count..
2560 * value-prof.c (gimple_value_profile_transformations):
2561 Do not use update_max_bb_count.
2563 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2565 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
2566 always use frequencies.
2568 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2570 * bb-reorder.c: Remove frequencies from comments.
2571 (better_edge_p): Use profile counts.
2572 (find_traces): Dump profile counts.
2573 (rotate_loop): Use profile counts.
2574 (find_traces_1_round): Likewise.
2575 (connect_better_edge_p): Use counts instead of probabilities for
2577 (copy_bb_p): Drop early check for non-0 frequency.
2578 (sanitize_hot_paths): Update comments.
2580 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2582 * ipa-split.c (struct split_point): Add count.
2583 (consider_split): Do not compute incoming frequency; compute incoming
2584 count and store it to split_point.
2585 (split_function): Set count of the call to split part correctly.
2587 2017-11-13 Carl Love <cel@us.ibm.com>
2589 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
2591 2017-11-13 Tom Tromey <tom@tromey.com>
2593 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
2595 2017-11-13 Carl Love <cel@us.ibm.com>
2597 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
2598 Add support for builtins:
2599 unsigned int vec_first_{,miss}_match_{,or_eos}index,
2600 vector {un,}signed {char,int,short},
2601 vector {un,}signed {char,int,short}) arguments.
2602 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
2603 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
2604 Add BU_P9V_AV_2 expansions for the builtins.
2605 * config/rs6000/altivec.h (vec_first_match_index,
2606 vec_first_mismatch_index, vec_first_match_or_eos_index,
2607 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
2608 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
2609 new extern declaration.
2610 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
2611 * config/rs6000/vsx.md (first_match_index_<mode>,
2612 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
2613 first_mismatch_or_eos_index_<mode>): Add define expand.
2614 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
2615 * doc/extend.texi: Update the built-in documenation file for the new
2618 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
2620 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
2621 into the min/max operations for _Float<N> and _Float<N>X types.
2623 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
2626 * dwarf2out.c (do_eh_frame): New static variable.
2627 (dwarf2out_begin_prologue): Set it.
2628 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
2630 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2632 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
2634 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
2636 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
2639 * debug.h (dwarf2out_do_eh_frame): Declare.
2640 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
2641 (dwarf2out_do_frame): Use it.
2642 (dwarf2out_do_cfi_asm): Likewise.
2643 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
2644 (dwarf2out_assembly_start): Likewise.
2645 (dwarf2out_begin_prologue): Fix comment.
2646 * toplev.c (compile_file): Always call dwarf2out_frame_finish
2647 if the target needs either debug or unwind DWARF2 info.
2648 * lto-opts.c (lto_write_options): Do not save -fexceptions,
2649 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
2650 -ftrapping-math, -ftrapv and -fwrapv.
2652 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2654 * cgraph.c (cgraph_edge::sreal_frequency): New function.
2655 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
2656 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
2657 (estimate_edge_size_and_time): Likewise.
2658 (ipa_merge_fn_summary_after_inlining): Likewise.
2659 * ipa-inline.c (cgraph_freq_base_rec): Remove.
2660 (compute_uninlined_call_time): Use sreal_frequency.
2661 (compute_inlined_call_time): Likewise.
2662 (ipa_inline): Do not initialize cgraph_freq_base_rec.
2663 * profile-count.c: Include sreal.h.
2664 (profile_count::to_sreal_scale): New.
2665 * profile-count.h: Forward declare sreal.
2666 (profile_count::to_sreal_scale): Declare.
2668 2017-11-13 Nathan Sidwell <nathan@acm.org>
2670 * diagnostic.c (maybe_line_and_column): New.
2671 (diagnostic_get_location_text): Use it.
2672 (diagnostic_report_current_module): Likewise.
2673 (test_diagnostic_get_location_text): Add tests.
2675 2017-11-13 Luis Machado <luis.machado@linaro.org>
2677 * doc/md.texi (Specifying processor pipeline description): Fix
2678 incorrect latency for the div instruction example.
2680 2017-11-13 Jakub Jelinek <jakub@redhat.com>
2682 PR tree-optimization/78821
2683 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
2684 that bit_not_p is the same.
2685 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
2686 (split_group): Count precisely bit_not_p bits in each statement.
2687 (invert_op): New function.
2688 (imm_store_chain_info::output_merged_store): Use invert_op to
2689 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
2690 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
2692 2017-11-13 Martin Liska <mliska@suse.cz>
2694 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
2695 (struct source_info): Likewise.
2696 (add_branch_counts): Likewise.
2697 (add_line_counts): Likewise.
2698 (function_summary): Likewise.
2699 (output_intermediate_line): Likewise.
2700 (generate_results): Likewise.
2702 2017-11-13 Martin Liska <mliska@suse.cz>
2704 * gcov.c (struct block_info): Remove typedef for block_t.
2705 (struct line_info): Likewise.
2706 (line_info::has_block): Likewise.
2707 (EXIT_BLOCK): Likewise.
2708 (unblock): Likewise.
2709 (circuit): Likewise.
2710 (get_cycles_count): Likewise.
2711 (process_file): Likewise.
2712 (read_graph_file): Likewise.
2713 (solve_flow_graph): Likewise.
2714 (find_exception_blocks): Likewise.
2715 (add_line_counts): Likewise.
2716 (accumulate_line_info): Likewise.
2717 (output_line_details): Likewise.
2719 2017-11-13 Martin Liska <mliska@suse.cz>
2721 * gcov.c (struct arc_info): Remove typedef for arc_t.
2722 (struct line_info): Likewise.
2723 (add_branch_counts): Likewise.
2724 (output_branch_count): Likewise.
2725 (function_info::~function_info): Likewise.
2726 (circuit): Likewise.
2727 (output_intermediate_line): Likewise.
2728 (read_graph_file): Likewise.
2729 (solve_flow_graph): Likewise.
2730 (find_exception_blocks): Likewise.
2731 (add_line_counts): Likewise.
2732 (accumulate_line_info): Likewise.
2733 (output_line_details): Likewise.
2734 (output_function_details): Likewise.
2736 2017-11-13 Martin Liska <mliska@suse.cz>
2738 * gcov.c (struct function_info): Remove typedef for function_t.
2739 (struct source_info): Likewise.
2740 (source_info::get_functions_at_location): Likewise.
2741 (solve_flow_graph): Likewise.
2742 (find_exception_blocks): Likewise.
2743 (add_line_counts): Likewise.
2744 (output_intermediate_file): Likewise.
2745 (process_file): Likewise.
2746 (generate_results): Likewise.
2747 (release_structures): Likewise.
2748 (read_graph_file): Likewise.
2749 (read_count_file): Likewise.
2750 (accumulate_line_counts): Likewise.
2751 (output_lines): Likewise.
2753 2017-11-13 Martin Liska <mliska@suse.cz>
2755 * gcov.c (function_info::function_info): Remove num_counts
2756 and add vector<gcov_type>.
2757 (function_info::~function_info): Use the vector.
2758 (process_file): Likewise.
2759 (read_graph_file): Likewise.
2760 (read_count_file): Likewise.
2761 (solve_flow_graph): Likewise.
2763 2017-11-13 Martin Liska <mliska@suse.cz>
2765 * gcov.c (function_info::is_artificial): New function.
2766 (process_file): Erase all artificial early.
2767 (generate_results): Skip as all artificial are already
2770 2017-11-13 Martin Liska <mliska@suse.cz>
2772 * gcov.c (read_graph_file): Store to global vector of functions.
2773 (read_count_file): Iterate the vector.
2774 (process_file): Likewise.
2775 (generate_results): Likewise.
2776 (release_structures): Likewise.
2778 2017-11-13 Jakub Jelinek <jakub@redhat.com>
2780 PR tree-optimization/82954
2781 * gimple-ssa-store-merging.c
2782 (imm_store_chain_info::coalesce_immediate_stores): If
2783 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
2785 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
2787 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
2788 Upddate call to ENDIAN_LANE_N.
2789 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
2790 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
2791 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
2792 and use aarch64_endian_lane_rtx.
2793 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
2795 2017-11-12 Tom de Vries <tom@codesourcery.com>
2797 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
2799 2017-11-12 Tom de Vries <tom@codesourcery.com>
2801 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
2803 2017-11-12 Tom de Vries <tom@codesourcery.com>
2805 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
2807 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
2808 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
2809 * defaults.h (ASM_OUTPUT_LABELREF): Same.
2811 2017-11-11 Martin Sebor <msebor@redhat.com>
2814 * doc/extend.texi (attribute nonstring): Remove spurious argument.
2817 * prefic.c (translate_name): Replace strncpy with memcpy to
2818 avoid -Wstringop-truncation.
2820 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
2822 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
2824 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
2826 * predict.c (maybe_hot_frequency_p): Do not use cfun.
2828 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
2830 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
2833 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
2835 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
2836 enable generating XXBRH if the value is in a vector register.
2837 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
2838 value is in a vector register.
2839 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
2840 register to register bswap64's instead of doing the GPR sequence
2841 used on previous machines.
2842 (bswapdi2_xxbrd): New insn.
2843 (bswapdi2_reg): Disallow on ISA 3.0.
2844 (register to register bswap64 splitter): Do not split the insn on
2845 ISA 3.0 systems that use XXBRD.
2847 2017-11-10 Martin Sebor <msebor@redhat.com>
2850 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
2852 (find_subframework_file): Same.
2854 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
2856 * auto-profile.c (afdo_indirect_call): Drop frequency.
2857 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
2858 (cgraph_node::create_edge): Drop frequency argument.
2859 (cgraph_node::create_indirect_edge): Drop frequency argument.
2860 (cgraph_edge::make_speculative): Drop frequency arguments.
2861 (cgraph_edge::resolve_speculation): Do not update frequencies
2862 (cgraph_edge::dump_edge_flags): Do not dump frequency.
2863 (cgraph_node::dump): Check consistency in IPA mode.
2864 (cgraph_edge::maybe_hot_p): Use IPA counter.
2865 (cgraph_edge::verify_count_and_frequency): Rename to ...
2866 (cgraph_edge::verify_count): ... this one; drop frequency checking.
2867 (cgraph_node::verify_node): Update.
2868 * cgraph.h (struct cgraph_edge): Drop frequency.
2869 (cgraph_edge::frequency): New function.
2870 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
2872 (cgraph_edge::rebuild_edges): Likewise.
2873 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
2874 (duplicate_thunk_for_node): Do not pass frequency.
2875 (cgraph_node::create_clone): Scale only counts.
2876 (cgraph_node::create_virtual_clone): Do not pass frequency.
2877 (cgraph_node::create_edge_including_clones): Do not pass frequency.
2878 (cgraph_node::create_version_clone): Do not pass frequency.
2879 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
2880 (cgraph_node::expand_thunk): Do not pass frequency.
2881 (cgraph_node::create_wrapper): Do not pass frequency.
2882 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
2884 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
2885 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
2886 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
2887 (gather_caller_stats): Use frequency function.
2888 (ipcp_cloning_candidate_p): Use frequency function.
2889 (ipcp_propagate_stage): Use frequency function.
2890 (get_info_about_necessary_edges): Use frequency function.
2891 (update_profiling_info): Update only IPA profile.
2892 (update_specialized_profile): Use frequency functoin.
2893 (perhaps_add_new_callers): Update only IPA profile.
2894 * ipa-devirt.c (ipa_devirt): Use IPA profile.
2895 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
2896 (dump_ipa_call_summary): Use frequency function.
2897 (estimate_edge_size_and_time): Use frequency function.
2898 (ipa_merge_fn_summary_after_inlining): Use frequency function.
2899 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
2900 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
2901 (update_noncloned_counts): ... ths one; scale counts only.
2902 (clone_inlined_nodes): Do not scale frequency.
2903 (inline_call): Do not pass frequency.
2904 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
2905 (compute_inlined_call_time): Use IPA profile.
2906 (want_inline_small_function_p): Use IPA profile.
2907 (want_inline_self_recursive_call_p): Use IPA profile.
2908 (edge_badness): Use IPA profile.
2909 (lookup_recursive_calls): Use IPA profile.
2910 (recursive_inlining): Do not pass frequency.
2911 (resolve_noninline_speculation): Do not update frequency.
2912 (inline_small_functions): Collect max of IPA profile.
2913 (dump_overall_stats): Dump IPA porfile.
2914 (dump_inline_stats): Dump IPA porfile.
2915 (ipa_inline): Collect IPA stats.
2916 * ipa-inline.h (clone_inlined_nodes): Update prototype.
2917 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
2918 (ipa_propagate_frequency): Use frequency function.
2919 (ipa_profile): Cleanup.
2920 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
2921 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
2922 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
2923 (input_node): Do not stream frequency.
2924 (input_edge): Do not stream frequency.
2925 (merge_profile_summaries): Scale only IPA profiles.
2926 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
2927 * predict.c (drop_profile): Do not recompute frequency.
2928 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
2929 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
2930 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
2931 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
2932 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
2933 * tree-inline.c (copy_bb): Do not scale frequency.
2934 (expand_call_inline): Do not scale frequency.
2935 (tree_function_versioning): Do not scale frequency.
2936 * ubsan.c (ubsan_create_edge): Do not pass frequency.
2938 2017-11-10 Julia Koval <julia.koval@intel.com>
2940 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
2941 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
2942 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
2943 (_mm256_maskz_gf2p8affine_epi64_epi8)
2944 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
2945 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
2946 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
2947 (__builtin_ia32_vgf2p8affineqb_v32qi)
2948 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
2949 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
2951 2017-11-10 Tamar Christina <tamar.christina@arm.com>
2955 (arm_option_override): Refactor.
2956 (arm_option_reconfigure_globals): New.
2957 (arm_options_perform_arch_sanity_checks): New.
2958 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
2960 (arm_options_perform_arch_sanity_checks): Likewise
2962 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
2964 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
2965 (power9-qpmul): New.
2966 * rs6000/rs6000.md ("type" attr): Add qmul.
2967 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
2968 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
2969 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
2971 2017-11-10 Martin Sebor <msebor@redhat.com>
2974 * builtins.c (compute_objsize): Handle arrays that
2975 compute_builtin_object_size likes to fail for. Make extern.
2976 * builtins.h (compute_objsize): Declare.
2977 (check_strncpy_sizes): New function.
2978 (expand_builtin_strncpy): Call check_strncpy_sizes.
2979 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
2980 -Wstringop-truncation.
2981 (gimple_fold_builtin_strncat): Same.
2982 * gimple.c (gimple_build_call_from_tree): Set call location.
2983 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
2984 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
2985 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
2986 (handle_builtin_strlen): Use strlen_to_stridx.
2987 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
2989 Use strlen_to_stridx.
2990 (pass_strlen::execute): Release strlen_to_stridx.
2991 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
2992 (-Wstringop-truncation): Document new option.
2994 2017-11-10 Martin Liska <mliska@suse.cz>
2996 PR gcov-profile/82702
2997 * gcov.c (main): Handle intermediate files in a different
2999 (get_gcov_intermediate_filename): New function.
3000 (output_gcov_file): Remove support of intermediate files.
3001 (generate_results): Allocate intermediate file.
3002 (release_structures): Clean-up properly fn_end.
3003 (output_intermediate_file): Start iterating with line 1.
3005 2017-11-10 Jakub Jelinek <jakub@redhat.com>
3007 PR tree-optimization/82929
3008 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
3009 ops_swapped_p non-static data member.
3010 (store_immediate_info::store_immediate_info): Clear it.
3011 (imm_store_chain_info::coalesce_immediate_stores): If swapping
3012 ops set ops_swapped_p.
3013 (count_multiple_uses): Handle ops_swapped_p.
3015 2017-11-10 Martin Liska <mliska@suse.cz>
3017 * coverage.c (coverage_init): Stream information about
3018 support of has_unexecuted_blocks.
3019 * doc/gcov.texi: Document that.
3020 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
3021 * gcov.c (read_graph_file): Likewise.
3022 (output_line_beginning): Fix a small issue with
3025 2017-11-10 Bin Cheng <bin.cheng@arm.com>
3027 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
3028 reference of trivial component.
3030 2017-11-10 Jakub Jelinek <jakub@redhat.com>
3033 * gimple-ssa-store-merging.c
3034 (pass_store_merging::terminate_all_aliasing_chains): For
3035 gimple_store_p stmts also call refs_output_dependent_p.
3037 PR rtl-optimization/82913
3038 * compare-elim.c (try_merge_compare): Punt if def_insn is not
3041 2017-11-09 Jeff Law <law@redhat.com>
3043 * vr-values.h: New file with vr_values class.
3044 * tree-vrp.c: Include vr-values.h
3045 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
3046 data objects into the vr_values class.
3047 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
3048 (get_value_range): Make it a member function within vr_values class.
3049 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
3050 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
3051 (op_with_constant_singleton_value_range): Likewise.
3052 (extract_range_for_var_from_comparison_expr): Likewise.
3053 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
3054 (extract_range_from_binary_expr): Likewise.
3055 (extract_range_from_unary_expr): Likewise.
3056 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
3057 (check_for_binary_op_overflow, extract_range_basic): Likewise.
3058 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
3059 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
3060 (compare_name_with_value, compare_names): Likewise.
3061 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
3062 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
3063 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
3064 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
3065 (extract_range_from_phi_node): Likewise.
3066 (simplify_truth_ops_using_ranges): Likewise.
3067 (simplify_div_or_mod_using_ranges): Likewise.
3068 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
3069 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
3070 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
3071 (simplify_float_conversion_using_ranges): Likewise.
3072 (simplify_internal_call_using_ranges): Likewise.
3073 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
3074 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
3075 poitner via x_vr_values for calls into gimple folder.
3076 (vrp_initialize_lattice): Make this the vr_values ctor.
3077 (vrp_free_lattice): Make this the vr_values dtor.
3078 (set_vr_value): New function.
3079 (class vrp_prop): Add vr_values data member. Add various member
3080 functions as well as member functions that delegate to vr_values.
3081 (check_array_ref): Make a member function within vrp_prop class.
3082 (search_for_addr_array, vrp_initialize): Likewise.
3083 (vrp_finalize): Likewise. Revamp to avoid direct access to
3084 vr_value, values_propagated, etc.
3085 (check_array_bounds): Extract vrp_prop class instance pointer from
3086 walk info structure. Use it to call member functions.
3087 (check_all_array_refs): Make a member function within vrp_prop class.
3088 Smuggle class instance pointer via walk info structure.
3089 (x_vr_values): New local static.
3090 (vrp_valueize): Use x_vr_values to get class instance.
3091 (vr_valueize_1): Likewise.
3092 (class vrp_folder): Add vr_values data member. Add various member
3093 functions as well as member functions that delegate to vr_values.
3094 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
3095 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
3096 class instance from vr_values. Use it to call member functions.
3097 (vrp_dom_walker): Add vr_values data member.
3098 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
3099 instance via x_vr_values.
3100 (identify_jump_threads): Accept vr_values as argument. Store
3101 it into the walker structure.
3102 (evrp_dom_walker): Add vr_values class data member. Add various
3104 (evrp_dom_walker::try_find_new_range): Use vr_values data
3105 member to access the memory allocator.
3106 (evrp_dom_walker::before_dom_children): Store vr_values class
3107 instance into the vrp_folder class.
3108 (evrp_dom_walker::push_value_range): Rework to avoid direct
3109 access to num_vr_values and vr_value.
3110 (evrp_dom_walker::pop_value_range): Likewise.
3111 (execute_early_vrp): Remove call to vrp_initialize_lattice.
3112 Use vr_values to get to dump_all_value_ranges member function.
3113 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
3114 and simplify_cond_using_ranges_2 via vrp_prop class instance.
3115 Pass vr_values class instance down to identify_jump_threads.
3116 Remove call to vrp_free_lattice.
3117 (debug_all_value_ranges): Remove.
3119 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
3120 (vrp_folder): Likewise.
3122 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
3123 Get it from the existing bitmap instead.
3124 (vrp_intersect_ranges_1): Likewise.
3126 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3128 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
3130 (store_immediate_info::store_immediate_info): Add bitnotp argument,
3131 set bit_not_p to it.
3132 (imm_store_chain_info::coalesce_immediate_stores): Break group
3133 if bit_not_p is different.
3134 (count_multiple_uses, split_group,
3135 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
3136 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
3137 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
3138 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
3139 to store_immediate_info ctor.
3141 2017-11-09 Jim Wilson <jimw@sifive.com>
3143 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
3144 (scan_prog_file): Likewise.
3146 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
3148 * bb-reorder.c (max_entry_frequency): Remove.
3149 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
3150 connect_traces, push_to_next_round_p): Remove prototypes.
3151 (find_traces_1_round): Use counts only.
3152 (push_to_next_round_p): Likewise.
3153 (find_traces): Likewise.
3154 (rotate_loop): Likewise.
3155 (find_traces_1_round): Likewise.
3156 (connect_traces): Likewise.
3157 (edge_order): Likewise.
3159 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
3161 * config/arm/arm.c (output_return_instruction): Add comments to
3162 indicate requirement for cmse_nonsecure_entry return to account
3163 for the size of clearing instruction output here.
3164 (thumb_exit): Likewise.
3165 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
3166 return in hardfloat mode.
3168 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
3170 * config/rs6000/rs6000.c (machine_function): Add a bool,
3171 "toc_is_wrapped_separately".
3172 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
3173 if it wasn't explicitly set or unset, we are optimizing for speed, and
3174 doing separate shrink-wrapping.
3175 (rs6000_get_separate_components): Enable the TOC component if
3176 saving the TOC register in the prologue.
3177 (rs6000_components_for_bb): Handle the TOC component.
3178 (rs6000_emit_prologue_components): Store the TOC register where needed.
3179 (rs6000_set_handled_components): Mark TOC as handled, if handled.
3180 (rs6000_emit_prologue): Don't save the TOC if that is already done.
3182 2017-11-09 Martin Jambor <mjambor@suse.cz>
3184 * ipa-param-manipulation.c: New file.
3185 * ipa-param-manipulation.h: Likewise.
3186 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
3187 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
3188 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
3189 (ipa_parm_adjustment): Likewise.
3190 (ipa_parm_adjustment_vec): Likewise.
3191 (ipa_get_vector_of_formal_parms): Moved declaration to
3192 ipa-param-manipulation.h.
3193 (ipa_get_vector_of_formal_parm_types): Likewise.
3194 (ipa_modify_formal_parameters): Likewise.
3195 (ipa_modify_call_arguments): Likewise.
3196 (ipa_combine_adjustments): Likewise.
3197 (ipa_dump_param_adjustments): Likewise.
3198 (ipa_modify_expr): Likewise.
3199 (ipa_get_adjustment_candidate): Likewise.
3200 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
3201 ipa-param-manipulation.c.
3202 (ipa_get_vector_of_formal_parm_types): Likewise.
3203 (ipa_modify_formal_parameters): Likewise.
3204 (ipa_modify_call_arguments): Likewise.
3205 (ipa_modify_expr): Likewise.
3206 (get_ssa_base_param): Likewise.
3207 (ipa_get_adjustment_candidate): Likewise.
3208 (index_in_adjustments_multiple_times_p): Likewise.
3209 (ipa_combine_adjustments): Likewise.
3210 (ipa_dump_param_adjustments): Likewise.
3211 * tree-sra.c: Also include ipa-param-manipulation.h
3212 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
3215 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3216 Alan Hayward <alan.hayward@arm.com>
3217 David Sherwood <david.sherwood@arm.com>
3219 * doc/sourcebuild.texi (vect_masked_store): Document.
3221 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3222 Alan Hayward <alan.hayward@arm.com>
3223 David Sherwood <david.sherwood@arm.com>
3225 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
3227 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3228 Alan Hayward <alan.hayward@arm.com>
3229 David Sherwood <david.sherwood@arm.com>
3231 * doc/sourcebuild.texi (vect_variable_length): Document.
3233 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3234 Alan Hayward <alan.hayward@arm.com>
3235 David Sherwood <david.sherwood@arm.com>
3237 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
3239 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3240 Alan Hayward <alan.hayward@arm.com>
3241 David Sherwood <david.sherwood@arm.com>
3243 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
3245 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3246 Alan Hayward <alan.hayward@arm.com>
3247 David Sherwood <david.sherwood@arm.com>
3249 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
3250 previously undocumented selectors.
3251 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
3253 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3255 * doc/rtl.texi (const_vector): Say that elements can be
3256 const_wide_ints too.
3257 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
3258 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
3259 (gen_vec_duplicate): Use it instead of CONSTANT_P.
3260 * optabs.c (expand_vector_broadcast): Likewise.
3262 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3263 Alan Hayward <alan.hayward@arm.com>
3264 David Sherwood <david.sherwood@arm.com>
3266 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
3267 scaled index even if the unscaled address was invalid.
3268 Don't increase the complexity of using a scale in that case.
3270 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3271 Alan Hayward <alan.hayward@arm.com>
3272 David Sherwood <david.sherwood@arm.com>
3274 * doc/rtl.texi: Rewrite the subreg rules so that they partition
3275 the inner register into REGMODE_NATURAL_SIZE bytes rather than
3276 UNITS_PER_WORD bytes.
3277 * emit-rtl.c (validate_subreg): Divide subregs into blocks
3278 based on REGMODE_NATURAL_SIZE of the inner mode.
3279 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
3280 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
3281 * expmed.c (lowpart_bit_field_p): Divide the value up into
3282 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
3283 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
3284 whether something is likely to occupy more than one register.
3286 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
3289 * ipa-inline-transform.c (update_noncloned_frequencies): Use
3290 profile_count::adjust_for_ipa_scaling.
3291 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
3292 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
3294 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
3296 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3298 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
3299 (split_group): Add total_orig and total_new arguments, estimate the
3300 number of statements related to the store group without store merging
3301 and with store merging.
3302 (imm_store_chain_info::output_merged_store): Adjust split_group
3303 callers, punt if estimated number of statements with store merging
3304 is not smaller than estimated number of statements without it.
3306 (handled_load): Remove has_single_use checks.
3307 (pass_store_merging::process_store): Likewise.
3309 2017-11-09 Richard Biener <rguenther@suse.de>
3311 PR tree-optimization/82902
3312 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
3314 2017-11-09 Martin Liska <mliska@suse.cz>
3317 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
3320 2017-11-09 Martin Liska <mliska@suse.cz>
3322 PR tree-optimization/82669
3323 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
3325 2017-11-09 Martin Liska <mliska@suse.cz>
3327 PR gcov-profile/48463
3328 * coverage.c (coverage_begin_function): Output also end locus
3329 of a function and information whether the function is
3331 * gcov-dump.c (tag_function): Parse and print the information.
3332 * gcov.c (INCLUDE_MAP): Add include.
3333 (INCLUDE_SET): Likewise.
3334 (struct line_info): Move earlier in the source file because
3335 of vector<line_info> in function_info structure.
3336 (line_info::line_info): Likewise.
3337 (line_info::has_block): Likewise.
3338 (struct source_info): Add new member index.
3339 (source_info::get_functions_at_location): New function.
3340 (function_info::group_line_p): New function.
3341 (output_intermediate_line): New function.
3342 (output_intermediate_file): Use the mentioned function.
3343 (struct function_start): New.
3344 (struct function_start_pair_hash): Likewise.
3345 (process_file): Add code that identifies group functions.
3346 Assign lines either to global or function scope.
3347 (generate_results): Skip artificial functions.
3348 (find_source): Assign index for each source file.
3349 (read_graph_file): Read new flag artificial and end_line.
3350 (add_line_counts): Assign it either to global of function scope.
3351 (accumulate_line_counts): Isolate core of the function to
3352 accumulate_line_info and call it for both function and global
3354 (accumulate_line_info): New function.
3355 (output_line_beginning): Fix GNU coding style.
3356 (print_source_line): New function.
3357 (output_line_details): Likewise.
3358 (output_function_details): Likewise.
3359 (output_lines): Iterate both source (global) scope and function
3361 (struct function_line_start_cmp): New class.
3362 * doc/gcov.texi: Reflect changes in documentation.
3364 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3367 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
3368 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
3369 and similarly for not instead of neg.
3371 2017-11-08 Andi Kleen <ak@linux.intel.com>
3373 * config/i386/i386.opt: Add -mforce-indirect-call.
3374 * config/i386/predicates.md: Check for flag_force_indirect_call.
3375 * doc/invoke.texi: Document -mforce-indirect-call
3377 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
3379 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
3381 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
3382 config/riscv/riscv.c (predict.h): New include.
3383 (riscv_slow_unaligned_access_p): No longer static.
3384 (riscv_block_move_straight): Add require.
3385 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
3387 2017-11-08 Jakub Jelinek <jakub@redhat.com>
3390 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
3391 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
3392 nonimmediate_operand predicate for operand 1 instead of
3395 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3397 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
3399 * config/aarch64/constraints.md (Uml): New constraint.
3400 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
3403 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3405 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
3406 of two vec_duplicates into a vec_concat.
3408 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3410 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
3411 Simplify vec_merge of vec_duplicate and vec_concat.
3412 * config/aarch64/constraints.md (Utq): New constraint.
3413 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
3416 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3418 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
3419 Simplify vec_merge of vec_duplicate and const_vector.
3420 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
3422 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
3423 mode iterator. Update predicate on operand 1 to
3424 handle non-const_vec constants. Delete constraints.
3425 (*aarch64_combinez_be<mode>): Likewise for operand 2.
3427 2017-11-08 Jakub Jelinek <jakub@redhat.com>
3429 PR tree-optimization/78821
3430 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
3432 (store_operand_info::store_operand_info): Initialize it to false.
3433 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
3434 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
3435 store in the group, and if chain_info is non-NULL, to ignore altogether
3437 (compatible_load_p): Fail if bit_not_p does not match.
3438 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
3439 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
3440 (pass_store_merging::process_store): Adjust
3441 terminate_all_aliasing_chains calls to pass NULL in all current spots,
3442 call terminate_all_aliasing_chains newly when adding a store into
3443 a chain with non-NULL chain_info.
3445 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
3447 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
3449 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
3451 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3453 (aarch64_layout_frame): Initialise emit_frame_chain.
3454 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
3455 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
3457 2017-11-08 Martin Liska <mliska@suse.cz>
3459 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
3460 of gimple_call_internal_p.
3462 2017-11-07 Tom de Vries <tom@codesourcery.com>
3464 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
3466 2017-11-07 Tom de Vries <tom@codesourcery.com>
3468 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
3471 2017-11-08 Martin Liska <mliska@suse.cz>
3474 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
3476 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
3478 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
3480 2017-11-07 Jakub Jelinek <jakub@redhat.com>
3483 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
3484 (*cmp<mode>_ccz_1): New insn with $k alternative.
3487 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
3488 RTX_COMM_COMPARE as commutative as well.
3489 (ix86_binary_operator_ok): Formatting fix.
3490 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
3491 *<code><mode>3<mask_name><round_saeonly_name>,
3492 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
3493 *<s>mul<mode>3_highpart<mask_name>,
3494 *vec_widen_umult_even_v16si<mask_name>,
3495 *vec_widen_umult_even_v8si<mask_name>,
3496 *vec_widen_umult_even_v4si<mask_name>,
3497 *vec_widen_smult_even_v16si<mask_name>,
3498 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
3499 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
3500 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
3501 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
3502 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
3503 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
3504 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
3505 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
3506 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
3507 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
3508 ix86_binary_operator_ok. Formatting fixes.
3509 (*<plusminus_insn><mode>3<mask_name><round_name>,
3510 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
3513 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
3515 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
3516 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
3517 eq and ne if TARGET_ISEL.
3518 (cmp): New code_iterator.
3519 (UNS, UNSU_, UNSIK): New code_attrs.
3520 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
3521 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
3523 ("eq<mode>3"): ... this.
3524 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
3526 ("ne<mode>3"): ... this.
3528 2017-11-07 Julia Koval <julia.koval@intel.com>
3531 * common/config/i386/i386-common.c
3532 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
3533 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
3534 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
3535 * config/i386/i386.opt: Ditto.
3536 * config/i386/i386.c (ix86_target_string): Ditto.
3537 (ix86_option_override_internal): Ditto.
3538 (ix86_init_mpx_builtins): Move MPX to args2.
3539 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
3540 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3541 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3542 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3543 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3544 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3545 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
3547 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
3550 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
3551 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
3552 (zero-extendsidi peephole2): Remove peephole.
3554 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
3557 * stor-layout.c: Include attribs.h.
3558 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
3559 explicit lookup of "aligned" attribute.
3561 2017-11-07 Andrew Waterman <andrew@sifive.com>
3563 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
3564 (riscv_expand_block_move): Likewise.
3565 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
3567 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
3568 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
3569 * config/riscv/riscv.c (riscv_block_move_straight): New function.
3570 (riscv_adjust_block_mem): Likewise.
3571 (riscv_block_move_loop): Likewise.
3572 (riscv_expand_block_move): Likewise.
3573 * config/riscv/riscv.md (movmemsi): New pattern.
3575 2017-11-07 Michael Clark <michaeljclark@mac.com>
3577 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
3578 (MUSL_DYNAMIC_LINKER): Likewise.
3580 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
3582 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
3583 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
3586 2017-11-07 Richard Biener <rguenther@suse.de>
3588 * match.pd: Fix build.
3590 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
3591 Jackson Woodruff <jackson.woodruff@arm.com>
3593 PR tree-optimization/71026
3594 * match.pd: Canonicalize negate in division.
3596 2017-11-07 Sudakshina Das <sudi.das@arm.com>
3599 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
3601 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
3603 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
3604 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
3606 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
3608 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
3609 non-scalar integral types.
3610 * match.pd (negate_expr_p): Handle MINUS_EXPR.
3611 (-(A-B), -(~A)): New transformations.
3613 2017-11-07 Tom de Vries <tom@codesourcery.com>
3615 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
3616 semicolon after "do {} while (0)".
3617 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3618 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3619 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3620 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3621 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3623 2017-11-07 Tom de Vries <tom@codesourcery.com>
3625 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
3626 after "do {} while (0)".
3627 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3628 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3629 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3630 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3631 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3633 2017-11-07 Tom de Vries <tom@codesourcery.com>
3636 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
3638 (arm_rtx_costs_internal): Add missing semicolon after
3639 HANDLE_NARROW_SHIFT_ARITH call.
3641 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
3643 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3644 disable isel if it was not set explicitly.
3646 2017-11-06 James Bowman <james.bowman@ftdichip.com>
3648 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
3649 (add_type_attribute) likewise.
3651 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
3653 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
3654 of struct ix86_frame.
3655 (ix86_initial_elimination_offset): Likewise.
3656 (ix86_expand_split_stack_prologue): Likewise.
3658 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
3660 * tree-vrp.h (enum value_range_type): Update stale comment.
3662 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3663 Alan Hayward <alan.hayward@arm.com>
3664 David Sherwood <david.sherwood@arm.com>
3666 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
3667 (aarch64_expand_vec_perm_const): Take the number of units too.
3668 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
3669 (aarch64_expand_vec_perm_const): Likewise.
3670 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
3671 (vec_perm<mode>): Update accordingly.
3673 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3674 Alan Hayward <alan.hayward@arm.com>
3675 David Sherwood <david.sherwood@arm.com>
3677 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
3678 Take the number of units too.
3679 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
3680 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
3681 but check for a vector mode before rather than after the call.
3682 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
3683 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
3684 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
3685 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
3686 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
3687 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
3688 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
3689 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
3690 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
3691 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
3692 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
3693 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
3694 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
3695 (aarch64_sqdmull2_n<mode>): Update accordingly.
3697 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3698 Alan Hayward <alan.hayward@arm.com>
3699 David Sherwood <david.sherwood@arm.com>
3701 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
3702 the number of units too.
3703 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
3704 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
3705 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
3706 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
3707 (vec_store_lanesxi<mode>): Update accordingly.
3709 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3710 Alan Hayward <alan.hayward@arm.com>
3711 David Sherwood <david.sherwood@arm.com>
3713 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
3714 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
3715 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
3716 of units rather than the mode.
3717 * config/aarch64/iterators.md (nunits): New mode attribute.
3718 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
3719 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
3720 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
3721 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
3722 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
3723 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
3724 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
3725 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
3726 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
3727 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
3728 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3729 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
3730 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
3731 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
3732 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
3733 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
3734 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
3735 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
3736 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
3737 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
3738 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
3739 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
3740 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
3741 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3742 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
3743 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
3744 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3745 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
3746 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
3747 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
3748 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
3749 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
3750 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
3751 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
3752 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
3753 (aarch64_simd_vec_setv2di): Likewise.
3755 2017-11-06 Carl Love <cel@us.ibm.com>
3757 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
3759 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
3760 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
3762 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
3764 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
3765 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
3766 Add power 8 macro expansions.
3767 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
3768 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
3769 power 8 instructions. (VSX_XXBR): Add iterator.
3771 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
3773 * config/arm/arm.md (predicable_short_it): Change default to "no",
3774 improve documentation, remove uses that are identical to the default.
3775 (enabled_for_depr_it): Rename to enabled_for_short_it.
3776 * gcc/config/arm/arm-fixed.md (predicable_short_it):
3777 Remove default uses.
3778 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
3779 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
3780 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
3781 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
3783 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
3786 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
3787 float128 helper macros, which are no longer used after deleting
3788 the old 'q' built-in functions, and moving the round to odd
3789 built-in functions to being special built-in functions.
3790 (BU_FLOAT128_2): Likewise.
3791 (BU_FLOAT128_1_HW): Likewise.
3792 (BU_FLOAT128_2_HW): Likewise.
3793 (BU_FLOAT128_3_HW): Likewise.
3794 (FABSQ): Delete old 'q' built-in functions.
3795 (COPYSIGNQ): Likewise.
3796 (SQRTF128_ODD): Move round to odd built-in functions to be
3797 special built-in functions, so that we can handle
3798 -mabi=ieeelongdouble.
3799 (TRUNCF128_ODD): Likewise.
3800 (ADDF128_ODD): Likewise.
3801 (SUBF128_ODD): Likewise.
3802 (MULF128_ODD): Likewise.
3803 (DIVF128_ODD): Likewise.
3804 (FMAF128_ODD): Likewise.
3805 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
3806 built-in names to 'f128'.
3807 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
3808 old 'q' built-in functions, as the machine independent code for
3809 'f128' built-in functions handles this.
3810 (rs6000_expand_builtin): Add expansion for float128 round to odd
3811 functions, keying off on -mabi=ieeelongdouble of whether to use
3812 the KFmode or TFmode variant.
3813 (rs6000_init_builtins): Initialize the _Float128 round to odd
3815 * doc/extend.texi (PowerPC Built-in Functions): Document the old
3816 _Float128 'q' built-in functions are now mapped into the new
3817 'f128' built-in functions.
3819 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
3821 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
3822 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
3824 2017-11-06 David Malcolm <dmalcolm@redhat.com>
3827 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
3828 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
3829 * toplev.c: Include "ipa-fnsummary.h".
3830 (toplev::finalize): Call ipa_fnsummary_c_finalize.
3832 2017-11-06 Jakub Jelinek <jakub@redhat.com>
3834 PR tree-optimization/82838
3835 * gimple-ssa-store-merging.c
3836 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
3837 on a separate gimple_seq which is then appended to seq.
3839 2017-11-06 Jeff Law <law@redhat.com>
3842 * config/i386/i386.c (PROBE_INTERVAL): Remove.
3843 (get_probe_interval): New functions.
3844 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
3845 (ix86_adjust_stack_and_probe): Likewise.
3846 (output_adjust_stack_and_probe): Likewise.
3847 (ix86_emit_probe_stack_range): Likewise.
3848 (ix86_expand_prologue): Likewise.
3850 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3852 PR tree-optimization/82816
3853 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
3854 if the modes of the two types are the same.
3855 (convert_plusminus_to_widen): Likewise.
3857 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3859 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
3861 (usadv16qi): New define_expand.
3862 (usadv8hi): New define_expand.
3864 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
3867 * ipa-inline-transform.c (update_noncloned_frequencies): Always
3869 (inline_transform): Likewise.
3870 * predict.c (counts_to_freqs): Remove useless conditional.
3871 * profile-count.h (profile_count::apply_scale): Move sanity check.
3872 * tree-inline.c (copy_bb): Always scale.
3873 (copy_cfg_body): Likewise.
3875 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
3878 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
3880 (*cmp_ite0): Add enabled_for_depr_it attribute.
3881 (*cmp_ite1): Likewise.
3883 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
3885 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
3888 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3890 * tree-vrp.c (vrp_int_const_binop): Return true on success and
3891 return the value by pointer.
3892 (extract_range_from_multiplicative_op_1): Update accordingly.
3893 Return as soon as an operation fails.
3895 2017-11-05 Tom de Vries <tom@codesourcery.com>
3898 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
3899 (DEF_SANITIZER_BUILTIN): ... here.
3900 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
3901 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
3903 2017-11-05 Tom de Vries <tom@codesourcery.com>
3906 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
3908 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
3909 ASM_OUTPUT_BEFORE_CASE_LABEL call.
3910 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
3912 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
3913 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
3914 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
3916 2017-11-05 Tom de Vries <tom@codesourcery.com>
3919 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
3922 2017-11-04 Michael Clark <michaeljclark@mac.com>
3924 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
3925 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
3927 (*addsi3_extended): Likewise.
3928 (*addsi3_extended2): Likewise.
3929 (<optab>si3): Likewise.
3930 (<optab>di3): Likewise.
3931 (<optab><mode>3): Likewise.
3932 (<*optabe>si3_internal): Likewise.
3933 (zero_extendqi<SUPERQI:mode>2): Likewise.
3934 (*add<mode>hi3): Likewise.
3935 (*xor<mode>hi3): Likewise.
3936 (<optab>di3): Likewise.
3937 (*<optab>si3_extend): Likewise.
3938 (*sge<u>_<X:mode><GPR:mode>): Likewise.
3939 (*slt<u>_<X:mode><GPR:mode>): Likewise.
3940 (*sle<u>_<X:mode><GPR:mode>): Likewise.
3942 2017-11-04 Andrew Waterman <andrew@sifive.com>
3944 * config/riscv/riscv.c (riscv_option_override): Conditionally set
3945 TARGET_STRICT_ALIGN based upon -mtune argument.
3947 2017-11-04 Andrew Waterman <andrew@sifive.com>
3949 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
3951 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
3953 * config/i386/i386.c (choose_basereg): Use optional scratch
3954 register and add assertion.
3955 (x86_emit_outlined_ms2sysv_save): Use scratch register when
3956 needed, and don't allocate stack.
3957 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
3958 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
3959 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
3961 2017-11-03 Jeff Law <law@redhat.com>
3963 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
3964 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
3965 to probe at the start of a noreturn function.
3967 2017-11-03 Jakub Jelinek <jakub@redhat.com>
3969 PR tree-optimization/78821
3970 * gimple-ssa-store-merging.c: Update the file comment.
3971 (MAX_STORE_ALIAS_CHECKS): Define.
3972 (struct store_operand_info): New type.
3973 (store_operand_info::store_operand_info): New constructor.
3974 (struct store_immediate_info): Add rhs_code and ops data members.
3975 (store_immediate_info::store_immediate_info): Add rhscode, op0r
3976 and op1r arguments to the ctor, initialize corresponding data members.
3977 (struct merged_store_group): Add load_align_base and load_align
3979 (merged_store_group::merged_store_group): Initialize them.
3980 (merged_store_group::do_merge): Update them.
3981 (merged_store_group::apply_stores): Pick the constant for
3982 encode_tree_to_bitpos from one of the two operands, or skip
3983 encode_tree_to_bitpos if neither operand is a constant.
3984 (class pass_store_merging): Add process_store method decl. Remove
3985 bool argument from terminate_all_aliasing_chains method decl.
3986 (pass_store_merging::terminate_all_aliasing_chains): Remove
3987 var_offset_p argument and corresponding handling.
3988 (stmts_may_clobber_ref_p): New function.
3989 (compatible_load_p): New function.
3990 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
3991 if there is overlap and rhs_code is not INTEGER_CST. For
3992 non-overlapping stores terminate group if rhs is not mergeable.
3993 (get_alias_type_for_stmts): Change first argument from
3994 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
3995 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
3996 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
3998 (get_location_for_stmts): Change first argument from
3999 auto_vec<gimple *> & to vec<gimple *> &.
4000 (struct split_store): Remove orig_stmts data member, add orig_stores.
4001 (split_store::split_store): Create orig_stores rather than orig_stmts.
4002 (find_constituent_stmts): Renamed to ...
4003 (find_constituent_stores): ... this. Change second argument from
4004 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
4005 to info structures rather than the statements.
4006 (split_group): Rename ALLOW_UNALIGNED argument to
4007 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
4008 it. Adjust find_constituent_stores caller.
4009 (imm_store_chain_info::output_merged_store): Handle rhs_code other
4010 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
4011 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
4012 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
4013 (mem_valid_for_store_merging): New function.
4014 (handled_load): New function.
4015 (pass_store_merging::process_store): New method.
4016 (pass_store_merging::execute): Use process_store method. Adjust
4017 terminate_all_aliasing_chains caller.
4019 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
4021 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
4022 Return true for more constants, symbols and label references.
4023 (aarch64_valid_floating_const): Remove unused function.
4025 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
4028 * config/aarch64/aarch64.c (aarch64_layout_frame):
4029 Undo forcing of LR at bottom of frame.
4031 2017-11-03 Jeff Law <law@redhat.com>
4034 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
4035 for int_registers_saved.
4037 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
4038 extracted from tree-ssa-dom.c.
4039 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
4040 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
4041 (record_equivalences_from_incoming_edge): Add additional argument
4042 to single_pred_edge_ignoring_loop_edges call.
4043 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
4044 (uncprop_dom_walker::before_dom_children): Add additional argument
4045 to single_pred_edge_ignoring_loop_edges call.
4046 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
4047 single_pred_edge_ignoring_loop_edges rather than open coding.
4048 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
4050 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
4052 * match.pd (-(-A)): Rewrite.
4054 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
4056 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
4057 (rs6000_emit_int_cmove): New declaration.
4058 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
4059 (rs6000_emit_sISEL): Delete.
4060 (rs6000_emit_int_cmove): Make non-static.
4061 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
4062 instead of rs6000_emit_sISEL.
4064 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
4066 * asan.c (create_cond_insert_point): Maintain profile.
4067 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
4069 * basic-block.h (struct basic_block_def): Remove frequency.
4070 (EDGE_FREQUENCY): Use to_frequency
4071 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
4073 (find_traces): Update to use to_frequency.
4074 (find_traces_1_round): Likewise; use only IPA counts.
4075 (bb_to_key): Likewise.
4076 (connect_traces): Use IPA counts only.
4077 (copy_bb_p): Update to use to_frequency.
4078 (fix_up_crossing_landing_pad): Likewise.
4079 (sanitize_hot_paths): Likewise.
4080 * bt-load.c (basic_block_freq): Likewise.
4081 * cfg.c (init_flow): Set count_max to uninitialized.
4082 (check_bb_profile): Remove frequencies; check counts.
4083 (dump_bb_info): Do not dump frequencies.
4084 (update_bb_profile_for_threading): Update counts only.
4085 (scale_bbs_frequencies_int): Likewise.
4086 (MAX_SAFE_MULTIPLIER): Remove.
4087 (scale_bbs_frequencies_gcov_type): Update counts only.
4088 (scale_bbs_frequencies_profile_count): Update counts only.
4089 (scale_bbs_frequencies): Update counts only.
4090 * cfg.h (struct control_flow_graph): Add count-max.
4091 (update_bb_profile_for_threading): Update prototype.
4092 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
4093 (find_many_sub_basic_blocks): Likewise.
4094 * cfgcleanup.c (try_forward_edges): Likewise.
4095 (try_crossjump_to_edge): Likewise.
4096 * cfgexpand.c (expand_gimple_cond): Likewise.
4097 (expand_gimple_tailcall): Likewise.
4098 (construct_init_block): Likewise.
4099 (construct_exit_block): Likewise.
4100 * cfghooks.c (verify_flow_info): Check consistency of counts.
4101 (dump_bb_for_graph): Do not dump frequencies.
4102 (split_block_1): Do not update frequencies.
4103 (split_edge): Do not update frequencies.
4104 (make_forwarder_block): Do not update frequencies.
4105 (duplicate_block): Do not update frequencies.
4106 (account_profile_record): Do not update frequencies.
4107 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
4108 for global heuristics.
4109 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
4110 (expected_loop_iterations_unbounded): Use counts only.
4111 * cfgloopmanip.c (scale_loop_profile): Simplify.
4112 (create_empty_loop_on_edge): Simplify
4114 (duplicate_loop_to_header_edge): Simplify
4115 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
4116 (update_br_prob_note): Take care of removing note when profile
4118 (relink_block_chain): Do not dump frequency.
4119 (rtl_account_profile_record): Use to_frequency.
4120 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
4121 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
4122 (cgraph_update_edges_for_call_stmt_node): Likewise.
4123 (cgraph_edge::verify_count_and_frequency): Update.
4124 (cgraph_node::verify_node): Temporarily disable frequency verification.
4125 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
4126 to_cgraph_frequency.
4127 (cgraph_edge::rebuild_edges): Convert to ipa counts.
4128 * cgraphunit.c (init_lowered_empty_function): Do not initialize
4130 (cgraph_node::expand_thunk): Update profile.
4131 * except.c (dw2_build_landing_pads): Do not update frequency.
4132 * final.c (compute_alignments): Use to_frequency.
4133 (dump_basic_block_info): Do not dump frequency.
4134 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
4135 (dump_gimple_bb_header): Do not dump frequency.
4136 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
4138 * gimple-streamer-in.c (input_bb): Do not stream frequency.
4139 * gimple-streamer-out.c (output_bb): Do not stream frequency.
4140 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
4141 (init_before_recovery): Do not update frequency.
4142 (sched_create_recovery_edges): Do not update frequency.
4143 * hsa-gen.c (convert_switch_statements): Do not update frequency.
4144 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
4145 (ipa_cp_c_finalize): Set max_count to uninitialized.
4146 * ipa-fnsummary.c (get_minimal_bb): Use counts.
4147 (param_change_prob): Use counts.
4148 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
4150 * ipa-split.c (consider_split): Use to_frequency.
4151 (split_function): Use to_frequency.
4152 * ira-build.c (loop_compare_func): Likewise.
4153 (mark_loops_for_removal): Likewise.
4154 (mark_all_loops_for_removal): Likewise.
4155 * loop-doloop.c (doloop_modify): Do not update frequency.
4156 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
4158 * lto-streamer-in.c (input_function): Update count_max.
4159 * omp-expand.c (expand_omp_taskreg): Update count_max.
4160 * omp-simd-clone.c (simd_clone_adjust): Update profile.
4161 * predict.c (maybe_hot_frequency_p): Use to_frequency.
4162 (maybe_hot_count_p): Use ipa counts only.
4163 (maybe_hot_bb_p): Simplify.
4164 (maybe_hot_edge_p): Simplify.
4165 (probably_never_executed): Do not take frequency argument.
4166 (probably_never_executed_bb_p): Do not pass frequency.
4167 (probably_never_executed_edge_p): Likewise.
4168 (combine_predictions_for_bb): Check that profile is nonzero.
4169 (propagate_freq): Do not set frequency.
4170 (drop_profile): Simplify.
4171 (counts_to_freqs): Simplify.
4172 (expensive_function_p): Use to_frequency.
4173 (propagate_unlikely_bbs_forward): Simplify.
4174 (determine_unlikely_bbs): Simplify.
4175 (estimate_bb_frequencies): Add hack to silence graphite issues.
4176 (compute_function_frequency): Use ipa counts.
4177 (pass_profile::execute): Update.
4178 (rebuild_frequencies): Use counts only.
4179 (force_edge_cold): Use counts only.
4180 * profile-count.c (profile_count::dump): Dump new count types.
4181 (profile_count::differs_from_p): Check compatiblity.
4182 (profile_count::to_frequency): New function.
4183 (profile_count::to_cgraph_frequency): New function.
4184 * profile-count.h (struct function): Declare.
4185 (enum profile_quality): Add profile_guessed_local and
4186 profile_guessed_global0.
4187 (class profile_proability): Decrease number of bits to 29;
4188 update from_reg_br_prob_note and to_reg_br_prob_note.
4189 (class profile_count: Update comment; decrease number of bits
4190 to 61. Check compatibility.
4191 (profile_count::compatible_p): New private member function.
4192 (profile_count::ipa_p): New member function.
4193 (profile_count::operator<): Handle global zero correctly.
4194 (profile_count::operator>): Handle global zero correctly.
4195 (profile_count::operator<=): Handle global zero correctly.
4196 (profile_count::operator>=): Handle global zero correctly.
4197 (profile_count::nonzero_p): New member function.
4198 (profile_count::force_nonzero): New member function.
4199 (profile_count::max): New member function.
4200 (profile_count::apply_scale): Handle IPA scalling.
4201 (profile_count::guessed_local): New member function.
4202 (profile_count::global0): New member function.
4203 (profile_count::ipa): New member function.
4204 (profile_count::to_frequency): Declare.
4205 (profile_count::to_cgraph_frequency): Declare.
4206 * profile.c (OVERLAP_BASE): Delete.
4207 (compute_frequency_overlap): Delete.
4208 (compute_branch_probabilities): Do not use compute_frequency_overlap.
4209 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
4210 * sched-ebb.c (rank): Use counts only.
4211 * shrink-wrap.c (handle_simple_exit): Use counts only.
4212 (try_shrink_wrapping): Use counts only.
4213 (place_prologue_for_one_component): Use counts only.
4214 * tracer.c (find_best_predecessor): Use to_frequency.
4215 (find_trace): Use to_frequency.
4216 (tail_duplicate): Use to_frequency.
4217 * trans-mem.c (expand_transaction): Do not update frequency.
4218 * tree-call-cdce.c: Do not update frequency.
4219 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
4220 (gimple_merge_blocks): Likewise.
4221 (gimple_split_edge): Likewise.
4222 (gimple_duplicate_sese_region): Likewise.
4223 (gimple_duplicate_sese_tail): Likewise.
4224 (move_sese_region_to_fn): Likewise.
4225 (gimple_account_profile_record): Likewise.
4226 (insert_cond_bb): Likewise.
4227 * tree-complex.c (expand_complex_div_wide): Likewise.
4228 * tree-eh.c (lower_resx): Update profile.
4229 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
4231 (initialize_cfun): Do not initialize frequencies
4232 (freqs_to_counts): Delete.
4233 (copy_cfg_body): Ignore count parameter.
4234 (copy_body): Update.
4235 (expand_call_inline): Update count_max.
4236 (optimize_inline_calls): Update count_max.
4237 (tree_function_versioning): Update count_max.
4238 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
4239 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
4241 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
4242 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
4243 (try_peel_loop): Likewise.
4244 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
4246 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
4247 (tree_transform_and_unroll_loop): Do not use frequencies
4248 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
4249 Use reliable prediction only.
4250 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
4251 * tree-ssa-sink.c (select_best_block): Use to_frequency.
4252 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
4253 probability scaling.
4254 * tree-ssa-threadupdate.c (create_block_for_threading): Do
4255 not update frequency
4256 (any_remaining_duplicated_blocks): Likewise.
4257 (update_profile): Likewise.
4258 (estimated_freqs_path): Delete.
4259 (freqs_to_counts_path): Delete.
4260 (clear_counts_path): Delete.
4261 (ssa_fix_duplicate_block_edges): Likewise.
4262 (duplicate_thread_path): Likewise.
4263 * tree-switch-conversion.c (gen_inbound_check): Use counts.
4264 * tree-tailcall.c (decrease_profile): Do not update frequency.
4265 (eliminate_tail_call): Likewise.
4266 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
4267 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
4268 (optimize_mask_stores): Likewise.
4269 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
4270 * ubsan.c (ubsan_expand_null_ifn): Update profile.
4271 (ubsan_expand_ptr_ifn): Update profile.
4272 * value-prof.c (gimple_ic): Simplify.
4273 * value-prof.h (gimple_ic): Update prototype.
4274 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
4275 * ipa-inline.c (compute_uninlined_call_time): Be sure that
4277 (want_inline_self_recursive_call_p): Likewise.
4278 (resolve_noninline_speculation): Only cummulate defined counts.
4279 (inline_small_functions): Use nonzero_p.
4280 (ipa_inline): Do not access freed node.
4282 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
4284 * config/aarch64/aarch64.c (aarch64_override_options_internal):
4285 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
4287 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
4289 * config/riscv/riscv.c (riscv_legitimize_move): Handle
4290 non-legitimate address.
4292 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
4294 * config/rs6000/rs6000.md (*lt0_disi): Delete.
4295 (*lt0_<mode>di, *lt0_<mode>si): New.
4297 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
4299 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
4300 TARGET_PAIRED_FLOAT.
4302 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
4303 Jim Wilson <jim.wilson@linaro.org>
4305 * config/aarch64/aarch64-cores.def (saphira): New CPU.
4306 * config/aarch64/aarch64-tune.md: Regenerated.
4307 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
4308 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
4310 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
4312 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
4313 (arc_expand_prologue): Restore blink for millicode.
4314 * config/arc/linux.h (LINK_EH_SPEC): Defined.
4316 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
4319 * config/i386/i386.c (ix86_vector_duplicate_value): Use
4320 gen_vec_duplicate after forcing the scalar into a register.
4322 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
4324 * combine (try_combine): Print the insns input to try_combine to the
4327 2017-11-02 Steve Ellcey <sellcey@cavium.com>
4330 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
4331 Remove second argument from aarch64_process_target_attr call.
4332 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
4334 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
4336 (aarch64_handle_attr_arch): Remove second argument.
4337 (aarch64_handle_attr_cpu): Ditto.
4338 (aarch64_handle_attr_tune): Ditto.
4339 (aarch64_handle_attr_isa_flags): Ditto.
4340 (aarch64_process_one_target_attr): Ditto.
4341 (aarch64_process_target_attr): Ditto.
4342 (aarch64_option_valid_attribute_p): Remove second argument.
4343 on aarch64_process_target_attr call.
4345 2017-11-02 David Malcolm <dmalcolm@redhat.com>
4347 * diagnostic.c: Include "selftest-diagnostic.h".
4348 (selftest::assert_location_text): New function.
4349 (selftest::test_diagnostic_get_location_text): New function.
4350 (selftest::diagnostic_c_tests): Call it.
4352 2017-11-02 David Malcolm <dmalcolm@redhat.com>
4354 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
4355 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
4356 (class selftest::test_diagnostic_context): Move to...
4357 * selftest-diagnostic.c: New file.
4358 * selftest-diagnostic.h: New file.
4360 2017-11-02 James Bowman <james.bowman@ftdichip.com>
4362 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
4363 offset range for FT32B.
4364 * config/ft32/ft32.h: option "mcompress" enables relaxation.
4365 * config/ft32/ft32.md: Add TARGET_NOPM.
4366 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
4367 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
4369 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
4371 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
4373 2017-11-02 Jeff Law <law@redhat.com>
4375 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
4376 virtual keyword on FINAL OVERRIDE members.
4378 * tree-ssa-propagate.h (ssa_propagation_engine): Group
4379 virtuals together. Add virtual destructor.
4380 (substitute_and_fold_engine): Similarly.
4382 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
4384 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
4386 2017-11-02 Richard Biener <rguenther@suse.de>
4388 PR tree-optimization/82795
4389 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
4391 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4393 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
4395 (gcc_GAS_CHECK_FEATURE): Remove.
4396 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
4397 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
4398 * configure: Regenerate.
4400 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
4402 * config/arc/arc.c (hwloop_optimize): Account for empty
4405 2017-11-02 Richard Biener <rguenther@suse.de>
4408 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
4409 Truncate ARRAY_REF index and element size.
4411 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
4413 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
4415 2017-11-01 Jeff Law <law@redhat.com>
4417 * tree-ssa-ccp.c (ccp_folder): New class derived from
4418 substitute_and_fold_engine.
4419 (ccp_folder::get_value): New member function.
4420 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
4421 (ccp_fold_stmt): Remove prototype.
4422 (ccp_finalize): Call substitute_and_fold from the ccp_class.
4423 * tree-ssa-copy.c (copy_folder): New class derived from
4424 substitute_and_fold_engine.
4425 (copy_folder::get_value): Renamed from get_value.
4426 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
4427 * tree-vrp.c (vrp_folder): New class derived from
4428 substitute_and_fold_engine.
4429 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
4430 (vrp_folder::get_value): New member function.
4431 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
4432 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
4433 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
4434 provide a class interface to folder/substitute routines.
4435 (ssa_prop_fold_stmt_fn): Remove typedef.
4436 (ssa_prop_get_value_fn): Likewise.
4437 (subsitute_and_fold): Remove prototype.
4438 (replace_uses_in): Likewise.
4439 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
4440 Renamed from replace_uses_in. Call the virtual member function
4441 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
4442 (substitute_and_fold_dom_walker): Remove initialization of
4443 data member entries for calbacks. Add substitute_and_fold_engine
4444 member and initialize it.
4445 (substitute_and_fold_dom_walker::before_dom_children0: Use the
4446 member functions for get_value, replace_phi_args_in c
4447 replace_uses_in, and fold_stmt calls.
4448 (substitute_and_fold_engine::substitute_and_fold): Renamed from
4449 substitute_and_fold. Remove assert. Update ctor call.
4451 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
4452 (ssa_prop_visit_phi_fn): Likewise.
4453 (class ssa_propagation_engine): New class to provide an interface
4455 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
4457 (ssa_prop_visit_phi): Likewise.
4458 (ssa_propagation_engine::simulate_stmt): Moved into class.
4459 Call visit_phi/visit_stmt from the class rather than via
4460 file scoped static variables.
4461 (ssa_propagation_engine::simulate_block): Moved into class.
4462 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
4463 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
4464 set file scoped statics for the visit_stmt/visit_phi callbacks.
4465 * tree-complex.c (complex_propagate): New class derived from
4466 ssa_propagation_engine.
4467 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
4468 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
4469 (tree_lower_complex): Call ssa_propagate via the complex_propagate
4471 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
4472 ssa_propagation_engine.
4473 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
4474 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
4475 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
4476 * tree-ssa-copy.c (copy_prop): New class derived from
4477 ssa_propagation_engine.
4478 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
4479 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
4480 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
4481 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
4482 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
4483 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
4484 (execute_vrp): Call ssa_propagate from the vrp_prop class.
4486 2017-11-01 Jakub Jelinek <jakub@redhat.com>
4488 PR rtl-optimization/82778
4489 PR rtl-optimization/82597
4490 * compare-elim.c (struct comparison): Add in_a_setter field.
4491 (find_comparison_dom_walker::before_dom_children): Remove killed
4492 bitmap and df_simulate_find_defs call, instead walk the defs.
4493 Compute last_setter and initialize in_a_setter. Merge definitions
4494 with first initialization for a few variables.
4495 (try_validate_parallel): Use insn_invalid_p instead of
4496 recog_memoized. Return insn rather than just the pattern.
4497 (try_merge_compare): Fix up comment. Don't uselessly test if
4498 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
4500 (execute_compare_elim_after_reload): Remove df_chain_add_problem
4503 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4504 Alan Hayward <alan.hayward@arm.com>
4505 David Sherwood <david.sherwood@arm.com>
4507 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
4508 aarch64_hard_regno_nregs to get the number of registers
4511 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4512 Alan Hayward <alan.hayward@arm.com>
4513 David Sherwood <david.sherwood@arm.com>
4515 * config/aarch64/constraints.md (Upl): Rename to...
4517 * config/aarch64/aarch64.md
4518 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
4521 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4522 Alan Hayward <alan.hayward@arm.com>
4523 David Sherwood <david.sherwood@arm.com>
4525 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
4526 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
4529 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4530 Alan Hayward <alan.hayward@arm.com>
4531 David Sherwood <david.sherwood@arm.com>
4533 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
4534 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
4535 (aarch64_evpc_dup): Generate rtl direcly, rather than using
4537 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
4538 of a single element.
4539 * config/aarch64/iterators.md: Add a comment above the permute
4540 unspecs to say that they are generated directly by
4541 aarch64_expand_vec_perm_const.
4542 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
4544 2017-11-01 Nathan Sidwell <nathan@acm.org>
4546 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
4548 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
4550 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
4551 and medany code models, and describe what they do.
4553 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4555 Revert accidental duplicate:
4557 * combine.c (can_change_dest_mode): Reject changes in
4558 REGMODE_NATURAL_SIZE.
4560 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
4562 PR rtl-optimization/64682
4563 PR rtl-optimization/69567
4564 PR rtl-optimization/69737
4565 PR rtl-optimization/82683
4566 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
4567 register mentioned in the note, drop the note, unless it came from I3,
4568 in which case it should go to I3 again.
4570 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4572 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
4573 and return false if not.
4574 (clear_bytes_written_by, live_bytes_read): Update accordingly.
4576 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4578 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
4579 range is known to be empty.
4581 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4582 Alan Hayward <alan.hayward@arm.com>
4583 David Sherwood <david.sherwood@arm.com>
4585 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
4586 and CONST_VECTOR_NUNITS instead of computing the number of units from
4587 the byte sizes of the vector and element.
4588 (simplify_binary_operation_1): Likewise.
4589 (simplify_const_binary_operation): Likewise.
4590 (simplify_ternary_operation): Likewise.
4592 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4593 Alan Hayward <alan.hayward@arm.com>
4594 David Sherwood <david.sherwood@arm.com>
4596 * var-tracking.c (INT_MEM_OFFSET): Replace with...
4597 (int_mem_offset): ...this new function.
4598 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
4599 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
4600 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
4603 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4604 Alan Hayward <alan.hayward@arm.com>
4605 David Sherwood <david.sherwood@arm.com>
4607 * lower-subreg.c (interesting_mode_p): New function.
4608 (compute_costs, find_decomposable_subregs, decompose_register)
4609 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
4610 (resolve_clobber, dump_choices): Use it.
4612 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4613 Alan Hayward <alan.hayward@arm.com>
4614 David Sherwood <david.sherwood@arm.com>
4616 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
4618 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4619 Alan Hayward <alan.hayward@arm.com>
4620 David Sherwood <david.sherwood@arm.com>
4622 * alias.c (find_base_value, find_base_term): Only process integer
4623 truncations. Check the precision rather than the size.
4625 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4626 Alan Hayward <alan.hayward@arm.com>
4627 David Sherwood <david.sherwood@arm.com>
4629 * machmode.h (is_narrower_int_mode): New function
4630 * optabs.c (expand_float, expand_fix): Use it.
4631 * dwarf2out.c (rotate_loc_descriptor): Likewise.
4633 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4634 Alan Hayward <alan.hayward@arm.com>
4635 David Sherwood <david.sherwood@arm.com>
4637 * rtl.h (narrower_subreg_mode): New function.
4638 * ira-color.c (update_costs_from_allocno): Use it.
4640 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4641 Alan Hayward <alan.hayward@arm.com>
4642 David Sherwood <david.sherwood@arm.com>
4644 * optabs-query.h (convert_optab_p): New function, split out from...
4645 (convert_optab_handler): ...here.
4646 (widening_optab_handler): Delete.
4647 (find_widening_optab_handler): Remove permit_non_widening parameter.
4648 (find_widening_optab_handler_and_mode): Likewise. Provide an
4649 override that operates on mode class wrappers.
4650 * optabs-query.c (widening_optab_handler): Delete.
4651 (find_widening_optab_handler_and_mode): Remove permit_non_widening
4652 parameter. Assert that the two modes are the same class and that
4653 the "from" mode is narrower than the "to" mode. Use
4654 convert_optab_handler instead of widening_optab_handler.
4655 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
4656 instead of widening_optab_handler.
4657 * expr.c (expand_expr_real_2): Update calls to
4658 find_widening_optab_handler.
4659 * optabs.c (expand_widen_pattern_expr): Likewise.
4660 (expand_binop_directly): Take the insn_code as a parameter.
4661 (expand_binop): Only call find_widening_optab_handler for
4662 conversion optabs; use optab_handler otherwise. Update calls
4663 to find_widening_optab_handler and expand_binop_directly.
4664 Use convert_optab_handler instead of widening_optab_handler.
4665 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
4666 find_widening_optab_handler and use scalar_mode rather than
4668 (convert_plusminus_to_widen): Likewise.
4670 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4671 Alan Hayward <alan.hayward@arm.com>
4672 David Sherwood <david.sherwood@arm.com>
4674 * machmode.h (fixed_size_mode): New class.
4675 * rtl.h (get_pool_mode): Return fixed_size_mode.
4676 * gengtype.c (main): Add fixed_size_mode.
4677 * target.def (get_raw_result_mode): Return a fixed_size_mode.
4678 (get_raw_arg_mode): Likewise.
4679 * doc/tm.texi: Regenerate.
4680 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
4681 * targhooks.c (default_get_reg_raw_mode): Likewise.
4682 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
4683 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
4684 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
4685 (msp430_get_raw_result_mode): Likewise.
4686 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
4687 * dbxout.c (dbxout_parms): Require fixed-size modes.
4688 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
4689 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
4690 * omp-low.c (lower_oacc_reductions): Likewise.
4691 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
4692 (simplify_subreg): Update accordingly.
4693 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
4694 (force_const_mem): Update accordingly. Return NULL_RTX for modes
4695 that aren't fixed-size.
4696 (get_pool_mode): Return a fixed_size_mode.
4697 (output_constant_pool_2): Take a fixed_size_mode.
4699 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4700 Alan Hayward <alan.hayward@arm.com>
4701 David Sherwood <david.sherwood@arm.com>
4703 * doc/rtl.texi (vec_series): Document.
4704 (const): Say that the operand can be a vec_series.
4705 * rtl.def (VEC_SERIES): New rtx code.
4706 * rtl.h (const_vec_series_p_1): Declare.
4707 (const_vec_series_p): New function.
4708 * emit-rtl.h (gen_const_vec_series): Declare.
4709 (gen_vec_series): Likewise.
4710 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
4711 (gen_vec_series): Likewise.
4712 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
4713 * simplify-rtx.c (simplify_unary_operation): Handle negations
4715 (simplify_binary_operation_series): New function.
4716 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
4717 (test_vector_ops_series): New function.
4718 (test_vector_ops): Call it.
4719 * config/powerpcspe/altivec.md (altivec_lvsl): Use
4720 gen_const_vec_series.
4721 (altivec_lvsr): Likewise.
4722 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
4724 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4725 Alan Hayward <alan.hayward@arm.com>
4726 David Sherwood <david.sherwood@arm.com>
4728 * doc/rtl.texi (const): Update description of address constants.
4729 Say that vector constants are allowed too.
4730 * common.md (E, F): Use CONSTANT_P instead of checking for
4732 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
4733 checking for CONST_VECTOR.
4734 * expmed.c (make_tree): Use build_vector_from_val for a CONST
4736 * expr.c (expand_expr_real_2): Check for vector modes instead
4737 of checking for CONST_VECTOR.
4738 * rtl.h (const_vec_p): New function.
4739 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
4740 (unwrap_const_vec_duplicate): Handle them here too.
4742 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4743 David Malcolm <dmalcolm@redhat.com>
4744 Alan Hayward <alan.hayward@arm.com>
4745 David Sherwood <david.sherwood@arm.com>
4747 * rtl.h (vec_duplicate_p): New function.
4748 * selftest-rtl.c (assert_rtx_eq_at): New function.
4749 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
4750 (assert_rtx_eq_at): Declare.
4751 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
4752 * selftest-run-tests.c (selftest::run_tests): Call it.
4753 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
4754 (simplify_unary_operation_1): Recursively handle vector duplicates.
4755 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
4757 (simplify_subreg): Handle subregs of vector duplicates.
4758 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
4759 (selftest::simplify_rtx_c_tests): New functions.
4761 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4762 Alan Hayward <alan.hayward@arm.com>
4763 David Sherwood <david.sherwood@arm.com>
4765 * emit-rtl.h (gen_const_vec_duplicate): Declare.
4766 (gen_vec_duplicate): Likewise.
4767 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
4769 (gen_const_vector): ...here.
4770 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
4771 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
4772 whose elements are all equal.
4773 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
4774 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
4775 (simplify_relational_operation): Likewise.
4776 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
4778 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
4779 (aarch64_expand_vector_init): Likewise.
4780 * config/arm/arm.c (neon_vdup_constant): Likewise.
4781 (neon_expand_vector_init): Likewise.
4782 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
4783 (arm_block_set_unaligned_vect): Likewise.
4784 (arm_block_set_aligned_vect): Likewise.
4785 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
4786 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
4787 (expand_vec_perm_even_odd_pack): Likewise.
4788 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
4789 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
4790 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
4791 gen_const_vec_duplicate.
4792 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
4793 * config/mips/mips.c (mips_gen_const_int_vector): Use
4794 gen_const_vec_duplicate.
4795 (mips_expand_vector_init): Use CONST0_RTX.
4796 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
4797 (define_split): Use gen_const_vec_duplicate.
4798 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
4799 (define_split): Use gen_const_vec_duplicate.
4800 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
4801 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
4802 * config/spu/spu.c (spu_const): Likewise.
4804 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4805 Alan Hayward <alan.hayward@arm.com>
4806 David Sherwood <david.sherwood@arm.com>
4808 * combine.c (can_change_dest_mode): Reject changes in
4809 REGMODE_NATURAL_SIZE.
4811 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
4813 * configure.ac (--enable-libssp): New.
4814 (gcc_cv_libc_provides_ssp): Check for explicit setting before
4815 trying to determine target-specific default. Adjust indentation.
4816 * configure: Regenerated.
4817 * doc/install.texi (Configuration): Expand --disable-libssp
4820 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
4822 config/i386/i386.c (ix86_expand_epilogue): Correct stack
4825 2017-10-31 Martin Jambor <mjambor@suse.cz>
4828 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
4830 2017-10-31 David Malcolm <dmalcolm@redhat.com>
4832 * auto-profile.c (autofdo_source_profile::read): Use
4833 UNKNOWN_LOCATION rather than 0.
4834 * diagnostic-core.h (warning_at_rich_loc): Rename to...
4835 (warning_at): ...this overload.
4836 (warning_at_rich_loc_n): Rename to...
4837 (warning_n): ...this overload.
4838 (error_at_rich_loc): Rename to...
4839 (error_at): ...this overload.
4840 (pedwarn_at_rich_loc): Rename to...
4841 (pedwarn): ...this overload.
4842 (permerror_at_rich_loc): Rename to...
4843 (permerror): ...this overload.
4844 (inform_at_rich_loc): Rename to...
4845 (inform): ...this overload.
4846 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
4847 (diagnostic_n_impl_richloc): Rename to...
4848 (diagnostic_n_impl): ...this rich_location *-based decl.
4849 (inform_at_rich_loc): Rename to...
4850 (inform): ...this, and add an assertion.
4851 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
4852 (warning_at_rich_loc): Rename to...
4853 (warning_at): ...this, and add an assertion.
4854 (warning_at_rich_loc_n): Rename to...
4855 (warning_n): ...this, and add an assertion.
4856 (warning_n): Update location_t-based implementation for removal of
4857 location_t-based diagnostic_n_impl.
4858 (pedwarn_at_rich_loc): Rename to...
4859 (pedwarn): ...this, and add an assertion.
4860 (permerror_at_rich_loc): Rename to...
4861 (permerror): ...this, and add an assertion.
4862 (error_n): Update for removal of location_t-based diagnostic_n_impl.
4863 (error_at_rich_loc): Rename to...
4864 (error_at): ...this, and add an assertion.
4865 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
4866 (driver::do_spec_on_infiles): Likewise.
4867 * substring-locations.c (format_warning_va): Update for renaming
4868 of inform_at_rich_loc.
4870 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
4872 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
4873 _Float<N>X built-in functions so that the variant without the
4874 "__builtin_" prefix is only enabled for the GNU C and Objective C
4875 languages when they are in non-strict ANSI/ISO mode.
4876 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
4877 * target.def (floatn_builtin_p): Add a target hook to control
4878 whether _Float<N> and _Float<N>X built-in functions without the
4879 "__builtin_" prefix are enabled, and return true for C and
4880 Objective C in the default hook. Include langhooks.h in
4882 * targhooks.h (default_floatn_builtin_p): Likewise.
4883 * targhooks.c (default_floatn_builtin_p): Likewise.
4884 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
4885 floatn_builtin_p target hook.
4886 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
4888 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
4889 Eric Botcazou <ebotcazou@adacore.com>
4891 PR rtl-optimization/81803
4892 * lra-constraints.c (curr_insn_transform): Also reload the whole
4893 register for a strict subreg no wider than a word if this is for
4894 a WORD_REGISTER_OPERATIONS target.
4896 2017-10-31 Jason Merrill <jason@redhat.com>
4898 * gdbinit.in: Skip over inlines from timevar.h.
4900 2017-10-31 Martin Liska <mliska@suse.cz>
4902 * doc/gcov.texi: Document new option.
4903 * gcov.c (print_usage): Likewise print it.
4904 (process_args): Support the argument.
4905 (format_count): New function.
4906 (format_gcov): Use the function.
4908 2017-10-31 Martin Liska <mliska@suse.cz>
4910 * gcov.c (struct name_map): do not use typedef.
4911 Define operator== and operator<.
4912 (name_search): Remove.
4913 (name_sort): Remove.
4914 (main): Do not allocate names.
4915 (process_file): Add vertical space.
4916 (generate_results): Use std::find.
4917 (release_structures): Do not release memory.
4918 (find_source): Use std::find.
4920 2017-10-31 Martin Liska <mliska@suse.cz>
4922 * gcov.c (struct line_info): Remove it's typedef.
4923 (line_info::line_info): Add proper ctor.
4924 (line_info::has_block): Do not use a typedef.
4925 (struct source_info): Do not use typedef.
4926 (circuit): Likewise.
4927 (get_cycles_count): Likewise.
4928 (output_intermediate_file): Iterate via vector iterator.
4929 (add_line_counts): Use std::vector methods.
4930 (accumulate_line_counts): Likewise.
4931 (output_lines): Likewise.
4933 2017-10-31 Martin Liska <mliska@suse.cz>
4935 * gcov.c (struct source_info): Remove typedef.
4936 (source_info::source_info): Add proper ctor.
4937 (accumulate_line_counts): Use struct, not it's typedef.
4938 (output_gcov_file): Likewise.
4939 (output_lines): Likewise.
4940 (main): Do not allocate an array.
4941 (output_intermediate_file): Use size of vector container.
4942 (process_file): Resize the vector.
4943 (generate_results): Do not preallocate, use newly added vector
4945 (release_structures): Do not release sources.
4946 (find_source): Use vector methods.
4947 (add_line_counts): Do not use typedef.
4949 2017-10-31 Martin Liska <mliska@suse.cz>
4951 * doc/gcov.texi: Document that.
4952 * gcov.c (add_line_counts): Mark lines with a non-executed
4954 (output_line_beginning): Handle such lines.
4955 (output_lines): Pass new argument.
4956 (output_intermediate_file): Print it in intermediate format.
4958 2017-10-31 Martin Liska <mliska@suse.cz>
4960 * color-macros.h: New file.
4961 * diagnostic-color.c: Factor out color related to macros to
4963 * doc/gcov.texi: Document -k option.
4964 * gcov.c (INCLUDE_STRING): Include string.h.
4965 (print_usage): Add -k option.
4966 (process_args): Parse it.
4967 (pad_count_string): New function.
4968 (output_line_beginning): Likewise.
4969 (DEFAULT_LINE_START): New macro.
4970 (output_lines): Support color output.
4972 2017-10-31 Martin Liska <mliska@suse.cz>
4974 PR gcov-profile/82633
4975 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
4976 their interaction with GCOV infrastructure.
4977 * configure.ac: Add -fkeep-{inline,static}-functions to
4979 * configure: Regenerate.
4981 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
4984 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
4986 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
4989 * config/rs6000/rs6000.md (allocate_stack): Force update interval
4990 into a register if it does not fit into an immediate offset field.
4992 2017-10-31 Olivier Hainque <hainque@adacore.com>
4994 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
4996 2017-10-31 Julia Koval <julia.koval@intel.com>
4998 * config.gcc: Add gfniintrin.h.
4999 * config/i386/gfniintrin.h: New.
5000 * config/i386/i386-builtin-types.def
5001 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
5002 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
5003 __builtin_ia32_vgf2p8affineinvqb_v32qi,
5004 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
5005 __builtin_ia32_vgf2p8affineinvqb_v16qi,
5006 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
5007 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
5008 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
5009 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
5010 V64QI_FTYPE_V64QI_V64QI_INT): New types.
5011 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5012 * config/i386/immintrin.h: Include gfniintrin.h.
5013 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
5015 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
5017 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
5019 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
5021 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
5022 (arm_ashldi3_1bit): Remove pattern.
5023 (ashrdi3): Remove shift by 1 expansion.
5024 (arm_ashrdi3_1bit): Remove pattern.
5025 (lshrdi3): Remove shift by 1 expansion.
5026 (arm_lshrdi3_1bit): Remove pattern.
5027 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
5028 cost of ashldi3 by 1.
5029 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
5030 (<shift>di3_neon): Likewise.
5032 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5034 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
5035 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
5036 and (*aarch64_simd_mov<VQ:mode>).
5037 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
5038 pattern alternative.
5039 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
5040 attributes to match pattern alternative.
5042 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
5044 * config.gcc (powerpc*-*-*): Add emmintrin.h.
5045 * config/rs6000/emmintrin.h: New file.
5046 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
5048 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
5050 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
5051 * (movdi_vfp_cortexa8): Remove pattern.
5053 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5055 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
5057 (Specific, alpha*-dec-osf5.1): Remove.
5058 (Specific, mips-sgi-irix5): Remove.
5059 (Specific, mips-sgi-irix6): Remove.
5061 2017-10-30 Jakub Jelinek <jakub@redhat.com>
5064 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
5065 arguments to clear_bit_region_be.
5067 2017-10-30 Jim Wilson <wilson@tuliptree.org>
5069 * gimplify.c: Include memmodel.h.
5071 2017-10-30 Martin Jambor <mjambor@suse.cz>
5073 * omp-grid.c (grid_attempt_target_gridification): Also insert a
5074 condition whether loop should be executed at all.
5076 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
5078 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
5079 gimple folding of vec_madd() intrinsics.
5080 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
5081 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
5082 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
5084 2017-10-30 Richard Biener <rguenther@suse.de>
5086 PR tree-optimization/82762
5088 2017-10-23 Richard Biener <rguenther@suse.de>
5090 PR tree-optimization/82129
5092 2017-08-01 Richard Biener <rguenther@suse.de>
5094 PR tree-optimization/81181
5095 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5096 (compute_antic): ... end of iteration here.
5098 2017-10-30 Joseph Myers <joseph@codesourcery.com>
5100 * doc/invoke.texi (C Dialect Options): Document -std=c17,
5101 -std=iso9899:2017 and -std=gnu17.
5102 * doc/standards.texi (C Language): Document C17 support.
5103 * doc/cpp.texi (Overview): Mention -std=c17.
5104 (Standard Predefined Macros): Document C11 and C17 values of
5105 __STDC_VERSION__. Do not refer to C99 support as incomplete.
5106 * doc/extend.texi (Inline): Do not list individual options for
5107 standards newer than C99.
5108 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
5110 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
5113 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
5115 * asan.c (asan_finish_file): Align asan globals array by shadow
5118 2017-10-30 Jakub Jelinek <jakub@redhat.com>
5121 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
5122 (struct store_immediate_info): Add bitregion_start and bitregion_end
5124 (store_immediate_info::store_immediate_info): Add brs and bre
5125 arguments and initialize bitregion_{start,end} from those.
5126 (struct merged_store_group): Add bitregion_start, bitregion_end,
5127 align_base and mask fields. Drop unnecessary struct keyword from
5128 struct store_immediate_info. Add do_merge method.
5129 (clear_bit_region_be): Use memset instead of loop storing zeros.
5130 (merged_store_group::do_merge): New method.
5131 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
5132 stores as long as the surrounding bitregions have no gaps.
5133 (merged_store_group::merge_overlapping): Use do_merge.
5134 (merged_store_group::apply_stores): Test that bitregion_{start,end}
5135 is byte aligned, rather than requiring that start and width are
5136 byte aligned. Drop unnecessary struct keyword from
5137 struct store_immediate_info. Allocate and populate also mask array.
5138 Make start of the arrays relative to bitregion_start rather than
5139 start and size them according to bitregion_{end,start} difference.
5140 (struct imm_store_chain_info): Drop unnecessary struct keyword from
5141 struct store_immediate_info.
5142 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
5143 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
5144 struct keyword from struct store_immediate_info.
5145 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
5146 between stores as long as the surrounding bitregions have no gaps.
5148 (struct split_store): Add orig non-static data member.
5149 (split_store::split_store): Initialize orig to false.
5150 (find_constituent_stmts): Return store_immediate_info *, non-NULL
5151 if there is exactly a single original stmt. Change stmts argument
5152 to pointer from reference, if NULL, don't push anything to it. Add
5153 first argument, use it to optimize skipping over orig stmts that
5154 are known to be before bitpos already. Simplify.
5155 (split_group): Return unsigned int count how many stores are or
5156 would be needed rather than a bool. Add allow_unaligned argument.
5157 Change split_stores argument from reference to pointer, if NULL,
5158 only do a dry run computing how many stores would be produced.
5159 Rewritten algorithm to use both alignment and misalign if
5160 !allow_unaligned and handle bitfield stores with gaps.
5161 (imm_store_chain_info::output_merged_store): Set start_byte_pos
5162 from bitregion_start instead of start. Compute allow_unaligned
5163 here, if true, do 2 split_group dry runs to compute which one
5164 produces fewer stores and prefer aligned if equal. Punt if
5165 new count is bigger or equal than original before emitting any
5166 statements, rather than during that. Remove no longer needed
5167 new_ssa_names tracking. Replace num_stmts with
5168 split_stores.length (). Use 32-bit stack allocated entries
5169 in split_stores auto_vec. Try to reuse original store lhs/rhs1
5170 if possible. Handle bitfields with gaps.
5171 (pass_store_merging::execute): Ignore bitsize == 0 stores.
5172 Compute bitregion_{start,end} for the stores and construct
5173 store_immediate_info with that. Formatting fixes.
5175 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
5178 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
5179 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
5181 2017-10-29 Jim Wilson <wilson@tuliptree.org>
5183 * gimplify.c: Include tm_p.h.
5185 * common.opt (gcoff): Re-add as ignored option.
5186 (gcoff1, gcoff2, gcoff3): Likewise.
5188 * Makefile.in (OBJS): Delete sdbout.o.
5189 (GTFILES): Delete $(srcdir)/sdbout.c.
5190 * debug.h: Delete sdb_debug_hooks.
5191 * final.c: Delete sdbout.h include.
5192 (final_scan_insn): Delete SDB_DEBUG check.
5193 (rest_of_clean_state): Likewise.
5194 * output.h: Delete sdb_begin_function_line.
5197 * toplev.c: Delete sdbout.h include.
5198 (process_options): Delete SDB_DEBUG check.
5199 * tree-core.h (tree_type_common): Delete pointer field of
5201 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
5202 TYPE_SYMTAB_POINTER.
5203 * tree.h (TYPE_SYMTAB_POINTER): Delete.
5204 (TYPE_SYMTAB_IS_POINTER): Delete.
5205 (TYPE_SYMTAB_IS_DIE): Renumber.
5206 * xcoffout.c: Refer to former sdbout.c file.
5207 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
5209 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
5210 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
5211 (Debugging Options): Delete -gcoff.
5212 (-gstabs): Delete SDB reference.
5214 (-gcoff@var{level}): Delete.
5215 * doc/passes.texi (Debugging information output): Delete SDB and
5216 sdbout.c references.
5217 * doc/tm.texi: Regenerate.
5218 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
5219 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
5221 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
5222 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
5223 (SDB_DEBUGGING_INFO): Delete.
5224 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
5225 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
5226 * target.def (output_source_filename): Delete COFF reference.
5228 * common.opt (gcoff): Delete.
5229 (gxcoff+): Update Negative chain.
5230 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
5232 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
5233 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
5234 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
5236 (expand_function_start): Change sdb reference to past tense.
5237 (expand_function_end): Change sdb reference to past tense.
5238 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
5239 * opts.c (debug_type_names): Delete coff entry.
5240 (common_handle_option): Delete OPT_gcoff case.
5241 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
5243 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
5244 * config/cris/cris.h: Delete SDB reference in comment.
5245 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
5246 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
5247 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
5248 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
5250 (ix86_expand_prologue): Likewise.
5251 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
5252 * config/ia64/ia64.h: Likewise.
5253 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
5254 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
5256 * config/mmix/mmix.h: Likewise.
5257 * config/nds32/nds32.c: Likewise.
5258 * config/stormy/storym16.h: Likewise.
5259 * config/visium/visium.h: Likewise.
5260 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
5262 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
5264 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
5265 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
5266 FRAME_POINTER_REGNUM point at high end of local var area.
5268 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
5270 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
5271 Move comment around. Do not reset best_edge for a copiable
5272 destination if the copy would cause a partition change.
5273 (better_edge_p): Remove redundant check.
5275 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
5277 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
5279 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5281 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
5282 for math functions that have _Float<N> and _Float<N>X variants.
5283 (mathfn_built_in_2): Add support for math functions that have
5284 _Float<N> and _Float<N>X variants.
5285 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
5286 (expand_builtin_mathfn_ternary): Add support for fma with
5287 _Float<N> and _Float<N>X variants.
5288 (expand_builtin): Likewise.
5289 (fold_builtin_3): Likewise.
5290 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
5291 create math function _Float<N> and _Float<N>X variants as external
5293 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
5294 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
5295 the __builtin_ prefix and if not strict ansi, without the prefix.
5296 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
5297 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
5298 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
5299 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
5300 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
5301 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
5302 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
5303 function signatures for fma _Float<N> and _Float<N>X variants.
5304 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
5305 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
5306 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
5307 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
5308 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
5309 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
5310 * gencfn-macros.c (print_case_cfn): Add support for math functions
5311 that have _Float<N> and _Float<N>X variants.
5312 (print_define_operator_list): Likewise.
5313 (fltfn_suffixes): Likewise.
5315 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
5316 for math functions that have _Float<N> and _Float<N>X variants.
5317 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
5318 and _Float<N>X variants.
5319 (COPYSIGN): Likewise.
5322 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
5323 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
5325 (integer_valued_read_call_p): Likewise.
5326 * fold-const-call.c (fold_const_call_ss): Likewise.
5327 (fold_const_call_sss): Add support for copysign, fmin, and fmax
5328 _Float<N> and _Float<N>X variants.
5329 (fold_const_call_ssss): Add support for fma _Float<N> and
5330 _Float<N>X variants.
5331 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
5332 support for copysign and fma _Float<N> and _Float<N>X variants.
5333 (backprop::process_builtin_call_use): Likewise.
5334 * tree-call-cdce.c (can_test_argument_range); Add support for
5335 sqrt _Float<N> and _Float<N>X variants.
5336 (edom_only_function): Likewise.
5337 (get_no_error_domain): Likewise.
5338 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
5339 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
5340 copysign _Float<N> and _Float<N>X variants.
5341 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
5342 handled by machine independent code.
5343 (FMAF128): Likewise.
5344 * doc/cpp.texi (Common Predefined Macros): Document defining
5345 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
5346 fma _Float<N> and _Float<N>X variants.
5348 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
5351 * config/i386/i386-modes.def (CCFPU): Remove definition.
5352 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
5353 (ix86_cc_modes_compatible): Ditto.
5354 (ix86_expand_carry_flag_compare): Ditto.
5355 (ix86_expand_int_movcc): Ditto.
5356 (ix86_expand_int_addcc): Ditto.
5357 (ix86_reverse_condition): Ditto.
5358 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
5359 Return true/false for unordered/ordered fp comparisons.
5360 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
5361 (ix86_prepare_fp_compare_args): Update for rename.
5362 (ix86_expand_fp_compare): Update for rename. Generate unordered
5363 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
5364 (ix86_expand_sse_compare_and_jump): Ditto.
5365 * config/i386/predicates.md (fcmov_comparison_operator):
5366 Remove CCFPU mode handling.
5367 (ix86_comparison_operator): Ditto.
5368 (ix86_carry_flag_operator): Ditto.
5369 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
5370 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
5371 (*cmpu<mode>_cc_i387): Ditto.
5372 (FPCMP): Remove mode iterator.
5373 (unord): Remove mode attribute.
5374 (unord_subst): New define_subst transformation
5375 (unord): New define_subst attribute.
5377 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
5378 (*cmpi<unord>xf_i387): Ditto.
5379 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
5380 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
5381 using unord_subst transformation.
5382 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
5383 (round_saeonly): Also handle CCFP mode.
5384 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
5385 Remove UNSPEC_SAHF unspec handling.
5387 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
5389 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
5391 2017-10-27 Jeff Law <law@redhat.com>
5393 * gimple-ssa-sprintf.c: Include domwalk.h.
5394 (class sprintf_dom_walker): New class, derived from dom_walker.
5395 (sprintf_dom_walker::before_dom_children): New function.
5396 (struct call_info): Moved into sprintf_dom_walker class
5397 (compute_formath_length, handle_gimple_call): Likewise.
5398 (sprintf_length::execute): Call the dominator walker rather
5399 than walking the statements.
5401 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
5402 gimple statement locations.
5403 (check_array_bounds): Corresponding changes. Get the statement's
5404 location directly from wi->stmt.
5406 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
5409 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
5411 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
5413 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
5414 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
5416 2017-10-27 Jakub Jelinek <jakub@redhat.com>
5419 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
5420 * config/i386/i386.c (maybe_get_pool_constant): Removed.
5421 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
5422 maybe_get_pool_constant.
5423 * config/i386/predicates.md (zero_extended_scalar_load_operand):
5426 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5428 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
5429 2.26 caveat. Update gas and gld versions.
5430 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
5433 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5435 * cgraph.h (set_malloc_flag): Declare.
5436 * cgraph.c (set_malloc_flag_1): New function.
5437 (set_malloc_flag): Likewise.
5438 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
5439 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
5441 (read_ipa_call_summary): Add support for reading is_return_callee.
5442 (write_ipa_call_summary): Stream is_return_callee.
5443 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
5444 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
5445 ipa-prop.h, ipa-fnsummary.h.
5446 (pure_const_names): Change to static.
5447 (malloc_state_e): Define.
5448 (malloc_state_names): Define.
5449 (funct_state_d): Add field malloc_state.
5450 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
5451 (check_retval_uses): New function.
5452 (malloc_candidate_p): Likewise.
5453 (analyze_function): Add support for malloc attribute.
5454 (pure_const_write_summary): Stream malloc_state.
5455 (pure_const_read_summary): Add support for reading malloc_state.
5456 (dump_malloc_lattice): New function.
5457 (propagate_malloc): New function.
5458 (warn_function_malloc): New function.
5459 (ipa_pure_const::execute): Call propagate_malloc and
5460 ipa_free_fn_summary.
5461 (pass_local_pure_const::execute): Add support for malloc attribute.
5462 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
5463 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
5465 2017-10-27 Martin Liska <mliska@suse.cz>
5467 PR gcov-profile/82457
5468 * doc/invoke.texi: Document that one needs a non-strict ISO mode
5469 for fork-like functions to be properly instrumented.
5471 2017-10-27 Richard Biener <rguenther@suse.de>
5474 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
5475 info when we redirected EH.
5477 2017-10-26 Michael Collison <michael.collison@arm.com>
5479 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
5481 (<optab>_trunchf<GPI:mode>2: New pattern.
5482 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
5483 * config/aarch64/iterators.md (wv): New mode attribute.
5484 (vf, VF): New mode attributes.
5485 (vgp, VGP): New mode attributes.
5486 (s): Update attribute with SImode and DImode prefixes.
5488 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
5490 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
5491 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
5492 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
5493 (nios2_option_overide): Initialize it. Don't allow R0-relative
5494 addressing with PIC.
5495 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
5496 (nios2_symbolic_constant_p): Likewise.
5497 (nios2_legitimate_address_p): Likewise.
5498 (nios2_r0rel_section_name_p): New.
5499 (nios2_symbol_ref_in_r0rel_data_p): New.
5500 (nios2_emit_move_sequence): Handle r0rel_constant_p.
5501 (r0rel_constant_p): New.
5502 (nios2_print_operand_address): Handle r0rel_constant_p.
5503 (nios2_cdx_narrow_form_p): Likewise.
5504 * config/nios2/nios2.opt (mr0rel-sec=): New option.
5505 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
5506 (Nios II Options): Document -mr0rel-sec.
5508 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
5510 * config/nios2/nios2.c: Include xregex.h.
5511 (nios2_gprel_sec_regex): New.
5512 (nios2_option_overide): Initialize it. Don't allow GP-relative
5513 addressing with PIC.
5514 (nios2_small_section_name_p): Check for regex match.
5515 * config/nios2/nios2.opt (mgprel-sec=): New option.
5516 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
5517 (Nios II Options): Document -mgprel-sec.
5519 2017-10-26 Jim Wilson <wilson@tuliptree.org>
5521 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
5523 2017-10-26 Tom de Vries <tom@codesourcery.com>
5525 PR tree-optimization/82707
5526 * gimple.c (gimple_copy): Fix unsharing of
5527 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
5529 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
5531 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
5532 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
5533 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
5534 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
5535 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
5536 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
5537 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
5538 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
5539 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
5540 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
5541 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
5542 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
5543 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
5544 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
5545 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
5546 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
5547 _mm512_mask_cmpunord_ps_mask): New intrinsics.
5549 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5551 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
5553 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
5554 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
5555 warning to rs6000.c. Remove the Undocumented flag, since it has
5557 (-mabi=ibmlongdouble): Likewise.
5558 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
5559 already set, set the default format for long double.
5560 (rs6000_debug_reg_global): Print whether long double is IBM or
5562 (rs6000_option_override_internal): Rework setting long double
5563 format. Only warn if the user is changing the long double default
5564 and they did not use -Wno-psabi.
5565 * doc/invoke.texi (PowerPC options): Update the documentation for
5566 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
5568 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5569 Alan Hayward <alan.hayward@arm.com>
5570 David Sherwood <david.sherwood@arm.com>
5572 * rtl.h (wider_subreg_mode): New function.
5573 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
5574 rather than an unsigned int *.
5575 * ira-color.c (regno_max_ref_width): Replace with...
5576 (regno_max_ref_mode): ...this new variable.
5577 (coalesced_pseudo_reg_slot_compare): Update accordingly.
5578 Use wider_subreg_mode.
5579 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
5580 rather than an unsigned int *.
5581 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
5582 (process_alt_operands): Likewise.
5583 (invariant_p): Likewise.
5584 * lra-spills.c (assign_mem_slot): Likewise.
5585 (add_pseudo_to_slot): Likewise.
5586 * lra.c (collect_non_operand_hard_regs): Likewise.
5587 (add_regs_to_insn_regno_info): Likewise.
5588 * reload1.c (regno_max_ref_width): Replace with...
5589 (regno_max_ref_mode): ...this new variable.
5590 (reload): Update accordingly. Update call to
5591 ira_sort_regnos_for_alter_reg.
5592 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
5593 (init_eliminable_invariants): Update to use regno_max_ref_mode.
5594 (scan_paradoxical_subregs): Likewise.
5596 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
5598 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
5599 (aarch64_frame): Add emit_frame_chain boolean.
5600 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
5601 Move eh_return case to aarch64_layout_frame.
5602 (aarch64_layout_frame): Initialize emit_frame_chain.
5603 (aarch64_expand_prologue): Use emit_frame_chain.
5605 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
5607 * config/aarch64/aarch64.c (aarch64_layout_frame):
5608 Ensure LR is always stored at the bottom of the callee-saves.
5609 Remove rarely used frame layout which saves callee-saves at top of
5610 frame, so the store of LR can be used as a valid probe in all cases.
5612 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
5614 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
5615 Improve unaligned TImode/TFmode base/offset split.
5617 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5618 Alan Hayward <alan.hayward@arm.com>
5619 David Sherwood <david.sherwood@arm.com>
5621 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
5622 * combine.c (find_single_use_1): Likewise.
5623 (expand_field_assignment): Likewise.
5624 (move_deaths): Likewise.
5625 * lra-constraints.c (simplify_operand_subreg): Likewise.
5626 (curr_insn_transform): Likewise.
5627 * lra.c (collect_non_operand_hard_regs): Likewise.
5628 (add_regs_to_insn_regno_info): Likewise.
5629 * rtlanal.c (reg_referenced_p): Likewise.
5630 (covers_regno_no_parallel_p): Likewise.
5632 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5634 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
5635 Don't print any bits outside the precision of the value.
5636 * wide-int.cc (test_printing): Add some new tests.
5638 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5640 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
5641 supports -xbrace_comment option.
5642 * configure: Regenerate.
5643 * config.in: Regenerate.
5644 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
5645 (ASM_CPU_SPEC): Use it.
5647 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5649 * target.def (static_rtx_alignment): New hook.
5650 * targhooks.h (default_static_rtx_alignment): Declare.
5651 * targhooks.c (default_static_rtx_alignment): New function.
5652 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
5653 * doc/tm.texi: Regenerate.
5654 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
5655 instead of targetm.constant_alignment. Remove call to
5657 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5658 (cris_preferred_mininum_alignment): New function, split out from...
5659 (cris_constant_alignment): ...here.
5660 (cris_static_rtx_alignment): New function.
5661 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
5663 (ix86_constant_alignment): ...here.
5664 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5665 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5666 (mmix_static_rtx_alignment): New function.
5667 * config/spu/spu.c (spu_static_rtx_alignment): New function.
5668 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5670 2017-10-26 Tamar Christina <tamar.christina@arm.com>
5673 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
5674 Add flag_trapping_math and flag_fp_int_builtin_inexact.
5676 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
5678 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
5679 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
5680 mark as a sign-extending load.
5681 (local_pic_load_u): Define.
5683 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
5686 * fold-const.c (operand_equal_for_comparison_p): Also return true
5687 if ARG0 is a simple variant of ARG1 with narrower precision.
5688 (fold_ternary_loc): Always pass unstripped operands to the predicate.
5690 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
5692 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
5694 * i386.h (processor_costs): Add gather_static, gather_per_elt,
5695 scatter_static, scatter_per_elt.
5696 * x86-tune-costs.h: Add new cost entries.
5698 2017-10-25 Richard Biener <rguenther@suse.de>
5700 * tree-ssa-sccvn.h (vn_eliminate): Declare.
5701 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
5702 class pass_fre): Move to ...
5703 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
5704 class pass_fre): ... here and adjust for statistics.
5706 2017-10-25 Jakub Jelinek <jakub@redhat.com>
5709 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
5710 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
5711 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
5712 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
5715 2017-10-25 Richard Biener <rguenther@suse.de>
5717 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
5718 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
5719 eliminate_push_avail, eliminate_insert): Move inside...
5720 (class eliminate_dom_walker): ... this class in preparation
5722 (fini_eliminate): Remove by merging with ...
5723 (eliminate): ... this function. Adjust for class changes.
5724 (pass_pre::execute): Remove fini_eliminate call.
5725 (pass_fre::execute): Likewise.
5727 2017-10-24 Jakub Jelinek <jakub@redhat.com>
5730 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
5731 (VPERMI2, VPERMI2I): New mode iterators.
5732 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
5733 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
5735 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
5736 mode iterator. Remove 3 old define_insn patterns.
5737 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
5738 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
5739 VPERMI2 mode iterator, remove the other two expanders.
5740 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
5741 to use VPERMI2 mode iterator, add another alternative for vpermi2*
5742 instructions, remove the other two patterns.
5743 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
5744 mode iterator, remove the other two patterns.
5745 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
5746 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
5747 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
5748 and adjust argument order accordingly.
5749 (ix86_expand_vec_perm): Adjust caller.
5750 (expand_vec_perm_1): Likewise.
5751 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
5752 (expand_vec_perm_vpermt2_vpshub2): ... this.
5753 (ix86_expand_vec_perm_const_1): Adjust caller.
5754 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
5757 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
5758 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
5759 (vec_shl_<mode>): Remove unused expander.
5760 (avx512bw_<shift_insn><mode>3): New define_insn.
5761 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
5762 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
5764 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
5767 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
5770 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
5772 PR rtl-optimization/82396
5773 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
5774 (autopref_multipass_init): Simplify initialization.
5775 (autopref_rank_data): Simplify sort order.
5776 * gcc/sched-int.h (autopref_multipass_data_): Remove
5777 multi_mem_insn_p, min_offset and max_offset.
5779 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
5782 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
5783 Check special value of flag_omit_frame_pointer.
5784 (aarch64_can_eliminate): Likewise.
5785 (aarch64_override_options_after_change_1): Simplify handling of
5786 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
5788 2017-10-24 Richard Biener <rguenther@suse.de>
5790 PR tree-optimization/82697
5791 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
5792 zero for conditional load and unconditional store.
5794 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
5796 * doc/install.texi: Document bootstrap-cet.
5798 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
5801 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
5802 ENDBR instruction at function entrance if function is only
5805 2017-10-24 Jakub Jelinek <jakub@redhat.com>
5808 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
5809 patterns to better describe from which operation the CF is computed.
5810 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
5811 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
5812 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
5813 is 0, use _0 suffixed expanders instead of emitting a comparison
5816 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
5818 * config/i386/i386.md(*movsf_internal, *movdf_internal):
5819 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
5821 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
5824 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
5825 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
5826 * loop-iv.c (iv_get_reaching_def): Likewise.
5827 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
5828 variable is promoted and the partition contains undefined values.
5830 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5832 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
5834 (nios2_address_cost): Define.
5835 (nios2_legitimize_address): Recognize (exp + constant) directly.
5836 (TARGET_ADDRESS_COST): Define.
5838 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5840 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
5841 (nios2_symbolic_memory_operand_p): Declare.
5842 (nios2_split_large_constant): Declare.
5843 (nios2_split_symbolic_memory_operand): Declare.
5844 * config/nios2/nios2.c: Adjust includes.
5845 (nios2_symbolic_constant_allowed): New.
5846 (nios2_symbolic_constant_p): New.
5847 (nios2_plus_symbolic_constant_p): New.
5848 (nios2_valid_addr_expr_p): Recognize addresses involving
5850 (nios2_legitimate_address_p): Likewise, also LO_SUM.
5851 (nios2_symbolic_memory_operand_p): New.
5852 (nios2_large_constant_p): New.
5853 (nios2_split_large_constant): New.
5854 (nios2_split_plus_large_constant): New.
5855 (nios2_split_symbolic_memory_operand): New.
5856 (nios2_legitimize_address): Code refactoring. Handle addresses
5857 involving symbolic constants.
5858 (nios2_emit_move_sequence): Likewise.
5859 (nios2_print_operand): Improve error output.
5860 (nios2_print_operand_address): Handle LO_SUM.
5861 (nios2_cdx_narrow_form_p): Likewise.
5862 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
5863 operands involving symbolic constants.
5864 (movhi_internal, movsi_internal): Likewise.
5865 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
5866 (extendhisi2, extendqi<mode>2): Likewise.
5868 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5870 * tree-pass.h (PROP_rtl_split_insns): Define.
5871 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
5873 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5875 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
5877 2017-10-23 Jakub Jelinek <jakub@redhat.com>
5880 * target.def (const_not_ok_for_debug_p): Default to
5881 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
5882 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
5883 * targhooks.c (default_const_not_ok_for_debug_p): New function.
5884 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
5885 which targetm.const_not_ok_for_debug_p returned true.
5886 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
5888 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
5890 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
5891 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
5892 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
5894 (ix86_const_not_ok_for_debug_p): New function.
5895 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
5896 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
5898 2017-10-23 David Malcolm <dmalcolm@redhat.com>
5901 * system.h: Conditionally include "unique-ptr.h" if
5902 INCLUDE_UNIQUE_PTR is defined.
5903 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
5904 of defining INCLUDE_UNIQUE_PTR before including "system.h".
5906 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
5908 * config/rl78/rl78.md: New define_expand "subdi3".
5910 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
5913 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
5914 DF_REF_INSN if DF_REF_INSN_INFO is false.
5916 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
5918 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
5919 xmm_move instead of sse_move.
5920 (sse_store_index): New function.
5921 (ix86_register_move_cost): Be more sensible about mismatch stall;
5922 model AVX moves correctly; make difference between sse->integer and
5924 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
5925 moves; make difference between SSE and AVX.
5926 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
5927 and zmm_move. Increase size of sse load and store tables;
5928 add unaligned load and store tables; add ssemmx_to_integer.
5929 * x86-tune-costs.h: Update all entries according to real
5930 move latencies from Agner Fog's manual and chip documentation.
5932 2017-10-23 Jakub Jelinek <jakub@redhat.com>
5935 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
5936 * config/i386/constraints.md (Wf): New constraint.
5937 * config/i386/i386.md (UNSPEC_SBB): New unspec.
5938 (cmp<dwi>_doubleword): Removed.
5939 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
5940 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
5941 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
5942 expand with cmp<dwi>_doubleword. For LTU and GEU use
5943 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
5945 * common.opt (gcolumn-info): Enable by default.
5946 * doc/invoke.texi (gcolumn-info): Document new default.
5948 2017-10-23 Richard Biener <rguenther@suse.de>
5950 PR tree-optimization/82672
5951 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
5952 Fold the stmt if we propagated into it.
5954 2017-10-23 Richard Biener <rguenther@suse.de>
5956 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
5957 (bitmap_remove_expr_from_set): ... this. All callers call this
5958 for non-constant values.
5959 (bitmap_set_subtract): Rename to...
5960 (bitmap_set_subtract_expressions): ... this. Adjust and
5962 (bitmap_set_contains_value): Remove superfluous check.
5963 (bitmap_set_replace_value): Inline into single caller ...
5964 (bitmap_value_replace_in_set): ... here and simplify.
5965 (dependent_clean): Merge into ...
5966 (clean): ... this using an overload. Adjust.
5967 (prune_clobbered_mems): Adjust.
5968 (compute_antic_aux): Likewise.
5969 (compute_partial_antic_aux): Likewise.
5971 2017-10-23 Richard Biener <rguenther@suse.de>
5973 PR tree-optimization/82129
5975 2017-08-01 Richard Biener <rguenther@suse.de>
5977 PR tree-optimization/81181
5978 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5979 (compute_antic): ... end of iteration here.
5981 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
5983 * target.def (starting_frame_offset): New hook.
5984 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
5985 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
5986 * doc/tm.texi.in: Regenerate.
5987 * hooks.h (hook_hwi_void_0): Declare.
5988 * hooks.c (hook_hwi_void_0): New function.
5989 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
5990 STARTING_FRAME_OFFSET.
5991 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
5992 * reload1.c (reload): Likewise.
5993 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
5994 instead of STARTING_FRAME_OFFSET.
5995 * function.c (try_fit_stack_local): Likewise.
5996 (assign_stack_local_1): Likewise
5997 (instantiate_virtual_regs): Likewise.
5998 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
5999 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
6000 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
6001 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
6002 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
6003 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
6004 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
6005 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
6006 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
6007 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
6008 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
6009 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
6010 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
6011 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
6012 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
6013 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
6014 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
6015 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
6016 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
6017 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
6018 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
6019 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
6020 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
6021 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
6022 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
6023 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
6024 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
6025 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
6026 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
6027 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
6028 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
6029 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
6030 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
6031 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
6032 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
6033 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
6034 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
6035 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
6036 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
6037 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
6038 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
6039 * config/avr/avr.c (avr_starting_frame_offset): Make static and
6040 return a HOST_WIDE_INT.
6041 (avr_builtin_setjmp_frame_value): Use it instead of
6042 STARTING_FRAME_OFFSET.
6043 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6044 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
6045 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
6047 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6048 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
6049 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
6050 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6051 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
6052 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
6053 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6054 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
6055 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
6056 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6057 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
6058 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
6060 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6061 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
6062 * config/mips/mips.c (mips_compute_frame_info): Refer to
6063 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
6064 (mips_starting_frame_offset): New function.
6065 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6066 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
6067 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
6068 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
6069 and return a HOST_WIDE_INT.
6070 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6071 (mmix_initial_elimination_offset): Refer to
6072 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
6073 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
6074 * config/pa/pa.c (pa_starting_frame_offset): New function.
6075 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
6076 (pa_expand_prologue): Likewise.
6077 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6078 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
6079 !FRAME_GROWS_DOWNWARD handling to...
6080 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6081 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
6082 !FRAME_GROWS_DOWNWARD handling to...
6083 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6084 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
6085 !FRAME_GROWS_DOWNWARD handling to...
6086 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6087 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
6089 (rs6000_starting_frame_offset): New function.
6090 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
6091 !FRAME_GROWS_DOWNWARD handling to...
6092 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6093 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
6094 !FRAME_GROWS_DOWNWARD handling to...
6095 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6096 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
6097 !FRAME_GROWS_DOWNWARD handling to...
6098 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6099 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
6100 (rs6000_starting_frame_offset): New function.
6101 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
6102 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
6103 * config/vax/vax.c (vax_starting_frame_offset): New function.
6104 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
6105 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6106 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
6107 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
6108 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6109 * system.h (STARTING_FRAME_OFFSET): Poison.
6111 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6113 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
6114 SCALAR_TYPE_MODE instead of TYPE_MODE.
6116 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6117 Alan Hayward <alan.hayward@arm.com>
6118 David Sherwood <david.sherwood@arm.com>
6120 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
6122 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6123 Alan Hayward <alan.hayward@arm.com>
6124 David Sherwood <david.sherwood@arm.com>
6126 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
6128 2017-10-23 Richard Biener <rguenther@suse.de>
6130 PR tree-optimization/82129
6131 * tree-ssa-pre.c (bitmap_set_and): Remove.
6132 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
6133 canonicalizing expressions in the set to those with lowest
6134 ID rather than taking that from the first edge.
6136 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6138 * combine.c (rtx_equal_for_field_assignment_p): Use
6139 byte_lowpart_offset.
6141 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6142 Alan Hayward <alan.hayward@arm.com>
6143 David Sherwood <david.sherwood@arm.com>
6145 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
6146 to a SUBREG_PROMOTED_VAR.
6148 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6149 Alan Hayward <alan.hayward@arm.com>
6150 David Sherwood <david.sherwood@arm.com>
6152 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
6153 (expand_debug_source_expr): Likewise.
6154 * combine.c (combine_simplify_rtx): Likewise.
6155 * cse.c (fold_rtx): Likewise.
6156 * optabs.c (expand_float): Likewise.
6157 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6158 (simplify_binary_operation_1): Likewise.
6160 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6161 Alan Hayward <alan.hayward@arm.com>
6162 David Sherwood <david.sherwood@arm.com>
6164 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
6165 (record_promoted_value): Likewise.
6166 * expr.c (expand_expr_real_2): Likewise.
6167 * ree.c (update_reg_equal_equiv_notes): Likewise.
6168 (combine_set_extension): Likewise.
6169 * rtlanal.c (low_bitmask_len): Likewise.
6170 * simplify-rtx.c (neg_const_int): Likewise.
6171 (simplify_binary_operation_1): Likewise.
6173 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6174 Alan Hayward <alan.hayward@arm.com>
6175 David Sherwood <david.sherwood@arm.com>
6177 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
6178 * regcprop.c (maybe_mode_change): Likewise.
6179 * reload1.c (alter_reg): Likewise.
6181 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6183 * inchash.h (inchash::hash::add_wide_int): New function.
6184 * lto-streamer-out.c (hash_tree): Use it.
6186 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6188 * inchash.h (inchash::hash::add_wide_int): Rename to...
6189 (inchash::hash::add_hwi): ...this.
6190 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
6191 (polymorphic_call_target_hasher::hash): Likewise.
6192 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
6193 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
6194 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
6195 * lto-streamer-out.c (hash_tree): Likewise.
6196 * optc-save-gen.awk: Likewise.
6197 * tree.c (add_expr): Likewise.
6199 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
6202 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
6203 for ordered inequality comparisons even with TARGET_IEEE_FP.
6205 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
6208 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
6209 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
6210 Expand with cmp<dwi>_doubleword.
6212 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6214 * extend.texi: Add x86 specific to 'nocf_check' attribute.
6215 List CET intrinsics.
6216 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
6217 specific to -fcf-protection option.
6219 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6221 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
6222 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
6223 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
6224 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
6225 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
6226 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
6227 (extra_objs): Add cet.o for Linux/x86 targets.
6228 (tmake_file): Add i386/t-cet for Linux/x86 targets.
6229 * config/i386/cet.c: New file.
6230 * config/i386/cetintrin.h: Likewise.
6231 * config/i386/t-cet: Likewise.
6232 * config/i386/cpuid.h (bit_SHSTK): New.
6233 (bit_IBT): Likewise.
6234 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
6235 pass IBT and SHSTK bits.
6236 * config/i386/i386-builtin-types.def
6237 (VOID_FTYPE_UNSIGNED_PVOID): New.
6238 (VOID_FTYPE_UINT64_PVOID): Likewise.
6239 * config/i386/i386-builtin.def: Add CET intrinsics.
6240 * config/i386/i386-c.c (ix86_target_macros_internal): Add
6241 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
6242 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
6243 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
6245 * config/i386/i386.c (rest_of_insert_endbranch): New.
6246 (pass_data_insert_endbranch): Likewise.
6247 (pass_insert_endbranch): Likewise.
6248 (make_pass_insert_endbranch): Likewise.
6249 (ix86_notrack_prefixed_insn_p): Likewise.
6250 (ix86_target_string): Add -mibt, -mshstk flags.
6251 (ix86_option_override_internal): Add flag_cf_protection
6253 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
6254 (ix86_print_operand): Add 'notrack' prefix output.
6255 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
6256 (ix86_expand_builtin): Expand CET intrinsics.
6257 (x86_output_mi_thunk): Add 'endbranch' instruction.
6258 * config/i386/i386.h (TARGET_IBT): New.
6259 (TARGET_IBT_P): Likewise.
6260 (TARGET_SHSTK): Likewise.
6261 (TARGET_SHSTK_P): Likewise.
6262 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
6263 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
6264 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
6265 (builtin_setjmp_setup): New pattern.
6266 (builtin_longjmp): Likewise.
6267 (rdssp<mode>): Likewise.
6268 (incssp<mode>): Likewise.
6269 (saveprevssp): Likewise.
6270 (rstorssp): Likewise.
6271 (wrss<mode>): Likewise.
6272 (wruss<mode>): Likewise.
6273 (setssbsy): Likewise.
6274 (clrssbsy): Likewise.
6275 (nop_endbr): Likewise.
6276 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
6278 * config/i386/immintrin.h: Include <cetintrin.h>.
6279 * config/i386/linux-common.h
6280 (file_end_indicate_exec_stack_and_cet): New prototype.
6281 (TARGET_ASM_FILE_END): New.
6283 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
6285 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
6286 latencies instead of having separate table; make difference between
6287 integer and float costs.
6288 * i386.h (processor_costs): Remove scalar_stmt_cost,
6289 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
6290 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
6292 * x86-tune-costs.h: Remove entries which has been removed in
6293 procesor_costs from all tables; make cond_taken_branch_cost
6294 and cond_not_taken_branch_cost COST_N_INSNS based.
6296 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
6298 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
6300 2017-10-20 Jakub Jelinek <jakub@redhat.com>
6302 * config/i386/i386.md (isa): Remove fma_avx512f.
6303 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
6304 <avx512>_fmadd_<mode>_mask3<round_name>,
6305 <avx512>_fmsub_<mode>_mask<round_name>,
6306 <avx512>_fmsub_<mode>_mask3<round_name>,
6307 <avx512>_fnmadd_<mode>_mask<round_name>,
6308 <avx512>_fnmadd_<mode>_mask3<round_name>,
6309 <avx512>_fnmsub_<mode>_mask<round_name>,
6310 <avx512>_fnmsub_<mode>_mask3<round_name>,
6311 <avx512>_fmaddsub_<mode>_mask<round_name>,
6312 <avx512>_fmaddsub_<mode>_mask3<round_name>,
6313 <avx512>_fmsubadd_<mode>_mask<round_name>,
6314 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
6315 (*vec_widen_umult_even_v16si<mask_name>,
6316 *vec_widen_smult_even_v16si<mask_name>): Likewise.
6317 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
6319 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6321 * extend.texi: Add 'nocf_check' documentation.
6322 * gimple.texi: Add second parameter to
6323 gimple_build_call_from_tree.
6324 * invoke.texi: Add -fcf-protection documentation.
6325 * rtl.texi: Add REG_CALL_NOTRACK documenation.
6327 2017-10-20 Richard Biener <rguenther@suse.de>
6329 PR tree-optimization/82473
6330 * tree-vect-loop.c (vectorizable_reduction): Properly get at
6331 the largest input type.
6333 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6335 * c-attribs.c (handle_nocf_check_attribute): New function.
6336 (c_common_attribute_table): Add 'nocf_check' handling.
6337 * gimple-parser.c: Add second argument NULL to
6338 gimple_build_call_from_tree.
6339 * attrib.c (comp_type_attributes): Check nocf_check attribute.
6340 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
6342 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
6343 * common.opt: Add fcf-protection flag.
6344 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
6345 * flag-types.h: Add enum cf_protection_level.
6346 * gimple.c (gimple_build_call_from_tree): Add second parameter.
6347 Add 'nocf_check' attribute propagation to gimple call.
6348 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
6349 (gimple_build_call_from_tree): Update prototype.
6350 (gimple_call_nocf_check_p): New function.
6351 (gimple_call_set_nocf_check): Likewise.
6352 * gimplify.c: Add second argument to gimple_build_call_from_tree.
6353 * ipa-icf.c: Add nocf_check attribute in statement hash.
6354 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
6355 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
6356 * toplev.c (process_options): Add flag_cf_protection handling.
6358 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6360 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
6362 2017-10-20 Richard Biener <rguenther@suse.de>
6364 PR tree-optimization/82603
6365 * tree-if-conv.c (predicate_mem_writes): Make sure to only
6366 remove false predicated stores.
6368 2017-10-20 Richard Biener <rguenther@suse.de>
6370 * graphite-isl-ast-to-gimple.c
6371 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
6372 Remove return value and simplify, dump copied stmt after lhs
6374 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
6375 Reduce dump verbosity.
6376 (gsi_insert_earliest): Likewise.
6377 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
6378 * graphite.c (print_global_statistics): Adjust dumping.
6379 (print_graphite_scop_statistics): Likewise.
6380 (print_graphite_statistics): Do not dump loops here.
6381 (graphite_transform_loops): But here.
6383 2017-10-20 Nicolas Roche <roche@adacore.com>
6385 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
6386 * configure: Regenerate.
6388 2017-10-20 Jakub Jelinek <jakub@redhat.com>
6391 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
6392 functions when optimizing replace GIMPLE_RETURN stmts with
6393 calls to __builtin_unreachable ().
6396 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
6397 for -fsanitize=thread link of executables.
6398 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
6399 link of executables.
6402 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
6404 (<shift_insn><mode>3<mask_name>): Change the last of the 3
6405 define_insns for logical vector shifts to use VI248_AVX512BW
6406 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
6407 condition, useless isa and prefix attributes. Change the first
6408 2 of these define_insns to ...
6409 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
6410 define_insn for avx512vl.
6411 (<shift_insn><mode>3): ... and this, new define_insn without
6412 masking for non-avx512vl.
6415 * config/i386/sse.md (*andnot<mode>3,
6416 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
6417 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
6418 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
6419 not applied use empty suffix even for TARGET_AVX512VL.
6420 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
6421 is applied, supply evex,evex or evex,evex,evex instead of just
6424 2017-10-20 Julia Koval <julia.koval@intel.com>
6426 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
6427 (OPTION_MASK_ISA_GFNI_UNSET): New.
6428 (ix86_handle_option): Handle OPT_mgfni.
6429 * config/i386/cpuid.h (bit_GFNI): New.
6430 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
6431 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
6432 * config/i386/i386.c (ix86_target_string): Add -mgfni.
6433 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
6434 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
6435 * config/i386/i386.opt: Add mgfni.
6437 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
6439 * config/msp430/msp430.c (msp430_option_override): Disable
6440 -fdelete-null-pointer-checks.
6441 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
6443 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6445 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
6446 of x87 and SSE instructions.
6448 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6450 * asan.c (create_cond_insert_point): Do not update edge count.
6451 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
6452 (afdo_propagate_circuit): Likewise.
6453 (afdo_calculate_branch_prob): Likewise.
6454 (afdo_annotate_cfg): Likewise.
6455 * basic-block.h (struct edge_def): Remove count.
6456 (edge_def::count): New accessor.
6457 * bb-reorder.c (rotate_loop): Update.
6458 (find_traces_1_round): Update.
6459 (connect_traces): Update.
6460 (sanitize_hot_paths): Update.
6461 * cfg.c (unchecked_make_edge): Update.
6462 (make_single_succ_edge): Update.
6463 (check_bb_profile): Update.
6464 (dump_edge_info): Update.
6465 (update_bb_profile_for_threading): Update.
6466 (scale_bbs_frequencies_int): Update.
6467 (scale_bbs_frequencies_gcov_type): Update.
6468 (scale_bbs_frequencies_profile_count): Update.
6469 (scale_bbs_frequencies): Update.
6470 * cfganal.c (connect_infinite_loops_to_exit): Update.
6471 * cfgbuild.c (compute_outgoing_frequencies): Update.
6472 (find_many_sub_basic_blocks): Update.
6473 * cfgcleanup.c (try_forward_edges): Update.
6474 (try_crossjump_to_edge): Update
6475 * cfgexpand.c (expand_gimple_cond): Update
6476 (expand_gimple_tailcall): Update
6477 (construct_exit_block): Update
6478 * cfghooks.c (verify_flow_info): Update
6479 (redirect_edge_succ_nodup): Update
6480 (split_edge): Update
6481 (make_forwarder_block): Update
6482 (duplicate_block): Update
6483 (account_profile_record): Update
6484 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
6485 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
6486 * cfgloopmanip.c (scale_loop_profile): Update.
6488 (lv_adjust_loop_entry_edge): Update.
6489 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
6490 (force_nonfallthru_and_redirect): Update.
6491 (purge_dead_edges): Update.
6492 (rtl_flow_call_edges_add): Update.
6493 * cgraphunit.c (init_lowered_empty_function): Update.
6494 (cgraph_node::expand_thunk): Update.
6495 * gimple-pretty-print.c (dump_probability): Update.
6496 (dump_edge_probability): Update.
6497 * gimple-ssa-isolate-paths.c (isolate_path): Update.
6498 * haifa-sched.c (sched_create_recovery_edges): Update.
6499 * hsa-gen.c (convert_switch_statements): Update.
6500 * ifcvt.c (dead_or_predicable): Update.
6501 * ipa-inline-transform.c (inline_transform): Update.
6502 * ipa-split.c (split_function): Update.
6503 * ipa-utils.c (ipa_merge_profiles): Update.
6504 * loop-doloop.c (add_test): Update.
6505 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
6506 * lto-streamer-in.c (input_cfg): Update.
6507 (input_function): Update.
6508 * lto-streamer-out.c (output_cfg): Update.
6509 * modulo-sched.c (sms_schedule): Update.
6510 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
6511 * predict.c (maybe_hot_edge_p): Update.
6512 (unlikely_executed_edge_p): Update.
6513 (probably_never_executed_edge_p): Update.
6514 (dump_prediction): Update.
6515 (drop_profile): Update.
6516 (propagate_unlikely_bbs_forward): Update.
6517 (determine_unlikely_bbs): Update.
6518 (force_edge_cold): Update.
6519 * profile.c (compute_branch_probabilities): Update.
6520 * reg-stack.c (better_edge): Update.
6521 * shrink-wrap.c (handle_simple_exit): Update.
6522 * tracer.c (better_p): Update.
6523 * trans-mem.c (expand_transaction): Update.
6524 (split_bb_make_tm_edge): Update.
6525 * tree-call-cdce.c: Update.
6526 * tree-cfg.c (gimple_find_sub_bbs): Update.
6527 (gimple_split_edge): Update.
6528 (gimple_duplicate_sese_region): Update.
6529 (gimple_duplicate_sese_tail): Update.
6530 (gimple_flow_call_edges_add): Update.
6531 (insert_cond_bb): Update.
6532 (execute_fixup_cfg): Update.
6533 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
6534 * tree-complex.c (expand_complex_div_wide): Update.
6535 * tree-eh.c (lower_resx): Update.
6536 (unsplit_eh): Update.
6537 (cleanup_empty_eh_move_lp): Update.
6538 * tree-inline.c (copy_edges_for_bb): Update.
6539 (freqs_to_counts): Update.
6540 (copy_cfg_body): Update.
6541 * tree-ssa-dce.c (remove_dead_stmt): Update.
6542 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
6543 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
6544 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
6545 (unloop_loops): Update.
6546 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
6547 * tree-ssa-loop-split.c (connect_loops): Update.
6548 (split_loop): Update.
6549 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
6550 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
6551 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
6552 * tree-ssa-reassoc.c (branch_fixup): Update.
6553 * tree-ssa-tail-merge.c (replace_block_by): Update.
6554 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
6555 (compute_path_counts): Update.
6556 (update_profile): Update.
6557 (recompute_probabilities): Update.
6558 (update_joiner_offpath_counts): Update.
6559 (estimated_freqs_path): Update.
6560 (freqs_to_counts_path): Update.
6561 (clear_counts_path): Update.
6562 (ssa_fix_duplicate_block_edges): Update.
6563 (duplicate_thread_path): Update.
6564 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
6565 (case_bit_test_cmp): Update.
6566 (collect_switch_conv_info): Update.
6567 (gen_inbound_check): Update.
6568 (do_jump_if_equal): Update.
6569 (emit_cmp_and_jump_insns): Update.
6570 * tree-tailcall.c (decrease_profile): Update.
6571 (eliminate_tail_call): Update.
6572 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
6573 (vect_do_peeling): Update.
6574 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
6575 * ubsan.c (ubsan_expand_null_ifn): Update.
6576 (ubsan_expand_ptr_ifn): Update.
6577 * value-prof.c (gimple_divmod_fixed_value): Update.
6578 (gimple_mod_pow2): Update.
6579 (gimple_mod_subtract): Update.
6580 (gimple_ic): Update.
6581 (gimple_stringop_fixed_value): Update.
6583 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
6586 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
6588 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
6590 PR rtl-optimization/82395
6591 * ira-color.c (allocno_priority_compare_func): Fix comparison step
6592 based on non_spilled_static_chain_regno_p.
6594 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
6596 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
6597 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
6598 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
6600 2017-10-19 Martin Sebor <msebor@redhat.com>
6602 PR tree-optimization/82596
6603 * tree.c (array_at_struct_end_p): Handle STRING_CST.
6605 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
6607 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
6608 (get_mem_refs_of_builtin_call): Likewise.
6609 * builtins.c (expand_builtin_apply): Adjust call to
6610 allocate_dynamic_stack_space.
6611 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
6612 the third argument to allocate_dynamic_stack_space, otherwise -1.
6613 (expand_builtin): Deal with all alloca variants.
6614 (is_inexpensive_builtin): Likewise.
6615 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
6616 * calls.c (special_function_p): Deal with all alloca variants.
6617 (initialize_argument_information): Adjust call to
6618 allocate_dynamic_stack_space.
6619 (expand_call): Likewise.
6620 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
6621 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
6622 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
6623 use it for the stack usage computation.
6624 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
6625 * function.c (gimplify_parameters): Call build_alloca_call_expr.
6626 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
6627 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
6628 (in_loop_p): Remove first argument and useless check.
6629 (pass_walloca::execute): Remove useless test and adjust call to above.
6630 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
6631 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
6632 (gimplify_call_expr): Deal with all alloca variants.
6633 * hsa-gen.c (gen_hsa_alloca): Likewise.
6634 (gen_hsa_insns_for_call): Likewise.
6635 * ipa-pure-const.c (special_builtin_state): Likewise.
6636 * tree-chkp.c (chkp_build_returned_bound): Likewise.
6637 * tree-object-size.c (alloc_object_size): Likewise.
6638 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
6639 (call_may_clobber_ref_p_1): Likewise.
6640 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
6641 (ccp_fold_stmt): Likewise.
6642 (optimize_stack_restore): Likewise.
6643 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
6644 (mark_all_reaching_defs_necessary_1): Likewise.
6645 (propagate_necessity): Likewise.
6646 (eliminate_unnecessary_stmts): Likewise.
6647 * tree.c (build_common_builtin_nodes): Build
6648 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
6649 (build_alloca_call_expr): New function.
6650 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
6651 (CASE_BUILT_IN_ALLOCA): Likewise.
6652 (build_alloca_call_expr): Declare.
6653 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
6655 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
6658 * dwarf2out.c (new_die_raw): New static inline function.
6659 (new_die): Use it to create the DIE.
6660 (add_AT_external_die_ref): Likewise.
6661 (clone_die): Likewise.
6662 (clone_as_declaration): Likewise.
6663 (dwarf2out_vms_debug_main_pointer): Likewise.
6664 (base_type_die): Likewise. Remove early return for corner cases.
6665 Do not call add_pubtype on the DIE here.
6666 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
6667 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
6668 typedefs for base types with DW_AT_endianity. Make sure a DIE with
6669 native order exists for base types, attach the DIE manually and call
6670 add_pubtype on it. Do not equate a reverse order DIE to the type.
6672 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
6674 * config/arm/arm.c (align_ok_ldrd_strd): New function.
6675 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
6677 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
6679 2017-10-19 Jakub Jelinek <jakub@redhat.com>
6681 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
6682 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
6683 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
6684 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
6685 * opts.c (sanitizer_opts): Add builtin.
6686 * ubsan.c (instrument_builtin): New function.
6687 (pass_ubsan::execute): Call it.
6688 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
6689 * doc/invoke.texi: Document -fsanitize=builtin.
6691 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
6692 builtins, store max (log2 (align), 0) into uchar field instead of
6693 align into uptr field.
6694 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
6695 store uchar 0 field instead of uptr 0 field.
6696 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
6697 instead of passing one address of struct with 2 locations pass
6698 two addresses of structs with 1 location each.
6699 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
6700 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
6701 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
6702 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
6703 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
6704 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
6705 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
6706 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
6708 2017-10-19 Martin Liska <mliska@suse.cz>
6711 * file-find.c (remove_prefix): Remove.
6712 * file-find.h (remove_prefix): Likewise.
6713 * gcc-ar.c: Remove smartness of lookup.
6715 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
6717 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
6718 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
6719 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
6721 2017-10-19 Jakub Jelinek <jakub@redhat.com>
6724 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
6725 (setcc + and to xor + setcc): New peephole2.
6727 2017-10-19 Tom de Vries <tom@codesourcery.com>
6729 * doc/sourcebuild.texi (Test Directives, Variants of
6730 dg-require-support): Add dg-require-stack-size.
6732 2017-10-19 Martin Liska <mliska@suse.cz>
6735 * gimplify.c (gimplify_decl_expr): Do not instrument variables
6736 that have a large alignment.
6737 (gimplify_target_expr): Likewise.
6739 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
6741 PR rtl-optimization/82602
6742 * ira.c (rtx_moveable_p): Return false for volatile asm.
6744 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
6747 * config/i386/i386-modes.def (CCGZ): New CC mode.
6748 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
6749 * config/i386/predicates.md (ix86_comparison_operator):
6751 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
6752 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
6753 with double-word subtraction.
6754 (put_condition_code): Handle CCGZmode.
6756 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
6758 * wide-int.cc (debug (const wide_int &)): New.
6759 (debug (const wide_int *)): New.
6760 (debug (const widest_int &)): New.
6761 (debug (const widest_int *)): New.
6763 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
6766 * lra-constraints.c (curr_insn_transform): Use non-input operand
6767 instead of output one for matched reload.
6769 2017-10-18 Bin Cheng <bin.cheng@arm.com>
6771 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
6772 (tree-ssa-loop-ivopts.h): New header file.
6773 (struct builtin_info): New fields.
6774 (classify_builtin_1): Compute and record base and offset parts for
6775 memset builtin partition by calling strip_offset.
6776 (offset_cmp, fuse_memset_builtins): New functions.
6777 (finalize_partitions): Fuse adjacent memset partitions by calling
6779 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
6780 Expose the interface.
6781 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
6783 2017-10-18 Bin Cheng <bin.cheng@arm.com>
6785 PR tree-optimization/82574
6786 * tree-loop-distribution.c (find_single_drs): New parameter. Check
6787 that data reference must be executed exactly once per iteration
6788 against the outermost loop in nest.
6789 (classify_partition): Update call to above function.
6791 2017-10-18 Richard Biener <rguenther@suse.de>
6793 PR tree-optimization/82591
6794 * graphite.c (graphite_transform_loops): Move code gen message
6796 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6797 Here. Handle scop_to_isl_ast failing.
6798 (scop_to_isl_ast): Limit the number of ISL operations.
6800 2017-10-18 Richard Biener <rguenther@suse.de>
6802 * graphite-isl-ast-to-gimple.c
6803 (translate_isl_ast_to_gimple::set_rename): Simplify.
6804 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
6805 (graphite_copy_stmts_from_block): ... here.
6806 (copy_bb_and_scalar_dependences): Simplify.
6807 (add_parameters_to_ivs_params): Canonicalize.
6808 (generate_entry_out_of_ssa_copies): Simplify.
6809 * graphite-sese-to-poly.c (extract_affine_name): Simplify
6810 by passing in ISL dimension.
6811 (parameter_index_in_region_1): Rename to ...
6812 (parameter_index_in_region): ... this.
6813 (extract_affine): Adjust assert, pass down parameter index.
6814 (add_param_constraints): Use range-info when available.
6815 (build_scop_context): Adjust.
6816 * sese.c (new_sese_info): Adjust.
6817 (free_sese_info): Likewise.
6818 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
6819 Remove unused typedefs.
6820 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
6822 2017-10-18 Martin Liska <mliska@suse.cz>
6824 * combine.c (simplify_compare_const): Add gcc_fallthrough.
6826 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
6828 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
6829 (s390_sched_init): Do not reset s390_sched_state if we entered the
6830 current basic block via a fallthru edge and all others are unlikely.
6832 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
6834 * config/s390/s390.c (NUM_SIDES): New variable.
6835 (LONGRUNNING_THRESHOLD): New variable.
6836 (LATENCY_FACTOR): New variable.
6837 (s390_sched_score): Decrease score for long-running instructions on
6839 (s390_sched_variable_issue): Perform bookkeeping for long-running
6842 2017-10-18 Richard Biener <rguenther@suse.de>
6844 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
6845 Simplify with removal of the parameter rename map.
6846 (set_rename): Likewise.
6847 (should_copy_to_new_region): Likewise.
6848 (graphite_copy_stmts_from_block): Likewise.
6849 (copy_bb_and_scalar_dependences): Remove initialization of
6850 unused copied_bb_map.
6852 (copy_internal_parameters): Likewise.
6853 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
6854 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
6855 Use INTEGRAL_TYPE_P.
6856 (parameter_index_in_region_1): Rename to ...
6857 (assign_parameter_index_in_region): ... this. Assert we have
6858 a parameter we handle.
6859 (scan_tree_for_params): Adjust.
6860 * sese.h (parameter_rename_map_t): Remove.
6861 (struct sese_info_t): Remove unused parameter_rename_map and
6862 copied_bb_map members.
6863 * sese.c (new_sese_info): Adjust.
6864 (free_sese_info): Likewise.
6866 2017-10-18 Martin Liska <mliska@suse.cz>
6869 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
6870 on an abnormal edge.
6872 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
6874 * doc/invoke.texi (ffunction-sections and fdata-sections):
6877 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
6879 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
6880 the use statement can throw internally.
6882 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
6884 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
6885 any RTX present on the RHS of a SET.
6886 * compare-elim.c (try_eliminate_compare): Restore comment.
6888 2017-10-17 Jakub Jelinek <jakub@redhat.com>
6890 * langhooks.h (struct lang_hooks): Document that tree_size langhook
6891 may be also called on tcc_type nodes.
6892 * langhooks.c (lhd_tree_size): Likewise.
6894 2017-10-17 David Malcolm <dmalcolm@redhat.com>
6896 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
6897 format_warning_at_substring.
6898 (maybe_warn): Convert source_range * param to a location_t. Pass
6899 UNKNOWN_LOCATION rather than NULL to fmtwarn.
6900 (format_directive): Remove code to extract source_ranges and
6901 source_range * in favor of just a location_t.
6902 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
6904 * substring-locations.c (format_warning_va): Convert
6905 source_range * param to a location_t.
6906 (format_warning_at_substring): Likewise.
6907 * substring-locations.h (format_warning_va): Likewise.
6908 (format_warning_at_substring): Likewise.
6910 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
6912 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
6914 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
6915 and scatter/gather ops.
6917 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
6918 vec_gather_load and vec_scatter_store.
6919 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
6920 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
6921 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
6922 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
6923 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
6924 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
6926 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
6928 * reg-stack.c (compare_for_stack_reg): Add bool argument.
6929 Detect FTST instruction and handle its register pops. Only pop
6930 second operand if can_pop_second_op is true.
6931 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
6932 set can_pop_second_op to false in the compare_for_stack_reg call.
6934 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
6935 output_fp_compare for stack register operands.
6936 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
6937 instructions here. Do not emit stack register pops here. Assert
6938 that FCOMPP pops next to top stack register. Rewrite function.
6940 2017-10-17 Nathan Sidwell <nathan@acm.org>
6943 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
6944 use DECL_ASSEMBLER_NAME_RAW.
6947 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
6950 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
6951 Wilco Dijkstra <wilco.dijkstra@arm.com>
6953 * builtins.c (expand_builtin_update_setjmp_buf): Add a
6954 converstion to Pmode from the buf_addr.
6956 2017-10-17 Richard Biener <rguenther@suse.de>
6958 * graphite-dependences.c (scop_get_reads_and_writes): Change
6959 output parameters to references.
6961 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
6963 PR 71026/tree-optimization
6964 * fold-const.c (distribute_real_division): Removed.
6965 (fold_binary_loc): Remove calls to distribute_real_divison.
6967 2017-10-17 Richard Biener <rguenther@suse.de>
6969 * graphite-scop-detection.c
6970 (scop_detection::stmt_has_simple_data_refs_p): Always use
6971 the full nest as region.
6972 (try_generate_gimple_bb): Likewise.
6973 * sese.c (scalar_evolution_in_region): Simplify now that
6974 SCEV can handle instantiation in regions.
6975 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
6976 in the non-loop part of a function if requested.
6978 2017-10-17 Richard Biener <rguenther@suse.de>
6980 PR tree-optimization/82563
6981 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
6983 (graphite_regenerate_ast_isl): Call it.
6984 * graphite-scop-detection.c (build_scops): Remove entry edge split.
6986 2017-10-17 Jakub Jelinek <jakub@redhat.com>
6988 PR tree-optimization/82549
6989 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
6990 Formatting fixes. Instead of calling make_bit_field_ref with negative
6993 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
6995 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
6996 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
6997 _mm_maskz_reduce_ss): New.
6998 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
6999 __builtin_ia32_reducess_mask): Ditto..
7000 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
7001 * config/i386/sse.md (reduces<mode>): Renamed to ...
7002 (reduces<mode><mask_scalar_name>): ... this.
7003 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
7005 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
7006 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
7008 2017-10-16 David Malcolm <dmalcolm@redhat.com>
7010 * Makefile.in (OBJS): Add unique-ptr-tests.o.
7011 * selftest-run-tests.c (selftest::run_tests): Call
7012 selftest::unique_ptr_tests_cc_tests.
7013 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
7014 * unique-ptr-tests.cc: New file.
7016 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
7019 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
7021 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
7022 (make_hard_regno_born, make_hard_regno_dead): Update
7023 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
7025 2017-10-16 Jeff Law <law@redhat.com>
7027 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
7029 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
7031 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
7033 2017-10-16 Olivier Hainque <hainque@adacore.com>
7035 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
7036 with_cpu if we were configured for an e500v2 target cpu name.
7038 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
7040 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
7041 * doc/invoke.texi: Document +nodsp as a valid extension for
7044 2017-10-16 Martin Liska <mliska@suse.cz>
7046 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
7047 (test_set_range): Likewise.
7048 (test_range_functions): Rename to ...
7049 (test_bit_in_range): ... this.
7050 (sbitmap_c_tests): Add new test.
7052 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7054 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
7056 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
7057 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
7059 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7061 * config/aarch64/aarch64-builtins.c
7062 (aarch64_types_quadopu_lane_qualifiers): New.
7063 (TYPES_QUADOPU_LANE): New.
7064 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
7065 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
7066 (aarch64_<sur>dot_laneq<vsi2qi>): New.
7067 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
7068 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
7069 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
7070 (Vdottype, DOTPROD): New.
7071 (sur): Add SDOT and UDOT.
7073 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7075 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
7076 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
7077 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
7079 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
7080 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
7081 Enable TARGET_DOTPROD.
7082 (cortex-a75.cortex-a55): Likewise.
7083 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
7085 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7087 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
7088 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
7090 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
7092 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
7093 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
7094 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
7095 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
7096 * config/arm/types.md (neon_dot, neon_dot_q): New.
7097 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
7099 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7101 * config/arm/arm.h (TARGET_DOTPROD): New.
7102 * config/arm/arm.c (arm_arch_dotprod): New.
7103 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
7104 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
7105 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
7106 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
7107 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
7108 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
7109 * doc/invoke.texi (armv8.2-a): Document dotprod
7111 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
7113 * i386.c (ix86_vec_cost): New function.
7114 (ix86_rtx_costs): Handle vector operations better.
7115 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
7116 * x86-tune-costs.h: Add new costs to all tables.
7118 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
7120 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
7122 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
7123 divsd, sqrtss and sqrtsd
7124 * x86-tune-costs.h: Add new entries to all costs.
7125 (znver1_cost): Fix to match real instruction latencies.
7127 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7128 Michael Collison <michael.collison@arm.com>
7130 * compare-elim.c: Include emit-rtl.h.
7131 (can_merge_compare_into_arith): New function.
7132 (try_validate_parallel): Likewise.
7133 (try_merge_compare): Likewise.
7134 (try_eliminate_compare): Call the above when no previous clobber
7136 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
7139 2017-10-14 Jakub Jelinek <jakub@redhat.com>
7143 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
7144 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
7148 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
7149 to be any operand_equal_p operands. For & (B - 1) require
7150 B to be power of 2. Recognize
7151 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
7153 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7156 * optabs.c (expand_memory_blockage): Fix call of
7157 targetm.have_memory_blockage.
7159 2017-10-14 Jakub Jelinek <jakub@redhat.com>
7162 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
7163 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
7164 objects to extra_objs instead of overwriting it.
7166 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7168 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
7169 Use any_fp_register_operand as operand[3] predicate. Simplify
7170 equality test for operands[2] and operands[4] memory location.
7171 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
7172 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
7173 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
7174 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
7175 any_fp_register_operand as operand[1] predicate. Simplify
7176 equality test for operands[0] and operands[3] memory location.
7177 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
7178 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
7179 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
7181 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7183 * target-insns.def: Add memory_blockage.
7184 * optabs.c (expand_memory_blockage): New function.
7185 (expand_asm_memory_barrier): Rename ...
7186 (expand_asm_memory_blockage): ... to this.
7187 (expand_mem_thread_fence): Call expand_memory_blockage
7188 instead of expand_asm_memory_barrier.
7189 (expand_mem_singnal_fence): Ditto.
7190 (expand_atomic_load): Ditto.
7191 (expand_atomic_store): Ditto.
7192 * doc/md.texi (Standard Pattern Names For Generation):
7193 Document memory_blockage instruction pattern.
7195 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
7197 * config/rl78/rl78.c (rl78_emit_libcall): New function.
7198 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
7199 * config/rl78/rl78.md: New define_expand "adddi3".
7201 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
7203 * cfghooks.c (verify_flow_info): Disable check that all probabilities
7206 2017-10-13 Jeff Law <law@redhat.com>
7208 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
7210 2017-10-13 Jakub Jelinek <jakub@redhat.com>
7213 * internal-fn.c (expand_mul_overflow): If both operands have
7214 the same highpart of -1 or 0 and the topmost bit of lowpart
7215 is different, overflow is if res <= 0 rather than res < 0.
7217 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
7219 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
7220 TARGET_P9_VECTOR code for unaligned_load case.
7222 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
7224 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
7226 2017-10-13 Nathan Sidwell <nathan@acm.org>
7228 * tree-core.h (tree_contains_struct): Make bool.
7229 * tree.c (tree_contains_struct): Likewise.
7230 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
7231 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
7232 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
7233 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
7234 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
7236 2017-10-13 Richard Biener <rguenther@suse.de>
7238 * graphite-isl-ast-to-gimple.c
7239 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
7240 parameters and dominance check.
7241 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
7242 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
7243 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
7244 Do not update SSA form here or do intermediate IL verification.
7245 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
7246 (graphite_initialize): Remove check on the number of loops in
7247 the function and inline into graphite_transform_loops.
7248 (graphite_finalize): Inline into graphite_transform_loops.
7249 (graphite_transform_loops): Perform SSA update and IL verification
7251 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
7253 2017-10-13 Richard Biener <rguenther@suse.de>
7255 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
7256 graphite_expression_type_precision): Avoid global constructor
7258 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
7259 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
7260 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
7261 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
7262 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
7263 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
7265 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
7268 * config/i386/i386.h (ix86_red_zone_size): New.
7269 * config/i386/i386.md (push peephole2s): Replace
7270 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
7272 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7273 Alan Hayward <alan.hayward@arm.com>
7274 David Sherwood <david.sherwood@arm.com>
7276 * combine.c (can_change_dest_mode): Reject changes in
7277 REGMODE_NATURAL_SIZE.
7279 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7280 Alan Hayward <alan.hayward@arm.com>
7281 David Sherwood <david.sherwood@arm.com>
7283 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
7284 (expand_debug_source_expr): Likewise.
7285 * combine.c (combine_simplify_rtx): Likewise.
7286 * cse.c (fold_rtx): Likewise.
7287 * fwprop.c (canonicalize_address): Likewise.
7288 * targhooks.c (default_shift_truncation_mask): Likewise.
7290 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7291 Alan Hayward <alan.hayward@arm.com>
7292 David Sherwood <david.sherwood@arm.com>
7294 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
7295 (widened_mode): Likewise.
7296 (expand_unop): Likewise.
7297 * ree.c (transform_ifelse): Likewise.
7298 (merge_def_and_ext): Likewise.
7299 (combine_reaching_defs): Likewise.
7300 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7302 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7303 Alan Hayward <alan.hayward@arm.com>
7304 David Sherwood <david.sherwood@arm.com>
7306 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
7307 * combine.c (gen_lowpart_for_combine): Likewise.
7308 * dwarf2out.c (rtl_for_decl_location): Likewise.
7309 * final.c (alter_subreg): Likewise.
7310 * rtlhooks.c (gen_lowpart_general): Likewise.
7311 (gen_lowpart_if_possible): Likewise.
7313 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7314 Alan Hayward <alan.hayward@arm.com>
7315 David Sherwood <david.sherwood@arm.com>
7317 * calls.c (expand_call): Use subreg_lowpart_offset.
7318 * cse.c (cse_insn): Likewise.
7319 * regcprop.c (copy_value): Likewise.
7320 (copyprop_hardreg_forward_1): Likewise.
7322 2017-10-13 Jakub Jelinek <jakub@redhat.com>
7325 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
7326 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
7327 =Q constraints to +Q and into insn condition add check
7328 that operands[0] and operands[1] are equal.
7329 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
7330 =Q constraints to +Q and into insn condition add check
7331 that operands[0] is equal to either operands[1] or operands[2].
7334 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
7335 instead of handling MINUS_EXPR twice (once for each argument),
7336 canonicalize operand order and handle just once, use rtype where
7337 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
7340 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
7341 any values of __C while still being pattern recognizable as a simple
7344 2017-10-13 Richard Biener <rguenther@suse.de>
7346 PR tree-optimization/82451
7348 2017-10-02 Richard Biener <rguenther@suse.de>
7350 PR tree-optimization/82355
7351 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
7352 a mapping for the enclosing loop but avoid generating one for
7354 (copy_bb_and_scalar_dependences): Remove premature codegen
7355 error on PHIs in blocks duplicated into multiple places.
7356 * graphite-scop-detection.c
7357 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
7358 in the region use it as loop and nest to analyze the DR in.
7359 (try_generate_gimple_bb): Likewise.
7360 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
7361 (add_loop_constraints): For blocks in a loop not in the region
7362 create a dimension with a single iteration.
7363 * sese.h (gbb_loop_at_index): Remove assert.
7365 * cfgloop.c (loop_preheader_edge): For the loop tree root
7366 return the single successor of the entry block.
7367 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7368 Reset the SCEV hashtable and niters.
7369 * graphite-scop-detection.c
7370 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
7371 assert that we only have POLYNOMIAL_CHREC that vary in loops
7372 contained in the region.
7373 (scop_detection::graphite_can_represent_expr): Adjust.
7374 (scop_detection::stmt_has_simple_data_refs_p): For loops
7375 not in the region set loop to NULL. The nest is now the
7376 entry edge to the region.
7377 (try_generate_gimple_bb): Likewise.
7378 * sese.c (scalar_evolution_in_region): Adjust for
7379 instantiate_scev change.
7380 * tree-data-ref.h (graphite_find_data_references_in_stmt):
7381 Make nest parameter the edge into the region.
7382 (create_data_ref): Likewise.
7383 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
7384 entry edge into a region and adjust instantiate_scev calls.
7385 (create_data_ref): Likewise.
7386 (graphite_find_data_references_in_stmt): Likewise.
7387 (find_data_references_in_stmt): Pass the loop preheader edge
7388 from the nest argument.
7389 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
7390 parameter the edge into the region.
7391 (instantiate_parameters): Use the loop preheader edge as entry.
7392 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
7394 (get_instantiated_value_entry): Make instantiate_below parameter
7395 the edge into the region.
7396 (instantiate_scev_name): Likewise. Adjust dominance checks,
7397 when we cannot use loop-based instantiation instantiate by
7398 walking use-def chains.
7399 (instantiate_scev_poly): Adjust.
7400 (instantiate_scev_binary): Likewise.
7401 (instantiate_scev_convert): Likewise.
7402 (instantiate_scev_not): Likewise.
7403 (instantiate_array_ref): Remove.
7404 (instantiate_scev_3): Likewise.
7405 (instantiate_scev_2): Likewise.
7406 (instantiate_scev_1): Likewise.
7407 (instantiate_scev_r): Do not blindly handle N-operand trees.
7408 Do not instantiate array-refs. Handle all constants and invariants.
7409 (instantiate_scev): Make instantiate_below parameter
7410 the edge into the region.
7411 (resolve_mixers): Use the loop preheader edge for the region
7412 parameter to instantiate_scev_r.
7413 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
7415 2017-10-13 Richard Biener <rguenther@suse.de>
7417 PR tree-optimization/82525
7418 * graphite-isl-ast-to-gimple.c
7419 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
7421 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
7422 Fail code generation when we cannot represent the isl integer.
7423 (binary_op_to_tree): Elide modulo operations that are no-ops
7424 in the type we code generate. Remove now superfluous code
7427 2017-10-13 Richard Biener <rguenther@suse.de>
7429 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
7430 (scop_detection::harmful_loop_in_region): Remove premature
7431 IV type restriction.
7432 (scop_detection::graphite_can_represent_scev): We can handle
7433 pointer IVs just fine.
7435 2017-10-13 Alan Modra <amodra@gmail.com>
7437 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
7438 "Clobbers and Scratch Registers". Add paragraph on
7439 alternative to clobbers for scratch registers and OpenBLAS
7442 2017-10-13 Alan Modra <amodra@gmail.com>
7444 * doc/extend.texi (Clobbers): Correct vax example. Delete old
7445 example of a memory input for a string of known length. Move
7446 commentary out of table. Add a number of new examples
7447 covering array memory inputs.
7449 2017-10-12 Martin Liska <mliska@suse.cz>
7451 PR tree-optimization/82493
7452 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
7453 (test_range_functions): New function.
7454 (sbitmap_c_tests): Likewise.
7455 * selftest-run-tests.c (selftest::run_tests): Run new tests.
7456 * selftest.h (sbitmap_c_tests): New function.
7458 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
7460 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
7462 * config/rs6000/amo.h: Fix spacing issue.
7464 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7467 * config/i386/i386.md (*ashl<mode>3_mask_1,
7468 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
7469 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
7472 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
7474 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
7475 (profile_probability): Set max_probability
7476 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
7478 * profile-count.c (profile_probability::differs_from_p): Do not
7479 rely on max_probaiblity == 10000
7481 2017-10-12 Jeff Law <law@redhat.com>
7483 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
7486 2017-10-12 Martin Sebor <msebor@redhat.com>
7490 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
7491 (handle_alias_pairs): Call it.
7492 * common.opt (-Wattribute-alias): New option.
7493 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
7494 * doc/invoke.texi (-Wattribute-alias): Document.
7496 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
7499 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
7501 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
7503 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
7504 (make_hard_regno_born, make_hard_regno_dead): Update
7505 bb_killed_pseudos and bb_gen_pseudos.
7507 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
7509 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
7511 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
7513 * config/alpha/alpha.c (alpha_split_conditional_move):
7514 Use std::swap instead of manually swapping.
7515 (alpha_stdarg_optimize_hook): Ditto.
7516 (alpha_canonicalize_comparison): Ditto.
7518 2017-10-12 Bin Cheng <bin.cheng@arm.com>
7520 * tree-loop-distribution.c (struct builtin_info): New struct.
7521 (struct partition): Refactor fields into struct builtin_info.
7522 (partition_free): Free struct builtin_info.
7523 (build_size_arg_loc, build_addr_arg_loc): Delete.
7524 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
7525 information from struct builtin_info.
7526 (find_single_drs): New function refactored from classify_partition.
7527 Also moved builtin validity checks to this function.
7528 (compute_access_range, alloc_builtin): New functions.
7529 (classify_builtin_st, classify_builtin_ldst): New functions.
7530 (classify_partition): Refactor code into functions find_single_drs,
7531 classify_builtin_st and classify_builtin_ldst.
7532 (distribute_loop): Don't do runtime alias check when distributing
7534 (find_seed_stmts_for_distribution): New function.
7535 (pass_loop_distribution::execute): Refactor code finding seed
7536 stmts into above function. Support distribution for the innermost
7537 two-level loop nest. Adjust dump information.
7539 2017-10-12 Bin Cheng <bin.cheng@arm.com>
7541 * tree-loop-distribution.c: Adjust the general comment.
7542 (NUM_PARTITION_THRESHOLD): New macro.
7543 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
7544 (classify_partition): Skip builtin pattern of loop nest's inner loop.
7545 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
7546 in call to build_partition_graph.
7547 (finalize_partitions): New parameter. Make loop distribution more
7548 conservative by fusing more partitions.
7549 (distribute_loop): Don't do runtime alias check in case of loop nest
7551 (find_seed_stmts_for_distribution): New function.
7552 (prepare_perfect_loop_nest): New function.
7553 (pass_loop_distribution::execute): Refactor code finding seed stmts
7554 and loop nest into above functions. Support loop nest distribution.
7555 Adjust dump information accordingly.
7557 2017-10-12 Bin Cheng <bin.cheng@arm.com>
7559 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
7560 and set PTYPE_SEQUENTIAL for merged partition.
7562 2017-10-12 Richard Biener <rguenther@suse.de>
7564 PR tree-optimization/69728
7566 2017-09-19 Richard Biener <rguenther@suse.de>
7568 PR tree-optimization/69728
7569 * graphite-sese-to-poly.c (schedule_error): New global.
7570 (add_loop_schedule): Handle empty domain by failing the
7572 (build_original_schedule): Handle schedule_error.
7574 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
7575 domain by returning an unchanged schedule.
7577 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7579 * genrecog.c (validate_pattern): For VEC_SELECT verify that
7580 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
7582 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
7584 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
7587 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7590 * expr.c (store_field): Don't optimize away bitsize == 0 store
7591 from CALL_EXPR with addressable return type.
7593 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
7595 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
7596 * config/rs6000/rs6000.md (sel): Delete mode attribute.
7597 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
7598 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
7599 TARGET_ISEL instead of TARGET_ISEL<sel>.
7601 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
7603 * config/rs6000/rs6000.c
7604 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
7606 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
7608 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
7610 (reg_or_cint_operand): Fix comment.
7611 (reg_or_zero_operand): New predicate.
7612 * config/rs6000/rs6000-protos.h (output_isel): Delete.
7613 * config/rs6000/rs6000.c (output_isel): Delete.
7614 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
7615 instead of reg_or_cint_operand. Output instruction directly (not via
7617 (isel_unsigned_<mode>): Ditto.
7618 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
7619 gpc_reg_operand. Add an instruction alternative for this. Output
7620 instruction directly.
7621 (*isel_reversed_unsigned_<mode>): Ditto.
7623 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
7625 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
7626 (TARGET_CANONICALIZE_COMPARISON): Define.
7628 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
7631 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
7632 Check whether the dest is REG before adding REG_EQUIV note.
7634 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
7637 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
7639 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
7640 (make_hard_regno_born, make_hard_regno_dead): Update
7641 bb_killed_pseudos and bb_gen_pseudos.
7643 2017-10-11 Nathan Sidwell <nathan@acm.org>
7645 * incpath.h (enum incpath_kind): Name enum, prefix values.
7646 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
7647 * incpath.c (heads, tails): Use INC_MAX.
7648 (add_env_var_paths, add_standard_paths): Use incpath_kind.
7649 (merge_include_chains, split_quote_chain,
7650 register_include_chains): Update incpath_kind names.
7651 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
7652 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
7654 (add_framework_path, darwin_register_objc_includes): Likewise.
7655 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
7657 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
7659 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
7660 Do not use float_operator operator predicate.
7661 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
7662 * config/i386/predicates.md (float_operator): Remove predicate.
7664 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
7666 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
7667 (*jccxf_i387): Ditto.
7668 (*jcc<mode>_i387): Ditto.
7669 (*jccu<mode>_i387): Ditto.
7670 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
7671 (*jcc_*_i387 splitters): Remove.
7672 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
7673 * config/i386/i386.c (ix86_split_fp_branch): Remove.
7674 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
7677 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
7679 * profile-count.h (slow_safe_scale_64bit): New function.
7680 (safe_scale_64bit): New inline.
7681 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
7682 * profile-count.c: Include wide-int.h
7683 (slow_safe_scale_64bit): New.
7685 2017-10-11 Nathan Sidwell <nathan@acm.org>
7687 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
7688 HAS_DECL_ASSEMBLER_NAME_P.
7689 * gimple-expr.c (gimple_decl_printable_name: Check
7690 HAS_DECL_ASSEMBLER_NAME_P too.
7691 * ipa-utils.h (type_in_anonymous_namespace_p): Check
7692 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
7693 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
7694 * passes.c (rest_of_decl_compilation): Check
7695 HAS_DECL_ASSEMBLER_NAME_P too.
7696 * recog.c (verify_changes): Likewise.
7697 * tree-pretty-print.c (dump_decl_name): Likewise.
7698 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
7700 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
7701 (SET_DECL_ASSEMBLER_NAME): Use it.
7702 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
7703 (COPY_DECL_ASSEMBLER_NAME): Likewise.
7704 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
7706 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
7708 * config.gcc (i386, x86_64): Add extra objects.
7709 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
7710 (ix86_min_insn_size): Declare.
7711 (ix86_issue_rate): Declare.
7712 (ix86_adjust_cost): Declare.
7713 (ia32_multipass_dfa_lookahead): Declare.
7714 (ix86_macro_fusion_p): Declare.
7715 (ix86_macro_fusion_pair_p): Declare.
7716 (ix86_bd_has_dispatch): Declare.
7717 (ix86_bd_do_dispatch): Declare.
7718 (ix86_core2i7_init_hooks): Declare.
7719 (ix86_atom_sched_reorder): Declare.
7720 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
7721 (COSTS_N_BYTES): Move to x86-tune-costs.h.
7722 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
7723 (rip_relative_addr_p): Rename to ...
7724 (ix86_rip_relative_addr_p): ... this one; export.
7725 (memory_address_length): Update.
7726 (ix86_issue_rate): Move to x86-tune-sched.c.
7727 (ix86_flags_dependent): Move to x86-tune-sched.c.
7728 (ix86_agi_dependent): Move to x86-tune-sched.c.
7729 (exact_dependency_1): Move to x86-tune-sched.c.
7730 (exact_store_load_dependency): Move to x86-tune-sched.c.
7731 (ix86_adjust_cost): Move to x86-tune-sched.c.
7732 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
7733 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
7734 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
7735 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
7736 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
7737 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
7738 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
7739 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
7740 (min_insn_size): Rename to ...
7741 (ix86_min_insn_size): ... this one; export.
7742 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
7743 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
7744 (core2i7_first_cycle_multipass_backtrack): Move to
7745 x86-tune-sched-core.c.
7746 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
7747 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
7748 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
7749 (ix86_avoid_jump_mispredicts): Update.
7750 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
7751 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
7752 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
7753 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
7754 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
7755 (MAX_INSN): Move to ix86-tune-sched-bd.c.
7756 (MAX_IMM): Move to ix86-tune-sched-bd.c.
7757 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
7758 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
7759 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
7760 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
7761 (MAX_STORE): Move to ix86-tune-sched-bd.c.
7762 (BIG): Move to ix86-tune-sched-bd.c.
7763 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
7764 (enum insn_path): Move to ix86-tune-sched-bd.c.
7765 (get_mem_group): Move to ix86-tune-sched-bd.c.
7766 (is_cmp): Move to ix86-tune-sched-bd.c.
7767 (dispatch_violation): Move to ix86-tune-sched-bd.c.
7768 (is_branch): Move to ix86-tune-sched-bd.c.
7769 (is_prefetch): Move to ix86-tune-sched-bd.c.
7770 (init_window): Move to ix86-tune-sched-bd.c.
7771 (allocate_window): Move to ix86-tune-sched-bd.c.
7772 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
7773 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
7774 (process_end_window): Move to ix86-tune-sched-bd.c.
7775 (allocate_next_window): Move to ix86-tune-sched-bd.c.
7776 (find_constant): Move to ix86-tune-sched-bd.c.
7777 (get_num_immediates): Move to ix86-tune-sched-bd.c.
7778 (has_immediate): Move to ix86-tune-sched-bd.c.
7779 (get_insn_path): Move to ix86-tune-sched-bd.c.
7780 (get_insn_group): Move to ix86-tune-sched-bd.c.
7781 (count_num_restricted): Move to ix86-tune-sched-bd.c.
7782 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
7783 (add_insn_window): Move to ix86-tune-sched-bd.c.
7784 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
7785 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
7786 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
7787 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
7788 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
7789 (do_dispatch): Move to ix86-tune-sched-bd.c.
7790 (has_dispatch): Move to ix86-tune-sched-bd.c.
7791 * config/i386/t-i386: Add new object files.
7792 * config/i386/x86-tune-costs.h: New file.
7793 * config/i386/x86-tune-sched-atom.c: New file.
7794 * config/i386/x86-tune-sched-bd.c: New file.
7795 * config/i386/x86-tune-sched-core.c: New file.
7796 * config/i386/x86-tune-sched.c: New file.
7798 2017-10-11 Liu Hao <lh_mouse@126.com>
7800 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
7801 the generic version below instead.
7802 (should_colorize): Recognize Windows consoles as terminals
7804 * pretty-print.c [__MINGW32__] (write_all): New function.
7805 [__MINGW32__] (find_esc_head): Likewise.
7806 [__MINGW32__] (find_esc_terminator): Likewise.
7807 [__MINGW32__] (eat_esc_sequence): Likewise.
7808 [__MINGW32__] (mingw_ansi_fputs): New function that handles
7810 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
7813 2017-10-11 Richard Biener <rguenther@suse.de>
7815 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
7816 Properly call analyze_scalar_evolution with the loop of the stmt.
7818 2017-10-11 Richard Biener <rguenther@suse.de>
7820 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
7821 * tree-core.h (tree_base): Add chrec_var union member.
7822 * tree.h (CHREC_VAR): Remove.
7823 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
7824 * tree-chrec.h (build_polynomial_chrec): Adjust.
7825 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
7826 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
7828 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
7830 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
7831 * match.pd: ... here.
7832 ((T) X == (T) Y): Relax condition.
7834 2017-10-11 Bin Cheng <bin.cheng@arm.com>
7836 PR tree-optimization/82472
7837 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
7839 (break_alias_scc_partitions): Update postorder number.
7841 2017-10-11 Martin Liska <mliska@suse.cz>
7844 * opts.c (parse_no_sanitize_attribute): Do not use error_value
7846 * opts.h (parse_no_sanitize_attribute): Remove last argument.
7848 2017-10-11 Martin Liska <mliska@suse.cz>
7850 * print-rtl.c (print_insn): Move declaration of idbuf
7851 to same scope as name.
7853 2017-10-11 Martin Liska <mliska@suse.cz>
7858 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
7861 2017-10-11 Martin Liska <mliska@suse.cz>
7864 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
7867 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
7869 * config.gcc (default_gnu_indirect_function): Default to yes for
7870 arm*-*-linux* with glibc.
7872 2017-10-11 Richard Biener <rguenther@suse.de>
7874 * tree-scalar-evolution.c (get_scalar_evolution): Handle
7875 default-defs and types we do not want to analyze.
7876 (interpret_loop_phi): Replace unreachable code with an assert.
7877 (compute_scalar_evolution_in_loop): Remove and inline ...
7878 (analyze_scalar_evolution_1): ... here, replacing condition with
7879 what makes the intent clearer. Remove handling of cases
7880 get_scalar_evolution now handles.
7882 2017-10-10 Jim Wilson <wilson@tuliptree.org>
7884 PR rtl-optimization/81434
7885 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
7886 comment for main loop. In sched_group_found if, also add checks for
7887 pass and min_cost_group.
7889 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
7891 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
7892 (rs6000_insn_cost): New function.
7893 * config/rs6000/rs6000.md (cost): New attribute.
7895 2017-10-10 Jakub Jelinek <jakub@redhat.com>
7896 H.J. Lu <hongjiu.lu@intel.com>
7900 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
7901 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
7902 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
7903 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
7904 to OPTION_MASK_ISA_AVX512VL - builtins that have both
7905 OPTION_MASK_ISA_MMX and some other bit set require both
7906 mmx and the ISAs without the mmx bit.
7907 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
7908 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
7909 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
7910 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
7911 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
7912 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
7913 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
7914 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
7915 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
7916 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
7917 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
7918 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
7919 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
7920 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
7921 Add OPTION_MASK_ISA_MMX.
7923 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
7925 * config.gcc (armv7*-*-freebsd*): New target.
7926 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
7928 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
7930 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
7931 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
7934 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
7936 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
7937 a template parameter.
7938 (WIDE_INT_REF_FOR): Update accordingly.
7939 * tree.h (wi::int_traits <const_tree>): Delete.
7940 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
7941 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
7942 (wi::tree_to_wide_ref): New typedef.
7943 (wi::to_wide): New function.
7944 * calls.c (get_size_range): Use wi::to_wide when operating on
7946 * cgraph.c (cgraph_node::create_thunk): Likewise.
7947 * config/i386/i386.c (ix86_data_alignment): Likewise.
7948 (ix86_local_alignment): Likewise.
7949 * dbxout.c (stabstr_O): Likewise.
7950 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
7951 * expr.c (const_vector_from_tree): Likewise.
7952 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
7953 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
7954 (fold_negate_expr_1, int_const_binop_1, const_binop)
7955 (fold_convert_const_int_from_real, optimize_bit_field_compare)
7956 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
7957 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
7958 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
7959 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
7960 (fold_not_const, round_up_loc): Likewise.
7961 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
7962 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
7963 (alloca_call_type): Likewise.
7964 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
7965 * godump.c (go_output_typedef): Likewise.
7966 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
7967 * internal-fn.c (get_min_precision): Likewise.
7968 * ipa-cp.c (ipcp_store_vr_results): Likewise.
7969 * ipa-polymorphic-call.c
7970 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
7971 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
7972 (ipa_modify_call_arguments): Likewise.
7973 * match.pd: Likewise.
7974 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
7975 * print-tree.c (print_node_brief, print_node): Likewise.
7976 * stmt.c (expand_case): Likewise.
7977 * stor-layout.c (layout_type): Likewise.
7978 * tree-affine.c (tree_to_aff_combination): Likewise.
7979 * tree-cfg.c (group_case_labels_stmt): Likewise.
7980 * tree-data-ref.c (dr_analyze_indices): Likewise.
7981 (prune_runtime_alias_test_list): Likewise.
7982 * tree-dump.c (dequeue_and_dump): Likewise.
7983 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
7984 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
7985 * tree-pretty-print.c (dump_generic_node): Likewise.
7986 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
7987 (simple_iv_with_niters): Likewise.
7988 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
7989 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
7990 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
7991 * tree-ssa-loop-niter.c (split_to_var_and_offset)
7992 (refine_value_range_using_guard, number_of_iterations_ne_max)
7993 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
7994 (get_cst_init_from_scev, record_nonwrapping_iv)
7995 (scev_var_range_cant_overflow): Likewise.
7996 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
7997 * tree-ssa-pre.c (compute_avail): Likewise.
7998 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
7999 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
8000 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
8001 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
8002 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
8003 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
8004 (dump_case_nodes, try_switch_expansion): Likewise.
8005 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
8006 (vect_do_peeling): Likewise.
8007 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
8008 * tree-vect-stmts.c (vectorizable_load): Likewise.
8009 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
8010 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
8011 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
8012 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
8013 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
8014 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
8015 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
8016 (evrp_dom_walker::before_dom_children): Likewise.
8017 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
8018 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
8019 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
8020 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
8021 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
8022 (get_range_pos_neg): Likewise.
8023 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
8024 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
8025 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
8026 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
8027 * config/avr/avr.c (avr_fold_builtin): Likewise.
8028 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
8029 * config/msp430/msp430.c (msp430_attr): Likewise.
8030 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
8031 * config/powerpcspe/powerpcspe-c.c
8032 (altivec_resolve_overloaded_builtin): Likewise.
8033 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
8034 (rs6000_expand_ternop_builtin): Likewise.
8035 * config/rs6000/rs6000-c.c
8036 (altivec_resolve_overloaded_builtin): Likewise.
8037 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
8038 (rs6000_expand_ternop_builtin): Likewise.
8039 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
8041 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8043 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
8044 when copying loop nest with only one inner loop.
8046 2017-10-10 Richard Biener <rguenther@suse.de>
8048 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
8049 blocks if SCEV is active.
8050 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
8052 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
8053 (scev_initialize): Assert we are not yet initialized.
8055 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8057 * tree-loop-distribution.c (generate_loops_for_partition): Remove
8058 inner loop's exit stmt by making it always exit the loop, otherwise
8059 we would generate an infinite empty loop.
8061 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8063 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
8064 renaming variables in new preheader if it's deleted.
8066 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8068 * tree-loop-distribution.c (struct partition): Remove unused field
8069 loops of the structure.
8070 (partition_alloc, partition_free): Ditto.
8071 (build_rdg_partition_for_vertex): Ditto.
8073 2017-10-09 Jeff Law <law@redhat.com>
8075 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
8076 return type to match prototype and documentation.
8078 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8080 * config/rs6000/rs6000.c (processor_costs): Move to ...
8081 * config/rs6000/rs6000.h: ... here.
8082 (rs6000_cost): Declare.
8084 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
8086 * except.c (setjmp_fn): New global variable.
8087 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
8088 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
8089 if DONT_USE_BUILTIN_SETJMP is defined.
8091 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8093 * target.def (insn_cost): New hook.
8094 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
8095 * doc/tm.texi: Regenerate.
8096 * rtlanal.c (insn_cost): Use the new hook.
8098 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8100 * combine.c (combine_validate_cost): Compute the new insn_cost,
8101 not just pattern_cost.
8102 (try_combine): Adjust comment.
8104 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8106 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
8108 * combine.c (uid_insn_cost): Adjust comment.
8109 (combine_validate_cost): Adjust comment. Use pattern_cost instead
8111 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
8112 * dse.c (find_shift_sequence): Ditto.
8113 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
8114 (bb_valid_for_noce_process_p): Use pattern_cost.
8115 * rtl.h (insn_rtx_cost): Delete.
8116 (pattern_cost): New prototype.
8117 (insn_cost): New prototype.
8118 * rtlanal.c (insn_rtx_cost): Rename to...
8119 (pattern_cost): ... this.
8122 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
8124 * config/i386/i386.md (*jcc_2): Remove insn pattern.
8125 (*jcc<mode>_0_r_i387): Ditto.
8126 (*jccxf_r_i387): Ditto.
8127 (*jcc<mode>_r_i387): Ditto.
8128 (*jccu<mode>_r_i387): Ditto.
8129 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
8130 (*jcc): Rename from *jcc_1.
8132 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8134 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
8135 deferred rescans after the lvx/stvx recombination pre-pass.
8137 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
8139 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
8140 memory operation instruction support.
8141 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
8142 (rs6000-ibm-aix[789]*): Likewise.
8143 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
8144 Document new functions.
8146 2017-10-09 Richard Biener <rguenther@suse.de>
8148 PR tree-optimization/82397
8149 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
8150 equality only for semantically equal trees.
8152 2017-10-09 Richard Biener <rguenther@suse.de>
8154 PR tree-optimization/82449
8155 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
8156 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
8157 allow constant addresses.
8158 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
8161 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8163 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
8166 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8169 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
8172 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8175 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
8177 2017-10-09 Jakub Jelinek <jakub@redhat.com>
8180 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
8181 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
8183 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
8185 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
8186 (WI_BINARY_PREDICATE_RESULT): Likewise.
8187 (wi::binary_traits::operator_result): New type.
8188 (wi::binary_traits::predicate_result): Likewise.
8189 (generic_wide_int::operator~, unary generic_wide_int::operator-)
8190 (generic_wide_int::operator==, generic_wide_int::operator!=)
8191 (generic_wide_int::operator&, generic_wide_int::and_not)
8192 (generic_wide_int::operator|, generic_wide_int::or_not)
8193 (generic_wide_int::operator^, generic_wide_int::operator+
8194 (binary generic_wide_int::operator-, generic_wide_int::operator*):
8196 (operator~, unary operator-, operator==, operator!=, operator&)
8197 (operator|, operator^, operator+, binary operator-, operator*): New
8199 * expr.c (get_inner_reference): Use wi::bit_and_not.
8200 * fold-const.c (fold_binary_loc): Likewise.
8201 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
8202 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
8203 (bit_value_binop): Likewise.
8204 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
8205 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
8206 (extract_range_from_binary_expr_1): Likewise.
8207 (masked_increment): Likewise.
8208 (simplify_bit_ops_using_ranges): Likewise.
8210 2017-10-09 Martin Jambor <mjambor@suse.cz>
8213 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
8214 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
8215 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
8217 (hsa_fixup_mov_insn_type): New function.
8218 (hsa_op_with_type::get_in_type): Use it.
8219 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
8220 immediates in an assert.
8221 (hsa_op_with_type::extend_int_to_32bit): New method.
8222 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
8223 types. Convert to dest type if necessary.
8224 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
8225 (reg_for_gimple_ssa): Pass false as min32int to
8226 hsa_type_for_scalar_tree_type.
8227 (gen_hsa_addr): Fixup type when creating addresable temporary.
8228 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
8229 (gen_hsa_unary_operation): Extend operands and convert to dest type if
8230 necessary. Call hsa_fixup_mov_insn_type.
8231 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
8232 extend operands and convert to dest type if necessary.
8233 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
8234 to dest type if necessary.
8235 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
8236 if conversion nt necessary and size matches.
8237 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
8238 to dest type if necessary.
8239 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
8240 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
8242 (gen_hsa_clrsb): Likewise.
8243 (gen_hsa_ffs): Likewise.
8244 (gen_hsa_divmod): Extend operands and convert to dest type if
8246 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
8248 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
8250 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
8251 Remove empty default arguments. Use a brace block as output
8253 (conditional return): Ditto.
8255 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
8256 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
8258 (group_ending_nop): Ditto.
8259 (doloop_end): Ditto.
8260 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
8261 (splitters for those): Ditto.
8263 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
8265 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
8266 a conditional jump (and the compare for it) so that pc_rtx is the
8268 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
8269 for the deleted and renamed ctr<mode>_internal[234] patterns.
8270 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
8271 Delete second conditional return pattern.
8272 (ctr<mode>_internal2): Delete this second bdnz pattern.
8273 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
8274 (ctr<mode>_internal4): Delete this second bdz pattern.
8276 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
8278 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
8279 (always_initialized_rtx_for_ssa_name_p): New predicate.
8280 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
8281 (finish_out_of_ssa): Free new field of SA.
8282 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
8283 * tree-ssa-coalesce.c: Include tree-ssa.h.
8284 (get_parm_default_def_partitions): Remove extern keyword.
8285 (get_undefined_value_partitions): New function.
8286 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
8287 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
8289 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
8291 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
8293 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
8295 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
8297 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
8298 for targets that preffer 128bit.
8300 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
8302 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
8304 2017-10-08 Olivier Hainque <hainque@adacore.com>
8306 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
8307 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
8309 (thumb_set_return_address): Likewise.
8311 2017-10-08 Olivier Hainque <hainque@adacore.com>
8313 * common/config/arm/arm-common.c (arm_except_unwind_info):
8314 Handle DWARF2_UNWIND_INFO.
8316 2017-10-07 Michael Collison <michael.collison@arm.com>
8318 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
8321 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
8323 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
8324 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
8325 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
8326 defined, force the creation of a new block for a dispatch label.
8328 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
8330 * invoke.texi (Wsuggest-attribute=cold): Document.
8331 * common.opt (Wsuggest-attribute=cold): New
8332 * ipa-pure-const.c (warn_function_cold): New function.
8333 * predict.c (compute_function_frequency): Use it.
8334 * predict.h (warn_function_cold): Declare.
8336 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
8338 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
8341 2017-10-06 Martin Liska <mliska@suse.cz>
8343 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
8344 keyword for member functions.
8345 (struct sanopt_tree_couple): New struct.
8346 (struct sanopt_tree_couple_hash): New function.
8347 (struct sanopt_ctx): Add new hash_map.
8348 (has_dominating_ubsan_ptr_check): New function.
8349 (record_ubsan_ptr_check_stmt): Likewise.
8350 (maybe_optimize_ubsan_ptr_ifn): Likewise.
8351 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
8352 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
8354 2017-10-06 Sudakshina Das <sudi.das@arm.com>
8357 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
8358 aarch64_simd_valid_immediate on CONST_VECTORs.
8359 (aarch64_reg_or_bic_imm): Likewise.
8361 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
8363 PR rtl-optimization/82396
8364 * haifa-sched.c (ready_sort_real): Disable qsort checking.
8366 2017-10-06 Sebastian Pop <sebpop@gmail.com>
8368 * graphite-dependences.c (scop_get_reads): Move code to...
8369 (scop_get_must_writes): Move code to...
8370 (scop_get_may_writes): Move code to...
8371 (scop_get_reads_and_writes): ... here.
8372 (scop_get_dependences): Call scop_get_reads_and_writes.
8374 2017-10-06 Jakub Jelinek <jakub@redhat.com>
8376 PR tree-optimization/82434
8377 * fold-const.h (can_native_encode_type_p,
8378 can_native_encode_string_p): Remove.
8379 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
8380 don't encode anything, just return what would be otherwise returned.
8381 (native_encode_fixed, native_encode_complex, native_encode_vector):
8383 (native_encode_string): Likewise. Inline by hand
8384 can_native_encode_string_p.
8385 (can_native_encode_type_p): Remove.
8386 (can_native_encode_string_p): Remove.
8387 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
8388 STRING_CSTs using can_native_encode_string_p, test all
8389 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
8390 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
8391 argument from native_encode_expr.
8392 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
8393 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
8396 2017-10-06 Richard Biener <rguenther@suse.de>
8398 PR tree-optimization/82397
8399 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
8400 operand_equal_p but rely on data_ref_compare_tree for detecting
8402 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
8403 to match up with dr_group_sort_cmp.
8405 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8408 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
8410 * config/s390/s390-builtin-types.def: Regenerate.
8412 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8415 * config/s390/s390-builtin-types.def: Regenerate.
8416 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
8417 Change flag from B_VXE to B_VX.
8418 (s390_vec_min_dbl): Remove B_VXE flag.
8420 2017-10-06 Richard Biener <rguenther@suse.de>
8422 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
8423 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
8424 translate_isl_ast_to_gimple::is_valid_rename,
8425 translate_isl_ast_to_gimple::get_rename,
8426 translate_isl_ast_to_gimple::get_def_bb_for_const,
8427 translate_isl_ast_to_gimple::get_new_name,
8428 translate_isl_ast_to_gimple::collect_all_ssa_names,
8429 translate_isl_ast_to_gimple::copy_loop_phi_args,
8430 translate_isl_ast_to_gimple::collect_all_ssa_names,
8431 translate_isl_ast_to_gimple::copy_loop_phi_args,
8432 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
8433 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
8434 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
8435 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
8436 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
8437 translate_isl_ast_to_gimple::copy_cond_phi_args,
8438 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
8439 translate_isl_ast_to_gimple::edge_for_new_close_phis,
8440 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
8441 translate_isl_ast_to_gimple::rename_uses,
8442 translate_isl_ast_to_gimple::rename_all_uses): Remove.
8443 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
8444 (set_rename_for_each_def): Likewise.
8445 (graphite_copy_stmts_from_block): Handle debug stmt resetting
8446 here. Handle rewriting SCEV analyzable uses here.
8447 (copy_bb_and_scalar_dependences): Generate code for PHI
8449 (graphite_regenerate_ast_isl): Adjust.
8450 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
8451 (add_write, add_read): New functions.
8452 (build_cross_bb_scalars_def): Use it and simplify.
8453 (build_cross_bb_scalars_use): Likewise.
8454 (graphite_find_cross_bb_scalar_vars): Inline into...
8455 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
8456 simulating out-of-SSA. Compute liveout and add dependencies.
8457 (build_scops): Force an empty entry block.
8458 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
8460 (sese_build_liveouts): Declare.
8461 (sese_trivially_empty_bb_p): Likewise.
8462 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
8463 compute liveout and debug_liveout.
8464 (sese_bad_liveouts_use): Remove.
8465 (sese_reset_debug_liveouts_bb): Likewise.
8466 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
8467 (sese_build_liveouts): Build liveout and debug_liveout and store
8469 (new_sese_info): Adjust.
8470 (free_sese_info): Likewise.
8471 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
8472 do not build liveout here.
8473 (move_sese_in_condition): Adjust region entry.
8474 (scev_analyzable_p): Match up with chrec_apply requirements.
8475 (sese_trivially_empty_bb_p): New.
8476 * tree-into-ssa.c (get_reaching_def): Properly support generating
8477 default-defs for incremental rewrite of anonymous names.
8479 2017-10-06 Richard Biener <rguenther@suse.de>
8481 * graphite-sese-to-poly.c (extract_affine): For casts increasing
8482 precision do not perform modulo reduction.
8484 2017-10-06 Richard Biener <rguenther@suse.de>
8486 PR tree-optimization/82436
8487 * tree-vect-slp.c (vect_supported_load_permutation_p): More
8488 conservatively choose the vectorization factor when checking
8489 whether we can perform the required load permutation.
8490 (vect_transform_slp_perm_load): Assert when we may not fail.
8492 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
8494 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
8495 message for incompatible -msdata=* and -mcall-* options.
8497 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
8499 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
8500 rate for post-reload scheduling.
8502 2017-10-05 Tamar Christina <tamar.christina@arm.com>
8504 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
8506 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
8508 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
8509 to improve monte carlo in scimark.
8511 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
8513 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
8514 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
8515 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
8516 pentium4_cost, nocona_cost): Set reassociation width to 1.
8517 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
8518 width to 2 for fp operations and 1 otherwise.
8519 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
8521 (atom_cost): Set reassociation width to 2.
8522 (slm_cost, generic_cost): Set fp reassociation width
8523 to 2 and 1 otherwise.
8524 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
8525 (core_cost): Set fp reassociation width to 4 and vector to 2.
8526 (ix86_reassociation_width): Rewrite using cost table; special case
8527 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
8528 and TARGET_AVX128_OPTIMAL.
8529 * config/i386/i386.h (processor_costs): Add
8530 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
8531 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
8532 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
8533 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
8534 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
8535 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
8537 2017-10-05 Nathan Sidwell <nathan@acm.org>
8539 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
8541 2017-10-05 Tamar Christina <tamar.christina@arm.com>
8543 * config/arm/arm.c (arm_test_fpu_data): New.
8544 (arm_run_selftests): Call arm_test_fpu_data.
8546 2017-10-04 Nathan Sidwell <nathan@acm.org>
8548 * toplev.c (toplev::main): Remove excess parens on pretty_printer
8550 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
8552 2017-10-04 Sudakshina Das <sudi.das@arm.com>
8554 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
8555 check type for aarch64_simd_valid_immediate.
8556 (aarch64_output_simd_mov_immediate): Update prototype.
8557 (aarch64_simd_valid_immediate): Update prototype.
8558 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
8559 support for ORR-immediate.
8560 (and<mode>3): modified pattern to add support for BIC-immediate.
8561 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
8562 now checks for valid immediate for BIC and ORR based on new enum
8564 (aarch64_output_simd_mov_immediate): Function now used to output
8565 BIC/ORR imm as well based on new enum argument.
8566 * config/aarch64/constraints.md (Do): New vector immediate constraint.
8568 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
8569 (aarch64_reg_or_bic_imm): Likewise.
8571 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8573 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
8574 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
8576 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
8580 PR rtl-optimization/82396
8581 * haifa-sched.c (autopref_multipass_init): Simplify
8583 (autopref_rank_data): Simplify sort order.
8584 * sched-int.h (autopref_multipass_data_): Remove
8585 multi_mem_insn_p, min_offset and max_offset.
8587 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8589 * doc/sourcebuild.texi: Document vect_peeling_profitable.
8591 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8593 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
8596 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8598 * doc/sourcebuild.texi: Document vect_long_mult.
8600 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
8602 PR tree-optimization/82413
8603 * fold-const.c (build_range_check): Use widest_int when comparing
8604 the maximum ETYPE value with HIGH.
8606 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
8608 PR rtl-optimization/82396
8609 * haifa-sched.c (autopref_multipass_init): Simplify
8611 (autopref_rank_data): Simplify sort order.
8612 * sched-int.h (autopref_multipass_data_): Remove
8613 multi_mem_insn_p, min_offset and max_offset.
8615 2017-10-04 Jakub Jelinek <jakub@redhat.com>
8617 PR tree-optimization/82381
8618 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
8619 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
8622 PR tree-optimization/82374
8623 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
8624 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
8625 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
8626 current_function_decl to the new decl.
8628 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
8630 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
8631 helper macro for IEEE float128 hardware built-in functions.
8632 (SQRTF128_ODD): Add built-in functions with the round-to-odd
8634 (TRUNCF128_ODD): Likewise.
8635 (ADDF128_ODD): Likewise.
8636 (SUBF128_ODD): Likewise.
8637 (MULF128_ODD): Likewise.
8638 (DIVF128_ODD): Likewise.
8639 (FMAF128_ODD): Likewise.
8640 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
8641 UNSPEC_TRUNC_ROUND_TO_ODD.
8642 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
8643 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
8644 floating point round to odd instructions.
8645 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
8646 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
8647 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
8648 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
8649 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
8650 (trunc<mode>sf2_hw): Change the truncate with round to odd
8651 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
8652 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
8653 to odd hardware instructions.
8654 (sub<mode>3_odd): Likewise.
8655 (mul<mode>3_odd): Likewise.
8656 (div<mode>3_odd): Likewise.
8657 (sqrt<mode>2_odd): Likewise.
8658 (fma<mode>4_odd): Likewise.
8659 (fms<mode>4_odd): Likewise.
8660 (nfma<mode>4_odd): Likewise.
8661 (nfms<mode>4_odd): Likewise.
8662 (trunc<mode>df2_odd): Change the truncate with round to odd
8663 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
8665 * doc/extend.texi (PowerPC built-in functions): Update documentation
8666 for existing IEEE float128-bit built-in functions. Add built-in
8667 functions that generate the IEEE 128-bit floating point round to
8670 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
8672 PR rtl-optimization/77729
8673 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
8674 to (X&(C1&~C2))|C2 transformations.
8676 2017-10-03 Martin Jambor <mjambor@suse.cz>
8678 PR tree-optimization/82363
8679 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
8680 mismatch, mark lacc written regardless of racc.
8682 2017-10-03 Jakub Jelinek <jakub@redhat.com>
8684 PR tree-optimization/82381
8685 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
8686 stmt_to_insert nor wheather SSA_NAMEs are default defs.
8687 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
8688 fallthrough into reassoc_stmt_dominates_stmt_p.
8691 * combine.c (combine_instructions): Don't combine in unreachable
8694 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
8697 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
8698 function to not use the have_cpu variable. Do not set cpu_index,
8699 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
8701 (rs6000_valid_attribute_p): Remove duplicate initializations of
8702 old_optimize and func_optimize.
8703 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
8704 (rs6000_activate_target_options): Make global.
8705 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
8708 2017-10-02 Jakub Jelinek <jakub@redhat.com>
8710 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
8711 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
8712 Set *poffset to 0 and *psize and *pmax_size to -1 if
8713 *poffset + *psize overflows in HOST_WIDE_INT.
8715 PR tree-optimization/82387
8716 PR tree-optimization/82388
8717 PR tree-optimization/82389
8718 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
8719 instead of live_bytes non-NULL.
8721 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
8724 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
8727 2017-10-02 Richard Biener <rguenther@suse.de>
8729 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
8730 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
8731 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
8733 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
8735 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
8736 requested precision matches the type's.
8737 * calls.c (alloc_max_size): Calculate the new candidate size as
8738 a widest_int and use wi::to_widest when comparing it with the
8739 current candidate size.
8740 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
8741 zero rather than integer_zero_node.
8742 * match.pd: Check for a no-op conversion before using wi::add
8743 rather than after. Use tree_to_uhwi when summing small shift
8744 counts into an unsigned int.
8746 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
8747 Alan Hayward <alan.hayward@arm.com>
8748 David Sherwood <david.sherwood@arm.com>
8751 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
8752 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
8753 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
8754 POINTER_AND_FP_REGS.
8756 2017-10-02 Richard Biener <rguenther@suse.de>
8758 PR tree-optimization/82355
8759 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
8760 a mapping for the enclosing loop but avoid generating one for
8762 (copy_bb_and_scalar_dependences): Remove premature codegen
8763 error on PHIs in blocks duplicated into multiple places.
8764 * graphite-scop-detection.c
8765 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
8766 in the region use it as loop and nest to analyze the DR in.
8767 (try_generate_gimple_bb): Likewise.
8768 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
8769 (add_loop_constraints): For blocks in a loop not in the region
8770 create a dimension with a single iteration.
8771 * sese.h (gbb_loop_at_index): Remove assert.
8773 2017-10-01 Kevin Buettner <kevinb@redhat.com>
8775 * omp-expand.c (adjust_context_scope): New function.
8776 (expand_parallel_call): Call adjust_context_scope.
8778 2017-10-01 Jeff Law <law@redhat.com>
8780 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
8781 dom_opt_dom_walker class with direct access to private members.
8782 Add comments. Call test_for_singularity.
8783 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
8784 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
8785 m_dummy_cond anymore.
8786 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
8788 (pass_dominator:execute): Build the dummy_cond here and pass it
8789 to the dom_opt_dom_walker ctor.
8790 (test_for_singularity): New function.
8792 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
8793 Maya Rashish <coypu@sdf.org>
8795 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
8796 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
8797 (alpha*-*-netbsd*) Use nbsd_tm_file.
8798 (arm*-*-netbsdelf*) Likewise.
8799 (i[34567]86-*-netbsdelf*) Likewise.
8800 (x86_64-*-netbsd*) Likewise.
8801 (mips*-*-netbsd*) Likewise.
8802 (powerpc-*-netbsd*) Likewise.
8803 (sh*-*-netbsd*) Likewise.
8804 (sparc-*-netbsdelf*) Likewise.
8805 (sparc64-*-netbsd*) Likewise.
8806 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
8808 (vax-*-netbsdelf*) Likewise.
8809 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
8810 (UINT_FAST8_TYPE) Likewise.
8811 (INT_FAST16_TYPE) Check CHAR_FAST16.
8812 (UINT_FAST16_TYPE) Likewise.
8814 2017-09-30 Jakub Jelinek <jakub@redhat.com>
8817 * config/i386/i386.md
8818 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
8819 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
8820 *divmodsi4_zext_2): New define_insn_and_split.
8821 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
8822 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
8823 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
8824 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
8825 New define_insn_and_split.
8826 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
8827 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
8828 operands[1] having DImode when mode is SImode.
8830 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
8831 always SImode for DIV and MOD in REG_EQUAL notes.
8833 2017-09-29 Yury Gribov <tetra2005@gmail.com>
8836 * match.pd: Fix handling of NaNs in pattern.
8838 2017-09-29 Jeff Law <law@redhat.com>
8840 * sbitmap.c (bitmap_bit_in_range_p): New function.
8841 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
8842 * tree-ssa-dse.c (live_bytes_read): New function.
8843 (dse_classify_store): Ignore reads of dead bytes.
8845 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
8846 typos and whitespace errors.
8847 * config/i386/predicates.md (address_no_seg_operand): Likewise.
8848 * config/s390/s390.c (s390_emit_prologue): Likewise.
8850 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
8853 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
8854 with a symbol for LRA.
8856 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
8858 PR rtl-optimization/82338
8859 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
8861 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
8863 * genmodes.c (calc_wider_mode): Suppress qsort macro.
8864 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
8865 (qsort_chk): Declare.
8866 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
8867 (qsort_chk): New function.
8869 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8871 PR tree-optimization/82337
8872 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
8873 phi definition if the PHI result appears in an abnormal PHI.
8874 (find_basis_for_base_expr): Don't record a basis if the LHS of the
8875 basis appears in an abnormal PHI.
8877 2017-09-29 Richard Biener <rguenther@suse.de>
8879 * graphite-isl-ast-to-gimple.c
8880 (translate_isl_ast_to_gimple::set_codegen_error): New function.
8881 (binary_op_to_tree): Use it.
8882 (get_rename_from_scev): Likewise.
8883 (copy_loop_phi_nodes): Likewise.
8884 (copy_bb_and_scalar_dependences): Likewise.
8885 (translate_pending_phi_nodes): Likewise.
8887 2017-09-29 Jakub Jelinek <jakub@redhat.com>
8890 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
8891 for movabsq $(i32 << shift), r64.
8893 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
8895 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
8896 index when encoding %esp as %rsp to avoid 0x67 prefix.
8898 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
8900 * config/i386/i386.md (*movsf_internal, *movdf_internal):
8901 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
8903 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
8905 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
8906 Extensions with more than 16 double VFP registers.
8907 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
8908 to_clear_mask and all code related to it. Replace the remaining
8909 entry by a sbitmap and adapt code accordingly.
8911 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
8913 * brig-builtins.def: Change pure attributes to const.
8915 2017-09-28 Joseph Myers <joseph@codesourcery.com>
8917 * config.gcc (default_gnu_indirect_function): Default to yes for
8918 sparc*-*-linux* with glibc.
8920 2017-09-28 Joseph Myers <joseph@codesourcery.com>
8922 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
8923 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
8924 when creating .init_array and .fini_array sections with priority
8927 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
8930 * config/aarch64/aarch64.c
8931 (aarch64_builtin_support_vector_misalignment): Always return false
8932 when misalignment is unknown.
8934 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
8936 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
8937 this function to return false if the definition used by the swap
8938 instruction is artificial, or if the memory address from which the
8939 constant value is loaded is not represented by a base address held
8940 in a register or if the base address register is a frame or stack
8941 pointer. Additionally, return false if the base address of the
8942 loaded constant is a SYMBOL_REF but is not considered to be a
8944 (replace_swapped_load_constant): New function.
8945 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
8946 loaded constant vector with a load of a swapped constant vector.
8948 2017-09-27 Carl Love <cel@us.ibm.com>
8950 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
8951 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
8952 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
8955 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
8957 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
8958 first, always call autopref_rank_data otherwise.
8960 2017-09-27 Richard Biener <rguenther@suse.de>
8962 * graphite-scop-detection.c (find_scop_parameters): Move
8963 loop bound handling ...
8964 (gather_bbs::before_dom_children): ... here, avoiding the need
8965 to build scop_info->loop_nest.
8966 (record_loop_in_sese): Remove.
8967 * sese.h (sese_info_t::loop_nest): Remove.
8968 * sese.c (new_sese_info): Do not allocate loop_nest.
8969 (free_sese_info): Do not free loop_nest.
8971 2017-09-27 Jakub Jelinek <jakub@redhat.com>
8974 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
8975 lhs from calls if the lhs has addressable type.
8977 2017-09-27 Richard Biener <rguenther@suse.de>
8979 * graphite.h (scop::max_alias_set): New member.
8980 * graphite-scop-detection.c: Remove references to non-existing
8981 --param in comments.
8982 (build_alias_sets): Record the maximum alias set used for drs.
8983 (build_scops): Support zero as unlimited for
8984 --param graphite-max-arrays-per-scop.
8985 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
8987 (build_poly_sr_1): ... here. Compute alias set based on the
8988 maximum alias set used for drs rather than
8989 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
8991 2017-09-27 Richard Biener <rguenther@suse.de>
8993 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
8994 --param loop-block-tile-size=0 to disable tiling.
8996 2017-09-27 Richard Biener <rguenther@suse.de>
8998 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
8999 (graphite-max-nb-scop-params): Document special value zero.
9000 * domwalk.h (dom_walker::STOP): New symbolical constant.
9001 (dom_walker::dom_walker): Add optional parameter for bb to
9003 (dom_walker::~dom_walker): Declare.
9004 (dom_walker::before_dom_children): Document STOP return value.
9005 (dom_walker::m_user_bb_to_rpo): New member.
9006 (dom_walker::m_bb_to_rpo): Likewise.
9007 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
9008 mapping here if not provided by the user.
9009 (dom_walker::~dom_walker): Free bb to RPO mapping if not
9010 provided by the user.
9011 (dom_walker::STOP): Define.
9012 (dom_walker::walk): Do not compute bb to RPO mapping here.
9013 Support STOP return value from before_dom_children to stop
9015 * graphite-optimize-isl.c (optimize_isl): If the schedule
9016 is the same still generate code if -fgraphite-identity
9017 or -floop-parallelize-all are given.
9018 * graphite-scop-detection.c: Include cfganal.h.
9019 (gather_bbs::gather_bbs): Get and pass through bb to RPO
9021 (gather_bbs::before_dom_children): Return STOP for BBs
9023 (build_scops): Compute bb to RPO mapping and pass it to
9024 the domwalk. Treat --param graphite-max-nb-scop-params=0
9025 as not limiting the number of params.
9026 * graphite.c (graphite_initialize): Remove limit on the
9027 number of basic-blocks in a function.
9028 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
9029 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
9030 default value of 10.
9032 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9034 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
9035 Adjust code to eliminate needing to do the shift right 32-bits
9036 operation after XSCVDPSPN.
9038 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9040 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
9041 ((X / Y) != 0 -> X >= Y): Likewise.
9043 2017-09-26 Carl Love <cel@us.ibm.com>
9045 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
9046 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
9047 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
9048 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
9049 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
9050 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
9051 definitions and overloading.
9052 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
9053 statement for P9V_BUILTIN_XST_LEN_R.
9054 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
9055 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
9056 define_expand and define_insn for the instructions and builtins.
9057 * doc/extend.texi: Update the built-in documentation file for the new
9059 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
9060 define_insn for the instructions
9062 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
9065 * gcc/config/netbsd-protos.h: New file.
9066 * gcc/config/netbsd.c: New file.
9067 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
9068 * gcc/config/t-netbsd: New file.
9069 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
9070 (tmake_file) Add t-netbsd.
9071 (extra_objs) Add netbsd.o.
9073 2017-09-26 Janus Weil <janus@gcc.gnu.org>
9077 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
9079 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9081 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
9082 sign extension from a vector register to a GPR by doing a 32-bit
9083 direct move and then an EXTSW.
9084 (extendsi<mode>2 splitter): Likewise.
9085 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
9086 right or vector extract after doing XSCVDPSPN. Use
9087 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
9089 (movdi_from_sf_zero_ext): Likewise.
9090 (reload_gpr_from_vsxsf): Likewise.
9091 (p8_mfvsrd_4_disf): Delete, no longer used.
9092 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
9093 then needing to move the SFmode to a GPR to use the XSCVDPSP
9094 instruction instead of FRSP and XSCVDPSPN.
9095 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
9096 it is adjacent to the other XSCVSPDP insns.
9097 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
9098 SFmode to be in traditional Altivec registers.
9099 (vsx_xscvdpspn): Eliminate useless alternative constraint.
9100 (vsx_xscvspdpn): Likewise.
9101 (vsx_xscvspdpn_scalar): Likewise.
9103 2017-09-26 Martin Jambor <mjambor@suse.cz>
9105 * tree-sra.c (compare_access_positions): Put integral types first,
9106 stabilize sorting of integral types, remove conditions putting
9107 non-full-precision integers last.
9108 (sort_and_splice_var_accesses): Disable scalarization if a
9109 non-integert would be represented by a non-full-precision integer.
9111 2017-09-26 Joseph Myers <joseph@codesourcery.com>
9113 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
9114 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
9115 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
9116 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
9117 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
9118 conditionals inside the function instead of around it. Call
9119 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
9120 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
9122 2017-09-26 Richard Biener <rguenther@suse.de>
9124 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
9126 (scop_detection::build_scop_breadth): ... this. Removed.
9127 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
9128 (scop_detection::harmful_stmt_in_bb): Likewise.
9129 (scop_detection::graphite_can_represent_stmt): Likewise.
9130 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
9131 (scop_detection::can_represent_loop): Remove recursion, fold in ...
9132 (scop_detection::can_represent_loop_1): ... this. Removed.
9133 (scop_detection::harmful_loop_in_region): Simplify after inlining
9134 the above and remove more quadraticness.
9135 (build_scops): Adjust.
9136 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
9139 2017-09-26 Jakub Jelinek <jakub@redhat.com>
9142 * config/i386/i386.c (ix86_print_operand_address_as): Only test
9143 REGNO (base) == SP_REG if base is a REG.
9146 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
9147 if it is different SSA_NAME.
9148 (optimize_range_tests_cmp_bitwise): New function.
9149 (optimize_range_tests): Call it.
9151 2017-09-26 Richard Biener <rguenther@suse.de>
9153 PR tree-optimization/82321
9154 * graphite.c (canonicalize_loop_closed_ssa): Properly check
9155 for the def being inside the loop.
9157 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9159 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
9161 * config/s390/s390-builtins.def: Fix constraint on op4.
9163 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9165 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
9166 independent expanders.
9167 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
9168 ("vec_ordered", "vec_unordered"): New expanders.
9170 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9172 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
9175 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9177 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
9178 vec_unpacks_lo_v16qi.
9179 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
9181 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9183 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
9184 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
9185 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
9187 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9189 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
9191 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
9193 ("*vec_slb<mode>"): New insn pattern.
9194 ("vec_shr_<mode>"): New expander.
9195 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
9196 and force the shift count operand to V16QImode.
9197 ("vec_srb<mode>"): Set shift count mode to V16QI.
9199 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9201 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
9202 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
9203 ("vec_widen_smult_hi_<mode>"): New expander definitions.
9205 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
9208 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
9210 2017-09-26 Richard Biener <rguenther@suse.de>
9212 PR tree-optimization/82320
9213 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
9216 2017-09-25 Jeff Law <law@redhat.com>
9218 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
9219 prototype for new argument.
9220 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
9221 mostly extracted from rs6000_emit_allocate_stack.
9222 (rs6000_emit_probe_stack_range_stack_clash): New function.
9223 (rs6000_emit_allocate_stack): Call
9224 rs6000_emit_probe_stack_range_stack_clash as needed.
9225 (rs6000_emit_probe_stack_range): Add additional argument
9226 to call to gen_probe_stack_range{si,di}.
9227 (output_probe_stack_range): New.
9228 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
9229 (output_probe_stack_range_stack_clash): New.
9230 (rs6000_emit_prologue): Emit notes into dump file as requested.
9231 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
9232 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
9233 Add additional operand and pass it to output_probe_stack_range.
9235 2017-09-25 Bin Cheng <bin.cheng@arm.com>
9237 PR tree-optimization/82163
9238 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
9239 (checking_verify_loop_closed_ssa): New parameter.
9240 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
9241 (check_loop_closed_ssa_stmt): Delete.
9242 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
9243 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
9244 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
9247 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
9249 * brig-builtins.def: Treat HSAIL barrier builtins as
9250 setjmp/longjump style functions.
9252 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
9254 * target.def (constant_alignment): New hook.
9255 * defaults.h (CONSTANT_ALIGNMENT): Delete.
9256 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
9257 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
9258 * doc/tm.texi: Regenerate.
9259 * targhooks.h (default_constant_alignment): Declare.
9260 (constant_alignment_word_strings): Likewise.
9261 * targhooks.c (default_constant_alignment): New function.
9262 (constant_alignment_word_strings): Likewise.
9263 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
9264 instead of CONSTANT_ALIGNMENT.
9265 * varasm.c (align_variable, get_variable_align, build_constant_desc)
9266 (force_const_mem): Likewise.
9267 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
9268 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
9269 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
9270 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9271 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
9273 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
9274 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9275 constant_alignment_word_strings.
9276 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
9277 (CONSTANT_ALIGNMENT): Likewise.
9278 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9279 (arm_constant_alignment): New function.
9280 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
9281 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9282 constant_alignment_word_strings.
9283 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
9284 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9285 constant_alignment_word_strings.
9286 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
9287 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9288 (cris_constant_alignment): New function.
9289 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
9290 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9291 (epiphany_constant_alignment): New function.
9292 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
9293 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9294 constant_alignment_word_strings.
9295 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
9296 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9297 constant_alignment_word_strings.
9298 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
9299 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9300 constant_alignment_word_strings.
9301 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
9302 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
9303 * config/i386/i386.c (ix86_constant_alignment): Make static.
9304 Use the same interface as the target hook.
9305 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9306 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
9307 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9308 constant_alignment_word_strings.
9309 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
9310 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
9311 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9312 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
9313 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9314 constant_alignment_word_strings.
9315 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
9316 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9317 constant_alignment_word_strings.
9318 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
9319 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9320 constant_alignment_word_strings.
9321 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
9322 * config/microblaze/microblaze.c (microblaze_constant_alignment):
9324 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9325 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
9326 * config/mips/mips.c (mips_constant_alignment): New function.
9327 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9328 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
9329 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
9330 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9331 (mmix_constant_alignment): Make static. Use the same interface
9333 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
9334 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9335 constant_alignment_word_strings.
9336 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
9337 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9338 constant_alignment_word_strings.
9339 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
9340 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9341 constant_alignment_word_strings.
9342 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
9343 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9344 (rs6000_constant_alignment): New function.
9345 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
9346 * config/riscv/riscv.c (riscv_constant_alignment): New function.
9347 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9348 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
9349 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9350 (rs6000_constant_alignment): New function.
9351 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
9352 * config/s390/s390.c (s390_constant_alignment): New function.
9353 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9354 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
9355 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9356 constant_alignment_word_strings.
9357 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
9358 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9359 (sparc_constant_alignment): New function.
9360 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
9361 * config/spu/spu.c (spu_constant_alignment): New function.
9362 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9363 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
9364 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9365 constant_alignment_word_strings.
9366 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
9367 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9368 constant_alignment_word_strings.
9369 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
9370 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9371 constant_alignment_word_strings.
9372 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
9373 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9374 (visium_constant_alignment): New function.
9375 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
9376 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9377 (xtensa_constant_alignment): New function.
9378 * system.h (CONSTANT_ALIGNMENT): Poison.
9380 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
9382 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
9383 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
9384 (rs6000_builtin_valid_without_lhs): New helper function.
9385 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9386 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
9388 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
9390 * target.h (vec_perm_indices): Use unsigned short rather than
9392 (auto_vec_perm_indices): Likewise.
9393 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
9394 Use unsigned int rather than unsigned char.
9395 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
9397 2017-09-25 Richard Biener <rguenther@suse.de>
9399 * cfgloop.h (sort_sibling_loops): Declare.
9400 * cfgloop.c (sort_sibling_loops_cmp): New helper.
9401 (sort_sibling_loops): New function sorting the sibling loop list
9403 * graphite.c (graphite_transform_loops): Sort sibling loops.
9405 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
9407 * target.def (vec_perm_const_ok): Change sel parameter to
9409 * optabs-query.c (can_vec_perm_p): Update accordingly.
9410 * doc/tm.texi: Regenerate.
9411 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
9412 auto_vec_perm_indices and remove separate nelt field.
9413 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
9414 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
9415 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
9416 (aarch64_expand_vec_perm_const): Update accordingly.
9417 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
9418 to vec_perm_indices.
9419 * config/arm/arm.c (expand_vec_perm_d): Change perm to
9420 auto_vec_perm_indices and remove separate nelt field.
9421 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
9422 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
9423 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
9425 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
9426 to vec_perm_indices.
9427 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
9428 sel to vec_perm_indices.
9429 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
9430 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
9431 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
9433 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
9436 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
9439 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
9440 on the FUNCTION_DECL function context if it has a DIE that is a
9443 2017-09-25 Richard Biener <rguenther@suse.de>
9445 PR tree-optimization/82285
9446 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
9449 2017-09-25 Tom de Vries <tom@codesourcery.com>
9453 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
9456 2017-09-25 Richard Biener <rguenther@suse.de>
9458 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
9459 ISL errors other than isl_error_quota happen. Dump if the
9460 schedule is the same.
9461 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
9462 errors instead of aborting inside ISL.
9464 2017-09-25 Iain Sandoe <iain@codesourcery.com>
9467 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
9468 of libgcc_eh for m64.
9469 * config/i386/darwin64.h: Likewise.
9471 2017-09-25 Richard Biener <rguenther@suse.de>
9474 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
9475 attribute for incomplete types nor twice for complete ones.
9477 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
9480 * config/i386/i386.c (ix86_print_operand_address_as): Encode
9481 %esp as %rsp to avoid 0x67 prefix if there is no index or base
9484 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
9487 * config/i386/i386.opt (mprefer-avx256): Use
9488 ix86_target_flags variable.
9489 * config/i386/i386.c (ix86_target_string): Move
9490 -mprefer-avx256 to flag2_opts.
9492 2017-09-22 Jakub Jelinek <jakub@redhat.com>
9495 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
9496 and x != -1 | y != -1 into (x & y) != -1.
9498 2017-09-22 Steve Ellcey <sellcey@cavium.com>
9500 * config.gcc: Add new case statement to set
9501 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
9502 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
9503 s390x-*-linux* case statements. Added aarch64 to the list of
9504 supported architectures.
9506 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9508 PR tree-optimization/82289
9509 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
9510 STMT_VINFO_RELEVANT_P.
9512 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9513 Alan Hayward <alan.hayward@arm.com>
9514 David Sherwood <david.sherwood@arm.com>
9516 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
9517 for VR_RANGE only; don't allow VR_ANTI_RANGE.
9518 (extract_range_from_binary_expr_1): Don't call
9519 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
9521 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9522 Alan Hayward <alan.hayward@arm.com>
9523 David Sherwood <david.sherwood@arm.com>
9525 * target.def (preferred_vector_alignment): New hook.
9526 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
9528 * doc/tm.texi: Regenerate.
9529 * targhooks.h (default_preferred_vector_alignment): Declare.
9530 * targhooks.c (default_preferred_vector_alignment): New function.
9531 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
9533 (DR_TARGET_ALIGNMENT): New macro.
9534 (aligned_access_p): Update commentary.
9535 (vect_known_alignment_in_bytes): New function.
9536 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
9538 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
9539 Calculate the misalignment based on the target alignment rather than
9541 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
9542 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
9543 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
9544 the target alignment, rather than masking the element misalignment
9545 with the number of elements in a vector. Also use the target
9546 alignment when calculating the maximum number of peels.
9547 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
9548 instead of TYPE_ALIGN_UNIT.
9549 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
9550 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
9551 (vect_create_addr_base_for_vector_ref): Update call accordingly.
9552 (vect_create_data_ref_ptr): Likewise.
9553 (vect_setup_realignment): Realign by ANDing with
9554 -DR_TARGET_MISALIGNMENT.
9555 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
9556 the number of peels based on DR_TARGET_ALIGNMENT.
9557 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
9558 with the guaranteed alignment boundary when deciding whether
9560 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
9561 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
9562 (ensure_base_align): Remove stmt_info parameter. Get the
9563 target base alignment from DR_TARGET_ALIGNMENT.
9564 (vectorizable_store): Update call accordingly. Interpret
9565 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
9567 (vectorizable_load): Likewise.
9569 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9570 Alan Hayward <alan.hayward@arm.com>
9571 David Sherwood <david.sherwood@arm.com>
9573 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
9574 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
9575 (vect_enhance_data_refs_alignment): Likewise.
9577 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
9579 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
9580 error. Only quit immediately if parsing is complete.
9581 (BEGIN): Initialize fatal_err and parse_done.
9582 (begin fpu, end fpu): Check number of arguments.
9583 (begin arch, end arch): Likewise.
9584 (begin cpu, end cpu): Likewise.
9585 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
9586 (optalias): Likewise.
9588 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
9590 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
9591 * config/arm/arm-isa.h: Delete. Move definitions to ...
9592 * arm-cpus.in: ... here. Use new feature and fgroup values.
9593 * config/arm/arm.c (arm_option_override): Use lower case for feature
9595 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
9596 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
9597 * config/arm/parsecpu.awk (END): Add new command 'isa'.
9599 (print_isa_bits_for): New function.
9600 (gen_isa): New function.
9601 (gen_comm_data): Use print_isa_bits_for.
9602 (define feature): New keyword.
9603 (define fgroup): New keyword.
9604 * config/arm/t-arm (TM_H): Remove.
9605 (GTM_H): Add arm-isa.h.
9606 (arm-isa.h): Add rule to generate file.
9607 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
9608 case for feature bit names.
9610 2017-09-22 Richard Biener <rguenther@suse.de>
9612 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
9614 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
9615 print of no dependency loops ...
9616 * graphite.c (graphite_transform_loops): ... here.
9617 (canonicalize_loop_closed_ssa_form): Work from inner to outer
9619 (same_close_phi_node, remove_duplicate_close_phi,
9620 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
9621 (canonicalize_loop_closed_ssa): ... here and simplify.
9622 * graphite-optimize-isl.c: Include tree-vectorizer.h.
9623 (optimize_isl): Use dump_printf_loc to tell when we stopped
9624 optimizing because of an ISL timeout.
9626 2017-09-22 Richard Biener <rguenther@suse.de>
9628 PR tree-optimization/82291
9629 * tree-if-conv.c (predicate_mem_writes): Make sure to
9630 remove writes in blocks predicated with false.
9632 2017-09-22 Richard Biener <rguenther@suse.de>
9634 * sese.c: Include cfganal.h.
9635 (if_region_set_false_region): Remove.
9636 (create_if_region_on_edge): Likewise.
9637 (move_sese_in_condition): Re-implement without destroying
9640 2017-09-22 Richard Biener <rguenther@suse.de>
9642 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
9643 Verify both BBs contain loop PHI nodes before dispatching to
9645 (graphite_regenerate_ast_isl): Do not recompute dominators,
9646 do not verify three times. Restructure for clarity.
9647 * graphite-scop-detection.c (same_close_phi_node,
9648 remove_duplicate_close_phi, make_close_phi_nodes_unique,
9649 defined_in_loop_p, canonicalize_loop_closed_ssa,
9650 canonicalize_loop_closed_ssa_form): Simplify, remove excess
9651 checking and SSA rewrite, move to ...
9652 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
9653 (graphite_initialize): Do not pass in ctx, do not reset the
9654 SCEV cache, compute only dominators.
9655 (graphite_transform_loops): Allocate ISL ctx after
9656 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
9657 Maintain post-dominators only around build_scops.
9658 * sese.c (if_region_set_false_region): Make static. Free
9659 and recompute dominators.
9660 (move_sese_in_condition): Assert we don't get called with
9661 post-dominators computed.
9662 * sese.h (if_region_set_false_region): Remove.
9664 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
9666 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
9667 mode attribute for TARGET_AVX512VL.
9669 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
9671 * config/i386/i386.opt (mprefer-avx256): New option.
9672 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
9674 (ix86_preferred_simd_mode): Return 256-bit AVX modes
9675 for TARGET_PREFER_AVX256.
9676 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
9678 2017-09-21 Jeff Law <law@redhat.com>
9680 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
9681 Fix dump output if the only stack space is for pushed registers.
9683 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9685 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
9688 2017-09-21 Martin Sebor <msebor@redhat.com>
9691 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
9692 code (in C++) or code that triggers warnings.
9694 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
9696 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
9698 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
9700 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
9701 * sched-rgn.c: Ditto.
9702 * sel-sched-ir.c: Ditto.
9704 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
9706 * toplev.h (set_random_seed): Adjust return type.
9707 * toplev.c (init_local_tick): Move eager initialization of random_seed
9708 to get_random_seed. Adjust comment.
9709 (init_random_seed): Inline to get_random_seed, delete.
9710 (get_random_seed): Initialize random_seed lazily.
9711 (set_random_seed): Do not return previous value.
9712 (print_switch_value): Do not call get_random_seed.
9714 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
9716 * cgraph.c (delete_function_version): New, broken out from...
9717 (cgraph_node::delete_function_version): ...here. Rename to
9718 cgraph_node::delete_function_version_by_decl. Update all uses.
9719 (cgraph_node::remove): Call delete_function_version.
9721 2017-09-21 Jakub Jelinek <jakub@redhat.com>
9724 * tree-inline.c (expand_call_inline): Emit clobber stmts for
9725 VAR_DECLs to which addressable non-volatile parameters are mapped
9726 and for id->retvar after the return value assignment. Clear
9727 id->retval and id->retbnd after inlining.
9729 2017-09-21 Richard Biener <rguenther@suse.de>
9731 PR tree-optimization/82276
9732 PR tree-optimization/82244
9733 * tree-vrp.c (build_assert_expr_for): Set
9734 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
9736 (remove_range_assertions): Revert earlier change.
9738 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
9741 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
9743 2017-09-21 Richard Biener <rguenther@suse.de>
9745 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
9746 Restore valid IL after code generation errors.
9747 * graphite.c (graphite_transform_loops): Diagnose code
9748 generation issues as MSG_MISSED_OPTIMIZATION and continue
9749 with processing SCOPs.
9751 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9752 Alan Hayward <alan.hayward@arm.com>
9753 David Sherwood <david.sherwood@arm.com>
9755 * calls.c (compute_argument_addresses): Use simplify_gen_binary
9756 rather than choosing between plus_constant and gen_rtx_<CODE>.
9757 * expr.c (emit_push_insn): Likewise.
9758 (expand_expr_real_2): Likewise.
9760 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9761 Alan Hayward <alan.hayward@arm.com>
9762 David Sherwood <david.sherwood@arm.com>
9764 * loop-unroll.c (split_iv): Call copy_rtx on the step.
9766 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9767 Alan Hayward <alan.hayward@arm.com>
9768 David Sherwood <david.sherwood@arm.com>
9770 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
9771 calling tree_to_uhwi.
9773 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9774 Alan Hayward <alan.hayward@arm.com>
9775 David Sherwood <david.sherwood@arm.com>
9777 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
9778 INTEGER_CST rather than a negative test for ADDR_EXPR.
9780 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9781 Alan Hayward <alan.hayward@arm.com>
9782 David Sherwood <david.sherwood@arm.com>
9784 * tree-vrp.c (extract_range_from_binary_expr_1): Check
9785 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
9787 2017-09-21 Richard Biener <rguenther@suse.de>
9789 PR tree-optimization/71351
9790 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
9791 graphite_create_new_loop_guard): Remove, fold remaining parts
9793 (translate_isl_ast_node_for): ... here and simplify.
9795 2017-09-21 Jakub Jelinek <jakub@redhat.com>
9798 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
9799 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
9800 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
9801 alternative always use QI mode, for -Os imov (=R,R) alternative
9802 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
9805 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9806 Jeff Law <law@redhat.com>
9808 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
9809 (allocate_stack_space): New function, partially extracted from
9811 (s390_emit_prologue): Track offset to most recent stack probe.
9812 Code to allocate space moved into allocate_stack_space.
9813 Dump actions when no stack is allocated.
9814 (s390_prologue_plus_offset): New function.
9815 (s390_emit_stack_probe): Likewise.
9817 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
9819 * common.opt (Wa, Wl, Wp, g, gz=): Add
9821 (gno-column-info): Remove.
9822 (gcolumn-info): Drop RejectNegative.
9824 (gno-record-gcc-switches): Remove.
9825 (grecord-gcc-switches): Drop RejectNegative.
9826 (gno-split-dwarf): Remove.
9827 (gsplit-dwarf): Drop RejectNegative.
9828 (gno-strict-dwarf): Remove.
9829 (gstrict-dwarf): Drop RejectNegative.
9830 * config/darwin.opt (gfull, gused): Add RejectNegative.
9831 * dwarf2out.c (gen_producer_string): Drop
9832 gno-record-gcc-switches handler.
9833 * optc-gen.awk: Add g to prefixes with negative forms.
9834 * opts-common.c (remapping_prefix_p): New.
9835 (find_opt): Check it.
9836 (generate_canonical_option): Test g prefix.
9837 (option_map): Add -gno- mapping.
9838 (add_misspelling_candidates): Check remapping_prefix_p.
9840 2017-09-20 Jeff Law <law@redhat.com>
9842 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
9843 thinko in stack clash protection support.
9845 * explow.c (compute_stack_clash_protection_loop_data): Use
9846 CONST_INT_P instead of explicit test. Verify object is a
9847 CONST_INT_P before looking at INTVAL.
9848 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
9849 instead of explicit test.
9851 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
9854 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
9855 address instead of to r1 and r11.
9857 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
9859 * config.gcc: Support "knm".
9860 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
9861 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
9863 * config/i386/i386.c (m_KNM): Define.
9864 (processor_target_table): Add "knm".
9866 (ix86_option_override_internal): Add "knm".
9867 (ix86_issue_rate): Add PROCESSOR_KNM.
9868 (ix86_adjust_cost): Ditto.
9869 (ia32_multipass_dfa_lookahead): Ditto.
9870 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
9871 (fold_builtin_cpu): Add M_INTEL_KNM.
9872 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
9873 (processor_type): Add PROCESSOR_KNM.
9874 * config/i386/x86-tune.def: Add m_KNM.
9875 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
9877 2017-09-20 Richard Biener <rguenther@suse.de>
9879 PR tree-optimization/80213
9880 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
9881 are allowed in empty BBs as well.
9882 (canonicalize_loop_closed_ssa): Also look for other complex
9884 (scop_detection::get_sese): Include the loop-closed PHI block
9886 (scop_detection::merge_sese): Remove code adding extra blocks.
9887 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
9888 (build_scops): Assert the final returned scop is invalid.
9890 2017-09-20 Richard Biener <rguenther@suse.de>
9892 PR tree-optimization/82264
9893 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
9895 (vn_phi_lookup): Likewise.
9896 (vn_phi_insert): Likewise.
9898 2017-09-20 Jakub Jelinek <jakub@redhat.com>
9900 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
9901 that fits into uhwi or shwi, add DW_AT_const_value regardless
9902 of early_dwarf without going through RTL, using add_AT_unsigned
9905 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
9906 (DEBUG_LTO_ABBREV_SECTION): Likewise.
9907 (DEBUG_LTO_MACINFO_SECTION): Likewise.
9908 (DEBUG_MACRO_SECTION): Likewise.
9909 (DEBUG_LTO_MACRO_SECTION): Likewise.
9910 (DEBUG_STR_DWO_SECTION): Likewise.
9911 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
9912 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
9913 (DEBUG_LTO_DWO_LINE_SECTION): Define.
9914 (DEBUG_LTO_LINE_STR_SECTION): Define.
9915 (init_sections_and_labels): Initialize debug_line_str_section
9916 variable. Initialize debug_loc_section for -gdwarf-5 to
9917 DEBUG_LOCLISTS_SECTION. Formatting fixes.
9919 2017-09-20 Richard Biener <rguenther@suse.de>
9921 * graphite-sese-to-poly.c (extract_affine): Properly handle
9922 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
9924 2017-09-20 Richard Biener <rguenther@suse.de>
9926 PR tree-optimization/81373
9927 * graphite-scop-detection.c (build_cross_bb_scalars_def):
9928 Force SESE live-out defs to be handled even if they are
9931 2017-09-19 Jeff Law <law@redhat.com>
9933 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
9934 nothing for stack adjustments with REG_STACK_CHECK.
9935 * sched-deps.c (parse_add_or_inc): Reject insns with
9936 REG_STACK_CHECK from dependency breaking.
9937 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
9938 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
9939 * reg-notes.def (STACK_CHECK): New note.
9941 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
9942 (ix86_expand_prologue): Dump stack clash info as needed.
9943 Call ix86_adjust_stack_and_probe_stack_clash as needed.
9945 * function.c (dump_stack_clash_frame_info): New function.
9946 * function.h (dump_stack_clash_frame_info): Prototype.
9947 (enum stack_clash_probes): New enum.
9949 * config/alpha/alpha.c (alpha_expand_prologue): Also check
9950 flag_stack_clash_protection.
9951 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
9952 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
9953 (arm_frame_pointer_required): Likewise.
9954 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
9955 (ia64_expand_prologue): Likewise.
9956 * config/mips/mips.c (mips_expand_prologue): Likewise.
9957 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
9958 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
9959 (sparc_flat_expand_prologue): Likewise.
9960 * config/spu/spu.c (spu_expand_prologue): Likewise.
9962 * explow.c: Include "params.h".
9963 (anti_adjust_stack_and_probe_stack_clash): New function.
9964 (get_stack_check_protect): Likewise.
9965 (compute_stack_clash_protection_loop_data): Likewise.
9966 (emit_stack_clash_protection_loop_start): Likewise.
9967 (emit_stack_clash_protection_loop_end): Likewise.
9968 (allocate_dynamic_stack_space): Use get_stack_check_protect.
9969 Use anti_adjust_stack_and_probe_stack_clash.
9970 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
9971 (emit_stack_clash_protection_loop_start): Likewise.
9972 (emit_stack_clash_protection_loop_end): Likewise.
9973 * rtl.h (get_stack_check_protect): Prototype.
9974 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
9975 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
9976 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
9978 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
9980 * doc/tm.texi: Rebuilt.
9981 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
9982 get_stack_check_protect.
9983 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
9984 * config/arm/arm.c (arm_expand_prologue): Likewise.
9985 (arm_frame_pointer_required): Likewise.
9986 * config/i386/i386.c (ix86_expand_prologue): Likewise.
9987 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
9988 * config/mips/mips.c (mips_expand_prologue): Likewise.
9989 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
9990 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
9991 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
9992 (sparc_flat_expand_prologue): Likewise.
9994 * common.opt (-fstack-clash-protection): New option.
9995 * flag-types.h (enum stack_check_type): Note difference between
9996 -fstack-check= and -fstack-clash-protection.
9997 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
9998 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
9999 * toplev.c (process_options): Issue warnings/errors for cases
10000 not handled with -fstack-clash-protection.
10001 * doc/invoke.texi (-fstack-clash-protection): Document new option.
10002 (-fstack-check): Note additional problem with -fstack-check=generic.
10003 Note that -fstack-check is primarily for Ada and refer users
10004 to -fstack-clash-protection for stack-clash-protection.
10005 Document new params for stack clash protection.
10007 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
10009 * config/i386/i386.c (ix86_split_long_move): Do not handle
10010 address used for LEA in a special way.
10012 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
10014 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
10016 2017-09-19 Martin Sebor <msebor@redhat.com>
10019 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
10020 of incompatible types.
10022 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
10024 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10025 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
10026 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10027 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
10029 2017-09-19 Richard Biener <rguenther@suse.de>
10031 PR tree-optimization/82244
10032 * tree-vrp.c (remove_range_assertions): Do not propagate
10033 a constant to abnormals but replace the assert with a copy.
10035 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
10037 PR rtl-optimization/57878
10038 PR rtl-optimization/68988
10039 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
10040 avoidance test involving non_reload_pseudos. Move frequency test
10041 below the general fragmentation avoidance test.
10043 2017-09-19 Richard Biener <rguenther@suse.de>
10045 PR tree-optimization/69728
10046 * graphite-sese-to-poly.c (schedule_error): New global.
10047 (add_loop_schedule): Handle empty domain by failing the
10049 (build_original_schedule): Handle schedule_error.
10051 2017-09-19 Richard Biener <rguenther@suse.de>
10053 * graphite-scop-detection.c (scop_detection::can_represent_loop):
10054 Do not iterate to sibling loops but only to siblings of inner
10057 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
10060 * config/m68k/m68k.md (moveq feeding equality comparison): Check
10061 that the registers are different.
10063 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
10065 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
10066 to processor_model and "amdfam17h" to arch_names_table.
10067 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
10069 2017-09-18 Jakub Jelinek <jakub@redhat.com>
10072 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
10074 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10075 Alan Hayward <alan.hayward@arm.com>
10076 David Sherwood <david.sherwood@arm.com>
10078 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
10080 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
10082 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
10083 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
10084 vect_schedule_slp_instance.
10085 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
10086 Update call to vect_slp_analyze_node_operations. Simplify return
10088 (vect_slp_analyze_bb_1): Update call accordingly.
10089 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
10090 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
10091 (vect_schedule_slp): Update call accordingly.
10093 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10094 Alan Hayward <alan.hayward@arm.com>
10095 David Sherwood <david.sherwood@arm.com>
10097 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
10098 with types that aren't in fact scalar.
10100 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10102 * tree-vect-slp.c (vect_record_max_nunits): New function,
10104 (vect_build_slp_tree_1): ...here.
10105 (vect_build_slp_tree_2): Call it for phis too.
10107 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10109 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
10110 to vect_get_vec_def_for_operand when getting the mask operand.
10112 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10113 Alan Hayward <alan.hayward@arm.com>
10114 David Sherwood <david.sherwood@arm.com>
10116 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
10119 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10120 Alan Hayward <alan.hayward@arm.com>
10121 David Sherwood <david.sherwood@arm.com>
10123 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
10124 calculation for vector booleans.
10126 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10127 Alan Hayward <alan.hayward@arm.com>
10128 David Sherwood <david.sherwood@arm.com>
10130 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
10132 (vect_transform_stmt): ...here.
10133 (vect_analyze_stmt): Use it instead of calling
10134 vectorizable_live_operation directly.
10136 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
10138 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
10139 non-SIMT targets in acc vector loops.
10141 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
10143 * configure.ac: Add arc and check if assembler supports gdwarf2.
10144 * configure: Regenerate.
10146 2017-09-18 Richard Biener <rguenther@suse.de>
10148 PR tree-optimization/82220
10149 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
10150 epilogue niters from the min_profitable_iters compute.
10152 2017-09-18 Jakub Jelinek <jakub@redhat.com>
10155 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
10156 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
10157 (ix86_init_pic_reg): Revert 2017-09-01 changes.
10159 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
10162 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
10163 switching to a new text section.
10165 2017-09-18 Richard Biener <rguenther@suse.de>
10167 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
10169 (build_alias_set): Reject aliases with no access function.
10171 2017-09-18 Richard Biener <rguenther@suse.de>
10173 PR tree-optimization/79622
10174 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
10176 (build_cross_bb_scalars_use): Likewise.
10178 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
10180 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
10182 2017-09-18 Alan Modra <amodra@gmail.com>
10185 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
10186 stack_pointer_rtx for count 0. Update comments. Break up
10187 large rtl expression.
10189 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
10191 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
10192 Increase to 20 bytes.
10193 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
10194 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
10195 or avx version of the stub.
10197 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
10200 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
10201 compute the minimum stack alignment. Also update preferred stack
10202 boundary for leaf functions.
10204 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
10206 PR tree-optimization/82228
10207 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
10210 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
10212 * common/config/nds32/nds32-common.c
10213 (nds32_option_optimization_table): Refine formatting.
10214 (nds32_option_optimization_table): Use -fsched-pressure and
10215 -fomit-frame-pointer for specific optimization level.
10217 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
10219 * config/nds32/nds32.c: Refine formatting and comments.
10220 * config/nds32/nds32.h: Likewise.
10221 * config/nds32/nds32.md: Likewise.
10222 * config/nds32/nds32-cost.c: Likewise.
10223 * config/nds32/nds32-isr.c: Likewise.
10224 * config/nds32/nds32-md-auxiliary.c: Likewise.
10225 * config/nds32/nds32-multiple.md: Likewise.
10226 * config/nds32/nds32-predicates.c: Likewise.
10228 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
10229 Jakub Jelinek <jakub@redhat.com>
10231 Add support for -std=c++2a.
10232 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
10234 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
10236 2017-09-15 Steve Ellcey <sellcey@cavium.com>
10239 * doc/extend.texi (Common Function Attributes): Add
10240 references to ARM, AArch64, and S/390 specific attributes.
10241 (Function Specific Option Pragmas): Add AArch64 and S/390
10242 to list of back ends that support the target pragma.
10244 2017-09-15 Nathan Sidwell <nathan@acm.org>
10246 * doc/standards.texi: Fix C++17 description. Update URLs for
10249 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
10251 * common.opt (Wcast-align=strict): New warning option.
10252 * doc/invoke.texi: Document -Wcast-align=strict.
10254 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
10256 * cgraph.h (cgraph_thunk_info): Add comments.
10257 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
10258 assert for VIRTUAL_* arguments stricter.
10260 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
10262 PR tree-optimization/71026
10263 * match.pd: Move RDIV patterns from fold-const.c
10264 * fold-const.c (distribute_real_division): Removed.
10265 (fold_binary_loc): Remove calls to distribute_real_divison.
10267 2017-09-15 Jakub Jelinek <jakub@redhat.com>
10269 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
10270 c++1z and gnu++1z as deprecated. Change other references to
10271 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
10272 Change -Wc++1z-compat to -Wc++17-compat.
10273 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
10274 * dwarf2out.c (highest_c_language): Handle C++17.
10275 (gen_compile_unit_die): Likewise.
10277 2017-09-15 Jakub Jelinek <jakub@redhat.com>
10279 PR rtl-optimization/82192
10280 * combine.c (make_extraction): Don't look through non-paradoxical
10281 SUBREGs or TRUNCATE if pos + len is or might be bigger than
10284 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
10285 Alan Hayward <alan.hayward@arm.com>
10286 David Sherwood <david.sherwood@arm.com>
10288 * target.def (function_arg_offset): New hook.
10289 * targhooks.h (default_function_arg_offset): Declare.
10290 * targhooks.c (default_function_arg_offset): New function.
10291 * function.c (locate_and_pad_parm): Use
10292 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
10293 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
10294 (TARGET_FUNCTION_ARG_OFFSET): ...this.
10295 * doc/tm.texi: Regenerate.
10296 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
10297 * config/spu/spu.c (spu_function_arg_offset): New function.
10298 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
10299 * system.h (FUNCTION_ARG_OFFSET): Poison.
10301 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
10302 Alan Hayard <alan.hayward@arm.com>
10303 David Sherwood <david.sherwood@arm.com>
10305 * target.def (truly_noop_truncation): New hook.
10306 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
10307 than TRULY_NOOP_TRUNCATION.
10308 * hooks.h (hook_bool_uint_uint_true): Declare.
10309 * hooks.c (hook_bool_uint_uint_true): New function.
10310 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
10311 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
10312 * doc/tm.texi: Regenerate.
10313 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
10314 rather than TRULY_NOOP_TRUNCATION in comments.
10315 (simplify_comparison): Likewise.
10316 (record_truncated_value): Likewise.
10317 * expmed.c (extract_bit_field_1): Likewise.
10318 (extract_split_bit_field): Likewise.
10319 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
10320 instead of TRULY_NOOP_TRUNCATION.
10321 * function.c (assign_parm_setup_block): Likewise.
10322 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
10323 * rtlhooks.c: Include target.h.
10324 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
10325 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
10326 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
10327 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
10328 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
10329 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
10330 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
10331 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
10332 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
10333 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
10334 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
10335 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
10336 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
10337 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
10338 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
10339 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
10340 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
10341 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
10342 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
10343 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
10344 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
10345 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
10346 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
10347 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
10348 * config/mips/mips.c (mips_truly_noop_truncation): New function.
10349 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10350 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
10351 rather than TRULY_NOOP_TRUNCATION in comments.
10352 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
10353 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
10354 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
10355 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
10356 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
10357 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
10358 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
10359 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
10360 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
10361 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
10362 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
10363 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
10364 rather than TRULY_NOOP_TRUNCATION in comments.
10365 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
10366 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
10367 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
10368 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
10369 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
10370 TRULY_NOOP_TRUNCATION condition.
10371 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
10372 (TRULY_NOOP_TRUNCATION): Delete.
10373 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
10374 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
10375 * config/spu/spu.c (spu_truly_noop_truncation): New function.
10376 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10377 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
10378 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
10379 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
10380 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10381 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
10382 rather than TRULY_NOOP_TRUNCATION in comments.
10383 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
10384 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
10385 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
10386 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
10387 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
10388 * system.h (TRULY_NOOP_TRUNCATION): Poison.
10390 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
10393 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
10394 (*cmp_ior): Likewise.
10395 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
10396 (*ior_scc_scc_cmp): Likewise.
10397 (*and_scc_scc): Likewise.
10398 (*and_scc_scc_cmp): Likewise.
10400 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
10401 Alan Hayard <alan.hayward@arm.com>
10402 David Sherwood <david.sherwood@arm.com>
10404 * target.def (can_change_mode_class): New hook.
10405 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
10406 (hard_regno_nregs): Likewise.
10407 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
10408 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
10409 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
10410 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
10411 (LOAD_EXTEND_OP): Update accordingly.
10412 * doc/tm.texi: Regenerate.
10413 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
10414 CANNOT_CHANGE_MODE_CLASS.
10415 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
10416 (REG_CAN_CHANGE_MODE_P): ...this new macro.
10417 * combine.c (simplify_set): Update accordingly.
10418 * emit-rtl.c (validate_subreg): Likewise.
10419 * recog.c (general_operand): Likewise.
10420 * regcprop.c (mode_change_ok): Likewise.
10421 * reload1.c (choose_reload_regs): Likewise.
10422 (inherit_piecemeal_p): Likewise.
10423 * rtlanal.c (simplify_subreg_regno): Likewise.
10424 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
10425 instead of CANNOT_CHANGE_MODE_CLASS.
10426 (reload_cse_simplify_operands): Likewise.
10427 * reload.c (push_reload): Use targetm.can_change_mode_class
10428 instead of CANNOT_CHANGE_MODE_CLASS.
10429 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
10430 REG_CANNOT_CHANGE_MODE_P.
10431 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10432 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
10433 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10434 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10435 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10436 (arm_can_change_mode_class): New function.
10437 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
10438 than CANNOT_CHANGE_MODE_CLASS in comments.
10439 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10440 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
10441 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
10442 (ix86_can_change_mode_class): ...this new function, inverting the
10443 sense of the return value.
10444 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10445 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10446 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10447 (ia64_can_change_mode_class): New function.
10448 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10449 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
10450 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
10451 (m32c_can_change_mode_class): ...this new function, inverting the
10452 sense of the return value.
10453 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10454 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10455 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
10456 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
10457 (mips_can_change_mode_class): ...this new function, inverting the
10458 sense of the return value.
10459 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10460 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10461 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10462 (msp430_can_change_mode_class): New function.
10463 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10464 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
10465 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10466 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10467 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10468 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
10469 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10470 (pa_cannot_change_mode_class): Replace with...
10471 (pa_can_change_mode_class): ...this new function, inverting the
10472 sense of the return value.
10473 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
10474 than CANNOT_CHANGE_MODE_CLASS in comments.
10475 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10476 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
10477 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10478 (pdp11_cannot_change_mode_class): Replace with...
10479 (pdp11_can_change_mode_class): ...this new function, inverting the
10480 sense of the return value.
10481 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10482 * config/powerpcspe/powerpcspe-protos.h
10483 (rs6000_cannot_change_mode_class_ptr): Delete.
10484 * config/powerpcspe/powerpcspe.c
10485 (rs6000_cannot_change_mode_class_ptr): Delete.
10486 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10487 (rs6000_option_override_internal): Assign to
10488 targetm.can_change_mode_class instead of
10489 rs6000_cannot_change_mode_class_ptr.
10490 (rs6000_cannot_change_mode_class): Replace with...
10491 (rs6000_can_change_mode_class): ...this new function, inverting the
10492 sense of the return value.
10493 (rs6000_debug_cannot_change_mode_class): Replace with...
10494 (rs6000_debug_can_change_mode_class): ...this new function.
10495 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10496 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
10497 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10498 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10499 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
10501 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
10502 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10503 (rs6000_option_override_internal): Assign to
10504 targetm.can_change_mode_class instead of
10505 rs6000_cannot_change_mode_class_ptr.
10506 (rs6000_cannot_change_mode_class): Replace with...
10507 (rs6000_can_change_mode_class): ...this new function, inverting the
10508 sense of the return value.
10509 (rs6000_debug_cannot_change_mode_class): Replace with...
10510 (rs6000_debug_can_change_mode_class): ...this new function.
10511 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10512 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
10513 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
10514 (s390_can_change_mode_class): ...this new function, inverting the
10515 sense of the return value.
10516 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10517 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10518 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
10519 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10520 (sh_cannot_change_mode_class): Replace with...
10521 (sh_can_change_mode_class): ...this new function, inverting the
10522 sense of the return value.
10523 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10524 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10525 (sparc_can_change_mode_class): New function.
10526 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10527 * config/spu/spu.c (spu_can_change_mode_class): New function.
10528 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10529 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10530 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10531 (visium_can_change_mode_class): New function.
10532 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
10534 2017-09-15 Richard Biener <rguenther@suse.de>
10536 PR tree-optimization/82217
10537 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
10538 but not undefined case.
10540 2017-09-15 Jakub Jelinek <jakub@redhat.com>
10543 * postreload.c (reload_cse_simplify_operands): Skip
10544 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
10546 2017-09-15 Richard Biener <rguenther@suse.de>
10548 PR tree-optimization/68823
10549 * graphite-scop-detection.c (build_alias_set): If we have a
10550 possible dependence check whether we can handle them by just
10551 looking at the DRs DR_ACCESS_FNs.
10552 (build_scops): If build_alias_set fails, fail the SCOP.
10554 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
10556 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
10557 to support float128 built-in functions that require the ISA 3.0
10559 (BU_FLOAT128_3_HW): Likewise.
10560 (SQRTF128): Add support for the IEEE 128-bit square root and fma
10561 built-in functions.
10562 (FMAF128): Likewise.
10564 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
10565 support for built-in functions that need the ISA 3.0 IEEE 128-bit
10566 floating point instructions.
10567 (rs6000_invalid_builtin): Likewise.
10568 (rs6000_builtin_mask_names): Likewise.
10569 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
10570 (RS6000_BTM_FLOAT128_HW): Likewise.
10571 (RS6000_BTM_COMMON): Likewise.
10572 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
10574 * doc/extend.texi (RS/6000 built-in functions): Document the
10575 IEEE 128-bit floating point square root and fused multiply-add
10576 built-in functions.
10578 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
10580 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
10581 reg (r2) isn't in the set of registers defined in the prologue.
10583 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10584 Alan Hayward <alan.hayward@arm.com>
10585 David Sherwood <david.sherwood@arm.com>
10587 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
10588 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
10589 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
10590 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
10591 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
10593 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
10594 max_vectorization_factor.
10595 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
10597 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10598 Alan Hayward <alan.hayward@arm.com>
10599 David Sherwood <david.sherwood@arm.com>
10601 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
10602 (vect_worthwhile_without_simd_p): Declare.
10603 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
10604 (vectorizable_reduction): Use it.
10605 * tree-vect-stmts.c (vectorizable_shift): Likewise.
10606 (vectorizable_operation): Likewise.
10608 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10609 Alan Hayward <alan.hayward@arm.com>
10610 David Sherwood <david.sherwood@arm.com>
10612 * tree-vectorizer.h (vect_get_num_copies): New function.
10613 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
10614 * tree-vect-loop.c (vectorizable_reduction): Likewise.
10615 (vectorizable_induction): Likewise.
10616 (vectorizable_live_operation): Likewise.
10617 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
10618 (vectorizable_bswap): Likewise.
10619 (vectorizable_call): Likewise.
10620 (vectorizable_conversion): Likewise.
10621 (vectorizable_assignment): Likewise.
10622 (vectorizable_shift): Likewise.
10623 (vectorizable_operation): Likewise.
10624 (vectorizable_store): Likewise.
10625 (vectorizable_load): Likewise.
10626 (vectorizable_condition): Likewise.
10627 (vectorizable_comparison): Likewise.
10628 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
10630 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10631 Alan Hayward <alan.hayward@arm.com>
10632 David Sherwood <david.sherwood@arm.com>
10634 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
10635 of vect_init_vector.
10637 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10638 Alan Hayward <alan.hayward@arm.com>
10639 David Sherwood <david.sherwood@arm.com>
10641 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
10642 an inline wrapper that provides a location.
10643 (gimple_build_vector): Likewise.
10644 * gimple-fold.c (gimple_build_vector_from_val): New function.
10645 (gimple_build_vector): Likewise.
10646 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
10647 functions to build the initial value. Always return a gimple value.
10648 (get_initial_defs_for_reduction): Likewise. Only compute
10650 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
10651 vect_init_vector on the results from get_initial_def(s)_for_reduction.
10652 (vectorizable_induction): Use gimple_build_vector rather than
10655 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10656 Alan Hayward <alan.hayward@arm.com>
10657 David Sherwood <david.sherwood@arm.com>
10659 * target.h (vec_perm_indices): New typedef.
10660 (auto_vec_perm_indices): Likewise.
10661 * optabs-query.h: Include target.h
10662 (can_vec_perm_p): Take a vec_perm_indices *.
10663 * optabs-query.c (can_vec_perm_p): Likewise.
10664 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
10665 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
10666 * tree-vect-generic.c (lower_vec_perm): Likewise.
10667 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
10668 (vect_grouped_load_supported): Likewise.
10669 (vect_shift_permute_load_chain): Likewise.
10670 (vect_permute_store_chain): Use auto_vec_perm_indices.
10671 (vect_permute_load_chain): Likewise.
10672 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
10673 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
10674 Update uses of can_vec_perm_p.
10675 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
10676 mode with a number of elements. Take a vec_perm_indices *.
10677 (vect_create_epilog_for_reduction): Update accordingly.
10678 Use auto_vec_perm_indices.
10679 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
10680 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
10681 (vect_transform_slp_perm_load): Likewise.
10682 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
10683 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
10684 (vect_gen_perm_mask_checked): Likewise.
10685 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
10686 (vect_gen_perm_mask_checked): Likewise.
10687 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
10688 (vectorizable_store): Likewise.
10689 (vectorizable_load): Likewise.
10690 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
10691 (vectorizable_bswap): Likewise.
10693 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10694 Alan Hayward <alan.hayward@arm.com>
10695 David Sherwood <david.sherwood@arm.com>
10697 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
10698 * tree.c (build_vector): Likewise.
10699 (build_vector_from_ctor): Update accordingly.
10700 (build_vector_from_val): Likewise.
10701 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
10702 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
10703 * tree-vect-generic.c (add_rshift): Likewise.
10704 (expand_vector_divmod): Likewise.
10705 (optimize_vector_constructor): Likewise.
10706 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
10707 (vect_transform_slp_perm_load): Likewise.
10708 (vect_schedule_slp_instance): Likewise.
10709 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
10710 (vectorizable_call): Likewise.
10711 (vect_gen_perm_mask_any): Likewise. Add elements in order.
10712 * expmed.c (make_tree): Likewise.
10713 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
10714 a vector passed to build_vector.
10715 (fold_convert_const): Likewise.
10716 (exact_inverse): Likewise.
10717 (fold_ternary_loc): Likewise.
10718 (fold_relational_const): Likewise.
10719 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
10720 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
10721 (const_unop): Likewise. Store the reduction accumulator in a
10722 variable rather than an array.
10723 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
10724 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
10725 the new vector, rather than constructing it after the input arrays.
10726 (native_interpret_vector): Use auto_vec<tree> when building
10727 a vector passed to build_vector. Add elements in order.
10728 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
10729 auto_vec<tree> when building a vector passed to build_vector.
10730 (vect_create_epilog_for_reduction): Likewise.
10731 (vectorizable_induction): Likewise.
10732 (get_initial_def_for_reduction): Likewise. Fix indentation of
10734 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
10736 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
10737 passed to build_vector.
10739 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10740 Alan Hayward <alan.hayward@arm.com>
10741 David Sherwood <david.sherwood@arm.com>
10743 * tree-core.h (tree_base::u): Add an "nelts" field.
10744 (tree_vector): Use VECTOR_CST_NELTS as the length.
10745 * tree.c (tree_size): Likewise.
10746 (make_vector): Initialize VECTOR_CST_NELTS.
10747 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
10748 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
10749 TYPE_VECTOR_SUBPARTS.
10750 * expr.c (const_vector_mask_from_tree): Consistently use "units"
10751 as the number of units, setting it from VECTOR_CST_NELTS.
10752 (const_vector_from_tree): Likewise.
10753 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
10754 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
10755 (fold_negate_expr_1): Likewise.
10756 (fold_convert_const): Likewise.
10757 (const_binop): Likewise. Differentiate the number of output and
10759 (const_unop): Likewise.
10760 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
10761 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
10762 in cases that did the opposite.
10764 2017-09-14 Richard Biener <rguenther@suse.de>
10766 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
10769 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
10771 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
10773 2017-09-14 Jakub Jelinek <jakub@redhat.com>
10776 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
10777 if and where to split a bb, except for splitting before debug insn
10778 sequences followed by non-label real insn. Delete debug insns
10779 in between basic blocks.
10781 * combine.c (make_compound_operation_int): Formatting fixes.
10783 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
10784 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10785 * config/netbsd.h (LINK_EH_SPEC): Likewise.
10786 * config/sol2.h (LINK_EH_SPEC): Likewise.
10787 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10788 * config/s390/linux.h (LINK_SPEC): Likewise.
10789 * config/freebsd.h (LINK_EH_SPEC): Likewise.
10790 * config/openbsd.h (LINK_EH_SPEC): Likewise.
10791 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10792 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
10793 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
10794 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10795 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
10796 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
10798 2017-09-13 Jakub Jelinek <jakub@redhat.com>
10800 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
10802 (ENDFILE_LINUX_SPEC): Likewise.
10803 (LINK_EH_SPEC): Likewise.
10804 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
10805 (LINK_OS_LINUX_SPEC32): Likewise.
10806 (LINK_OS_LINUX_SPEC64): Likewise.
10807 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
10808 (LINK_OS_LINUX_SPEC): Likewise.
10810 2017-09-13 Martin Liska <mliska@suse.cz>
10812 PR middle-end/82154
10813 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
10814 CASE_HIGH is NULL_TREE.
10816 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
10817 Alan Hayward <alan.hayward@arm.com>
10818 David Sherwood <david.sherwood@arm.com>
10820 * target.def (secondary_memory_needed): New hook.
10821 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
10822 instead of SECONDARY_MEMORY_NEEDED.
10823 (secondary_memory_needed_mode): Likewise.
10824 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
10825 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
10826 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
10827 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
10828 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
10829 * doc/tm.texi: Regenerate.
10830 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
10831 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
10832 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10833 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
10834 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
10835 * config/i386/i386.c (inline_secondary_memory_needed): Put the
10836 mode argument first and change the reg_class arguments to reg_class_t.
10837 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
10838 Make static. Update the call to inline_secondary_memory_needed.
10839 (ix86_register_move_cost): Update the call to
10840 inline_secondary_memory_needed.
10841 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10842 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
10844 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
10845 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
10847 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
10848 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
10849 * config/mips/mips.c (mips_secondary_memory_needed): Make static
10850 and match hook interface. Add comment from mips.h.
10851 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10852 * config/mmix/mmix.md (truncdfsf2): Refer to
10853 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
10855 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
10856 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
10857 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10858 (pa_secondary_memory_needed): New function.
10859 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
10860 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
10861 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10862 (pdp11_secondary_memory_needed): Make static and match hook interface.
10863 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
10864 * config/powerpcspe/powerpcspe-protos.h
10865 (rs6000_secondary_memory_needed_ptr): Delete.
10866 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
10868 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10869 (rs6000_option_override_internal): Assign to
10870 targetm.secondary_memory_needed rather than
10871 rs6000_secondary_memory_needed_ptr.
10872 (rs6000_secondary_memory_needed): Match hook interface.
10873 (rs6000_debug_secondary_memory_needed): Likewise.
10874 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
10875 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
10876 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
10877 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10878 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
10879 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
10881 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
10882 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10883 (rs6000_option_override_internal): Assign to
10884 targetm.secondary_memory_needed rather than
10885 rs6000_secondary_memory_needed_ptr.
10886 (rs6000_secondary_memory_needed): Match hook interface.
10887 (rs6000_debug_secondary_memory_needed): Likewise.
10888 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
10889 * config/s390/s390.c (s390_secondary_memory_needed): New function.
10890 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10891 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
10892 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10893 (sparc_secondary_memory_needed): New function.
10894 * lra-constraints.c (check_and_process_move): Refer to
10895 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
10897 (curr_insn_transform): Likewise.
10898 (process_alt_operands): Use targetm.secondary_memory_needed
10899 instead of TARGET_SECONDARY_MEMORY_NEEDED.
10900 (check_secondary_memory_needed_p): Likewise.
10901 (choose_split_class): Likewise.
10902 * reload.c: Unconditionally include code that was previously
10903 conditional on SECONDARY_MEMORY_NEEDED.
10904 (push_secondary_reload): Use targetm.secondary_memory_needed
10905 instead of TARGET_SECONDARY_MEMORY_NEEDED.
10906 (push_reload): Likewise.
10907 * reload1.c: Unconditionally include code that was previously
10908 conditional on SECONDARY_MEMORY_NEEDED.
10909 (choose_reload_regs): Use targetm.secondary_memory_needed
10910 instead of TARGET_SECONDARY_MEMORY_NEEDED.
10911 (gen_reload): Likewise.
10912 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
10914 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
10915 Alan Hayward <alan.hayward@arm.com>
10916 David Sherwood <david.sherwood@arm.com>
10918 * target.def (secondary_memory_needed_mode): New hook:
10919 * targhooks.c (default_secondary_memory_needed_mode): Declare.
10920 * targhooks.h (default_secondary_memory_needed_mode): New function.
10921 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
10922 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
10923 * doc/tm.texi: Regenerate.
10924 * lra-constraints.c (check_and_process_move): Use
10925 targetm.secondary_memory_needed_mode instead of
10926 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
10927 (curr_insn_transform): Likewise.
10928 * reload.c (get_secondary_mem): Likewise.
10929 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10930 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
10932 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10933 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10934 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
10935 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10936 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
10938 * config/powerpcspe/powerpcspe-protos.h
10939 (rs6000_secondary_memory_needed_mode): Delete.
10940 * config/powerpcspe/powerpcspe.c
10941 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10942 (rs6000_secondary_memory_needed_mode): Make static.
10943 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10944 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
10946 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
10948 (rs6000_secondary_memory_needed_mode): Make static.
10949 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10950 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
10951 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10952 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10953 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
10955 (sparc_secondary_memory_needed_mode): New function.
10956 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
10958 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
10960 * config/aarch64/constraints.md (Umq): New constraint.
10961 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
10963 (mov<mode>): Update condition.
10965 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10967 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
10968 when bitposition is the same.
10970 2017-09-13 Richard Biener <rguenther@suse.de>
10972 * dwarf2out.c (output_die_symbol): Remove.
10973 (output_die): Do not output a DIEs symbol.
10975 2017-09-13 Richard Biener <rguenther@suse.de>
10977 PR middle-end/82128
10978 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
10979 default-def to avoid breaking iterator update with the weird
10980 interaction with cgraph_update_edges_for_call_stmt_node.
10982 2017-09-13 Richard Biener <rguenther@suse.de>
10984 * tree-cfg.c (verify_gimple_assign_binary): Add verification
10985 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
10986 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
10987 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
10989 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
10991 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
10992 Disable pc relative literal load irrespective of
10993 TARGET_FIX_ERR_A53_84341 for default.
10995 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
10997 * config/sparc/sparc.c (output_return): Output the source location of
10998 the insn in the delay slot, if any.
10999 (output_sibcall): Likewise.
11001 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
11004 * common.opt (-static-pie): New alias.
11005 (shared): Negate static-pie.
11006 (-no-pie): Update help text.
11008 (static-pie): New option.
11009 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
11010 -static-pie support.
11011 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
11012 (LINK_EH_SPEC): Likewise.
11013 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11014 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
11015 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
11016 * gcc.c (LINK_COMMAND_SPEC): Likewise.
11017 (init_gcc_specs): Likewise.
11018 (init_spec): Likewise.
11019 (display_help): Update help message for -pie.
11020 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
11023 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
11025 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
11026 (movdi_aarch64): Likewise.
11027 (movti_aarch64): Likewise.
11029 2017-09-12 Simon Wright <simon@pushface.org>
11032 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
11033 calculation of the minor version, always output as 0.
11035 2017-09-12 Jakub Jelinek <jakub@redhat.com>
11038 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
11039 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
11040 on it early, rather than manual conversion late. For
11041 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
11042 instead of performing manual conversion.
11044 2017-09-12 Carl Love <cel@us.ibm.com>
11046 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
11047 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
11049 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11050 VMULOSW): Add definitions.
11051 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11052 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11053 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11054 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11055 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11057 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
11059 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
11061 (movti_aarch64): Likewise.
11062 (movdf_aarch64): Likewise.
11063 (movtf_aarch64): Likewise.
11064 (load_pairdi): Likewise.
11065 (store_pairdi): Likewise.
11066 (load_pairdf): Likewise.
11067 (store_pairdf): Likewise.
11068 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
11069 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
11070 (ldr_got_small_<mode>): Likewise.
11071 (ldr_got_small_28k_<mode>): Likewise.
11072 (ldr_got_tiny): Likewise.
11073 * config/aarch64/iterators.md (ldst_sz): New.
11074 (ldpstp_sz): Likewise.
11075 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
11077 (thunderx_load): Split load_8 to load_16.
11078 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
11080 (thunderx2t99_storepair_basic): Split store_8 to store_16.
11081 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
11082 (xgene1_store_pair): Split store_8 to store_16.
11083 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
11084 (falkor_st_0_st_sd): Split store_8 to store_16.
11086 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
11088 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
11089 and store1/2/3/4 to store_4/8/12/16.
11090 * config/aarch64/aarch64.md: Update for rename.
11091 * config/arm/arm.md: Likewise.: Likewise.
11092 * config/arm/arm.c: Likewise.
11093 * config/arm/thumb1.md: Likewise.
11094 * config/arm/thumb2.md: Likewise.
11095 * config/arm/vfp.md: Likewise.
11096 * config/arm/arm-generic.md: Likewise.
11097 * config/arm/arm1020e.md: Likewise.
11098 * config/arm/arm1026ejs.md: Likewise.
11099 * config/arm/arm1136jfs.md: Likewise.
11100 * config/arm/arm926ejs.md: Likewise.
11101 * config/arm/cortex-a15.md: Likewise.
11102 * config/arm/cortex-a17.md: Likewise.
11103 * config/arm/cortex-a5.md: Likewise.
11104 * config/arm/cortex-a53.md: Likewise.
11105 * config/arm/cortex-a57.md: Likewise.
11106 * config/arm/cortex-a7.md: Likewise.
11107 * config/arm/cortex-a8.md: Likewise.
11108 * config/arm/cortex-a9.md: Likewise.
11109 * config/arm/cortex-m4.md: Likewise.
11110 * config/arm/cortex-m7.md: Likewise.
11111 * config/arm/cortex-r4.md: Likewise.
11112 * config/arm/exynos-m1.md: Likewise.
11113 * config/arm/fa526.md: Likewise.
11114 * config/arm/fa606te.md: Likewise.
11115 * config/arm/fa626te.md: Likewise.
11116 * config/arm/fa726te.md: Likewise.
11117 * config/arm/fmp626.md: Likewise.
11118 * config/arm/iwmmxt.md: Likewise.
11119 * config/arm/ldmstm.md: Likewise.
11120 * config/arm/marvell-pj4.md: Likewise.
11121 * config/arm/xgene1.md: Likewise.
11122 * config/aarch64/thunderx.md: Likewise.
11123 * config/aarch64/thunderx2t99.md: Likewise.
11124 * config/aarch64/falkor.md: Likewise.
11126 2017-09-12 Martin Liska <mliska@suse.cz>
11128 * attribs.c (private_lookup_attribute): New function.
11129 * attribs.h (private_lookup_attribute): Declared here.
11130 (lookup_attribute): Called from this place.
11132 2017-09-12 Richard Biener <rguenther@suse.de>
11134 PR tree-optimization/82157
11135 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
11136 stmts with side-effects.
11138 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11139 Alan Hayward <alan.hayward@arm.com>
11140 David Sherwood <david.sherwood@arm.com>
11142 * target.def (hard_regno_nregs): New hook.
11143 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
11144 * targhooks.h (default_hard_regno_nregs): Declare.
11145 * targhooks.c (default_hard_regno_nregs): New function.
11146 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
11147 (TARGET_HARD_REGNO_NREGS): ...this hook.
11148 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
11149 (CLASS_MAX_NREGS): Likewise.
11150 * doc/tm.texi: Regenerate.
11151 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
11152 instead of HARD_REGNO_NREGS.
11153 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
11154 HARD_REGNO_NREGS in the comment.
11155 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
11156 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
11157 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
11158 Return an unsigned int.
11159 (TARGET_HARD_REGNO_NREGS): Redefine.
11160 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
11161 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
11162 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
11163 (arc_hard_regno_nregs): New function.
11164 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
11165 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
11166 (arm_hard_regno_nregs): New function.
11167 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
11168 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
11169 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
11170 (TARGET_HARD_REGNO_NREGS): Redefine.
11171 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
11172 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
11173 (HARD_REGNO_NREGS): Delete.
11174 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
11175 (cr16_hard_regno_nregs): New function.
11176 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
11177 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
11178 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
11179 (cris_hard_regno_nregs): New function.
11180 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
11181 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
11182 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
11183 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
11184 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
11185 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
11186 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
11187 (frv_hard_regno_nregs): Make static. Take and return an
11189 (frv_class_max_nregs): Remove outdated copy of documentation.
11190 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
11191 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
11192 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
11193 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
11194 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
11195 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
11196 (TARGET_HARD_REGNO_NREGS): Redefine.
11197 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
11198 (CLASS_MAX_NREGS): Update comment.
11199 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
11200 (ia64_hard_regno_nregs): New function.
11201 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
11202 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
11203 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
11204 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
11205 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
11207 (m32c_hard_regno_nregs): Likewise. Make static.
11208 (TARGET_HARD_REGNO_NREGS): Redefine.
11209 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
11210 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
11211 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
11212 (m68k_hard_regno_nregs): New function.
11213 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
11214 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
11215 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
11216 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
11217 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
11218 Take and return an unsigned int.
11219 (TARGET_HARD_REGNO_NREGS): Redefine.
11220 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
11221 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
11222 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
11223 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
11224 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
11225 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
11226 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
11227 (msp430_hard_regno_nregs): Make static. Take and return an
11229 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
11230 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
11231 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
11232 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
11233 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
11234 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
11235 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
11236 (TARGET_HARD_REGNO_NREGS): Redefine.
11237 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
11238 (PA_HARD_REGNO_NREGS): ...this.
11239 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
11240 (PA_HARD_REGNO_NREGS): ...this.
11241 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
11242 (pa_hard_regno_nregs): New function.
11243 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
11244 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
11245 (pdp11_hard_regno_nregs): New function.
11246 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
11247 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
11248 (rs6000_hard_regno_nregs_hook): New function.
11249 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
11250 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
11251 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
11252 Take and return an unsigned int. Move earlier in file.
11253 (TARGET_HARD_REGNO_NREGS): Redefine.
11254 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
11255 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
11256 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
11257 (rl78_hard_regno_nregs): Make static. Take and return an
11259 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
11260 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
11261 (rs6000_hard_regno_nregs_hook): New function.
11262 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
11263 * config/rx/rx.c (rx_hard_regno_nregs): New function.
11264 (TARGET_HARD_REGNO_NREGS): Redefine.
11265 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
11266 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
11267 instead of HARD_REGNO_NREGS.
11268 (s390_hard_regno_nregs): New function.
11269 (s390_hard_regno_mode_ok): Add comment from s390.h.
11270 (TARGET_HARD_REGNO_NREGS): Redefine.
11271 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
11272 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
11273 (sh_hard_regno_nregs): New function.
11274 (sh_pass_in_reg_p): Use it.
11275 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
11276 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
11277 (sparc_hard_regno_nregs): New function.
11278 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
11279 * config/spu/spu.c (spu_hard_regno_nregs): New function.
11280 (spu_function_arg_advance): Use it, supplying a valid register number.
11281 (TARGET_HARD_REGNO_NREGS): Redefine.
11282 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
11283 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
11284 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
11285 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
11286 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
11287 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
11288 (CLASS_MAX_NREGS): Remove copy of old documentation.
11289 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
11290 (visium_hard_regno_nregs): New function.
11291 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
11292 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
11293 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
11294 xtensa_hard_regno_nregs): New function.
11295 * system.h (HARD_REGNO_NREGS): Poison.
11297 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11299 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
11300 hard_regno_nregs instead of HARD_REGNO_NREGS.
11301 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
11302 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
11303 (c6x_expand_epilogue): Likewise.
11304 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
11305 (frv_read_iacc_argument): Likewise.
11306 * config/sh/sh.c: Include regs.h.
11307 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
11308 (regs_used): Likewise.
11309 (output_stack_adjust): Likewise.
11310 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
11311 * expmed.c: Include regs.h.
11312 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
11313 * ree.c: Include regs.h.
11314 (combine_reaching_defs): Use hard_regno_nregs instead of
11316 (add_removable_extension): Likewise.
11318 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11320 * regs.h (hard_regno_nregs): Turn into a function.
11321 (end_hard_regno): Update accordingly.
11322 * caller-save.c (setup_save_areas): Likewise.
11323 (save_call_clobbered_regs): Likewise.
11324 (replace_reg_with_saved_mem): Likewise.
11325 (insert_restore): Likewise.
11326 (insert_save): Likewise.
11327 * combine.c (can_change_dest_mode): Likewise.
11328 (move_deaths): Likewise.
11329 (distribute_notes): Likewise.
11330 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
11331 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
11332 (rs6000_split_multireg_move): Likewise.
11333 (rs6000_register_move_cost): Likewise.
11334 (rs6000_memory_move_cost): Likewise.
11335 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
11336 (rs6000_split_multireg_move): Likewise.
11337 (rs6000_register_move_cost): Likewise.
11338 (rs6000_memory_move_cost): Likewise.
11339 * cselib.c (cselib_reset_table): Likewise.
11340 (cselib_lookup_1): Likewise.
11341 * emit-rtl.c (set_mode_and_regno): Likewise.
11342 * function.c (aggregate_value_p): Likewise.
11343 * ira-color.c (setup_profitable_hard_regs): Likewise.
11344 (check_hard_reg_p): Likewise.
11345 (calculate_saved_nregs): Likewise.
11346 (assign_hard_reg): Likewise.
11347 (improve_allocation): Likewise.
11348 (calculate_spill_cost): Likewise.
11349 * ira-emit.c (modify_move_list): Likewise.
11350 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
11351 (ira_hard_reg_in_set_p): Likewise.
11352 * ira.c (setup_reg_mode_hard_regset): Likewise.
11353 (clarify_prohibited_class_mode_regs): Likewise.
11354 (check_allocation): Likewise.
11355 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11356 (lra_setup_reg_renumber): Likewise.
11357 (setup_try_hard_regno_pseudos): Likewise.
11358 (spill_for): Likewise.
11359 (assign_hard_regno): Likewise.
11360 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
11361 * lra-constraints.c (in_class_p): Likewise.
11362 (lra_constraint_offset): Likewise.
11363 (simplify_operand_subreg): Likewise.
11364 (lra_constraints): Likewise.
11365 (split_reg): Likewise.
11366 (split_if_necessary): Likewise.
11367 (invariant_p): Likewise.
11368 (inherit_in_ebb): Likewise.
11369 * lra-lives.c (process_bb_lives): Likewise.
11370 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
11371 (get_hard_regs): Likewise.
11372 (do_remat): Likewise.
11373 * lra-spills.c (assign_spill_hard_regs): Likewise.
11374 * mode-switching.c (create_pre_exit): Likewise.
11375 * postreload.c (reload_combine_recognize_pattern): Likewise.
11376 * recog.c (peep2_find_free_register): Likewise.
11377 * regcprop.c (kill_value_regno): Likewise.
11378 (set_value_regno): Likewise.
11379 (copy_value): Likewise.
11380 (maybe_mode_change): Likewise.
11381 (find_oldest_value_reg): Likewise.
11382 (copyprop_hardreg_forward_1): Likewise.
11383 * regrename.c (check_new_reg_p): Likewise.
11384 (regrename_do_replace): Likewise.
11385 * reload.c (push_reload): Likewise.
11386 (combine_reloads): Likewise.
11387 (find_dummy_reload): Likewise.
11388 (operands_match_p): Likewise.
11389 (find_reloads): Likewise.
11390 (find_equiv_reg): Likewise.
11391 (reload_adjust_reg_for_mode): Likewise.
11392 * reload1.c (count_pseudo): Likewise.
11393 (count_spilled_pseudo): Likewise.
11394 (find_reg): Likewise.
11395 (clear_reload_reg_in_use): Likewise.
11396 (free_for_value_p): Likewise.
11397 (allocate_reload_reg): Likewise.
11398 (choose_reload_regs): Likewise.
11399 (reload_adjust_reg_for_temp): Likewise.
11400 (emit_reload_insns): Likewise.
11401 (delete_output_reload): Likewise.
11402 * rtlanal.c (subreg_get_info): Likewise.
11403 * sched-deps.c (sched_analyze_reg): Likewise.
11404 * sel-sched.c (init_regs_for_mode): Likewise.
11405 (mark_unavailable_hard_regs): Likewise.
11406 (choose_best_reg_1): Likewise.
11407 (verify_target_availability): Likewise.
11408 * valtrack.c (dead_debug_insert_temp): Likewise.
11409 * var-tracking.c (track_loc_p): Likewise.
11410 (emit_note_insn_var_location): Likewise.
11411 * varasm.c (make_decl_rtl): Likewise.
11412 * reginfo.c (choose_hard_reg_mode): Likewise.
11413 (init_reg_modes_target): Refer directly to
11414 this_target_regs->x_hard_regno_nregs.
11416 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11418 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
11419 instead of hard_regno_nregs.
11421 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11423 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
11424 end_hard_regno instead of hard_regno_nregs.
11425 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
11426 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
11427 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
11428 * ira-color.c (improve_allocation): Likewise.
11429 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11430 * lra-lives.c (mark_regno_live): Likewise.
11431 (mark_regno_dead): Likewise.
11432 * lra-remat.c (operand_to_remat): Likewise.
11433 * lra.c (collect_non_operand_hard_regs): Likewise.
11434 * postreload.c (reload_combine_note_store): Likewise.
11435 (move2add_valid_value_p): Likewise.
11436 * reload.c (regno_clobbered_p): Likewise.
11438 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11440 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
11442 * config/v850/v850.c (v850_reorg): Likewise.
11443 * reload.c (refers_to_regno_for_reload_p): Likewise.
11444 (find_equiv_reg): Likewise.
11445 * reload1.c (reload_reg_reaches_end_p): Likewise.
11447 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11449 * caller-save.c (add_used_regs): Use REG_NREGS instead of
11451 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
11452 * config/arm/arm.c (output_move_neon): Likewise.
11453 (arm_attr_length_move_neon): Likewise.
11454 (neon_split_vcombine): Likewise.
11455 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
11456 (c6x_mark_reg_written): Likewise.
11457 (c6x_dwarf_register_span): Likewise.
11458 * config/i386/i386.c (ix86_save_reg): Likewise.
11459 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
11460 (rws_access_reg): Likewise.
11461 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
11462 * mode-switching.c (create_pre_exit): Likewise.
11463 * ree.c (combine_reaching_defs): Likewise.
11464 (add_removable_extension): Likewise.
11465 * regcprop.c (find_oldest_value_reg): Likewise.
11466 (copyprop_hardreg_forward_1): Likewise.
11467 * reload.c (reload_inner_reg_of_subreg): Likewise.
11468 (push_reload): Likewise.
11469 (combine_reloads): Likewise.
11470 (find_dummy_reload): Likewise.
11471 (reload_adjust_reg_for_mode): Likewise.
11472 * reload1.c (find_reload_regs): Likewise.
11473 (forget_old_reloads_1): Likewise.
11474 (reload_reg_free_for_value_p): Likewise.
11475 (reload_adjust_reg_for_temp): Likewise.
11476 (emit_reload_insns): Likewise.
11477 (delete_output_reload): Likewise.
11478 * sel-sched.c (choose_best_reg_1): Likewise.
11479 (choose_best_pseudo_reg): Likewise.
11481 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11482 Alan Hayward <alan.hayward@arm.com>
11483 David Sherwood <david.sherwood@arm.com>
11485 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
11486 * target.def (slow_unaligned_access): New hook.
11487 * targhooks.h (default_slow_unaligned_access): Declare.
11488 * targhooks.c (default_slow_unaligned_access): New function.
11489 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
11490 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
11491 * doc/tm.texi: Regenerate.
11492 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
11493 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
11494 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
11496 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
11497 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
11499 (rs6000_slow_unaligned_access): New function.
11500 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
11501 (expand_block_compare): Likewise.
11502 (expand_strn_compare): Likewise.
11503 (rs6000_rtx_costs): Likewise.
11504 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
11505 (riscv_slow_unaligned_access): Likewise.
11506 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
11507 (riscv_slow_unaligned_access_p): ...this and make static.
11508 (riscv_option_override): Update accordingly.
11509 (riscv_slow_unaligned_access): New function.
11510 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
11511 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
11512 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
11513 (rs6000_slow_unaligned_access): New function.
11514 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
11515 (rs6000_rtx_costs): Likewise.
11516 * config/rs6000/rs6000-string.c (expand_block_compare)
11517 (expand_strn_compare): Use targetm.slow_unaligned_access instead
11518 of SLOW_UNALIGNED_ACCESS.
11519 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
11520 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
11521 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
11522 of SLOW_UNALIGNED_ACCESS.
11523 * expmed.c (simple_mem_bitfield_p): Likewise.
11524 * expr.c (alignment_for_piecewise_move): Likewise.
11525 (emit_group_load_1): Likewise.
11526 (emit_group_store): Likewise.
11527 (copy_blkmode_from_reg): Likewise.
11528 (emit_push_insn): Likewise.
11529 (expand_assignment): Likewise.
11530 (store_field): Likewise.
11531 (expand_expr_real_1): Likewise.
11532 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
11533 * lra-constraints.c (simplify_operand_subreg): Likewise.
11534 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11535 * gimple-ssa-store-merging.c: Likewise in block comment at start
11537 * tree-ssa-strlen.c: Include target.h.
11538 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
11539 of SLOW_UNALIGNED_ACCESS.
11540 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
11542 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11544 PR rtl-optimization/82185
11545 * expmed.c (emit_store_flag_int): Only test tem if it has been
11548 2017-09-12 Richard Biener <rguenther@suse.de>
11550 PR middle-end/82149
11551 * match.pd ((FTYPE) N CMP CST): Fix typo.
11553 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
11555 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
11557 (mips_near_type_p): Add 'short_call' attribute as a synonym
11559 * doc/extend.texi (short_call): Document new function attribute.
11561 2017-09-12 Jakub Jelinek <jakub@redhat.com>
11564 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
11565 assertion check that in the condition.
11566 (get_atomic_generic_size): Likewise. Before testing if parameter
11567 has pointer type, if it has array type, call for C++
11568 default_conversion to perform array-to-pointer conversion.
11570 2017-09-12 Richard Biener <rguenther@suse.de>
11572 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
11573 for operations we cannot scalarize.
11575 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
11577 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
11578 vectors heap vectors. Clean up comments.
11579 Make visited_bbs a reference.
11580 (profitable_jump_thread_path): Make GC
11581 vectors heap vectors. Clean up comments.
11583 (convert_and_register_jump_thread_path): Make GC vectors heap
11585 (check_subpath_and_update_thread_path): Same. Clean up comments.
11586 Make visited_bbs a reference.
11587 (handle_phi): Abstract common code to to
11588 register_jump_thread_path_if_profitable.
11589 Rename VAR_BB to DEF_BB.
11591 Make GC vectors heap vectors.
11592 Make visited_bbs a reference.
11593 (handle_assignment): Same.
11594 (register_jump_thread_path_if_profitable): New.
11595 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
11597 Make GC vectors heap vectors. Clean up comments.
11598 Make visited_bbs a reference.
11599 (find_jump_threads_backwards): Make visited_bbs live in the stack.
11600 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
11603 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
11606 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
11607 words of E_DImode object are reachable by xtensa_uimm8x4 access.
11609 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
11611 Revert r251800 and r251799.
11613 2017-09-11 Martin Jambor <mjambor@suse.cz>
11616 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
11617 arguments in advance.
11618 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
11619 use it to find predecessor edges.
11620 (naive_outof_ssa): Collect vector of predecessors.
11622 2017-09-08 Jason Merrill <jason@redhat.com>
11624 PR c++/70029 - ICE with ref-qualifier and -flto
11625 * langhooks.h (struct lang_hooks_for_types): Add
11626 copy_lang_qualifiers.
11627 * attribs.c (build_type_attribute_qual_variant): Use it.
11628 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
11630 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
11631 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
11633 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
11636 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
11637 (*mulsi3_sp64): New instruction.
11638 (mulsi3): New expander.
11640 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
11642 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
11644 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11646 * sancov.c: Include memmodel.h.
11648 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
11651 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
11654 2017-09-07 Carl Love <cel@us.ibm.com>
11656 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
11657 the sldi instruction.
11659 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
11661 * sancov.c: Include tm_p.h.
11663 2017-09-07 Jakub Jelinek <jakub@redhat.com>
11666 * output.h (switch_to_other_text_partition): New declaration.
11667 * varasm.c (switch_to_other_text_partition): New function.
11668 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
11669 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
11670 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
11671 to the other text partition before emitting LCL label and switch back
11672 after emitting the word after it.
11674 2017-09-07 Richard Biener <rguenther@suse.de>
11676 * passes.def (pass_split_crit_edges): Remove instance before PRE.
11677 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
11678 critical edges here, after loop init.
11679 (pass_data_pre): Remove PROP_no_crit_edges flags.
11680 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
11681 for valueization of call args to avoid leaking VN_TOP.
11682 (visit_use): Assert we do not visit default defs.
11683 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
11684 Use error_mark_node to more easily detect leaking VN_TOP.
11685 All default-defs are varying, not VN_TOP. Mark them visited.
11686 (run_scc_vn): Make code match comment.
11688 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11690 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
11691 OPTION_MASK_FLOAT128_KEYWORD.
11692 (POWERPC_MASKS): Likewise.
11693 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
11694 support for the -mfloat128-type option, and make -mfloat128
11695 default on PowerPC Linux systems. Define or undefine
11696 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
11697 Define __float128 to be __ieee128 if IEEE 128-bit support is
11698 enabled, or undefine it.
11699 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
11700 Delete defining __FLOAT128_TYPE__.
11701 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
11702 -mfloat128-type option and make -mfloat128 default on PowerPC
11704 (TARGET_FLOAT128_TYPE): Likewise.
11705 (-mfloat128-type): Likewise.
11706 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11707 Delete the -mfloat128-type option and make -mfloat128 default on
11708 PowerPC Linux systems. Always use __ieee128 to be the keyword for
11709 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
11710 128-bit floating point is enabled. Change tests from using
11711 -mfloat128-type to -mfloat128.
11712 (rs6000_mangle_type): Use the correct mangling for the __float128
11713 type even if normal long double is restricted to 64-bits.
11714 (floatn_mode): Enable the _Float128 type by default on VSX Linux
11716 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
11717 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
11718 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
11720 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
11721 documentation for -mfloat128.
11723 2017-09-06 Olivier Hainque <hainque@adacore.com>
11725 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
11727 2017-09-06 Wish Wu <wishwu007@gmail.com>
11728 Jakub Jelinek <jakub@redhat.com>
11730 * asan.c (initialize_sanitizer_builtins): Add
11731 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
11732 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
11733 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
11734 BT_FN_VOID_UINT64_PTR variables.
11735 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
11736 (BT_FN_VOID_UINT16_UINT16): Likewise.
11737 (BT_FN_VOID_UINT32_UINT32): Likewise.
11738 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
11739 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
11740 (BT_FN_VOID_UINT64_PTR): Likewise.
11741 * common.opt (flag_sanitize_coverage): New variable.
11742 (fsanitize-coverage=trace-pc): Remove.
11743 (fsanitize-coverage=): Add.
11744 * flag-types.h (enum sanitize_coverage_code): New enum.
11745 * fold-const.c (fold_range_test): Disable non-short-circuit
11746 optimization if flag_sanitize_coverage.
11747 (fold_truth_andor): Likewise.
11748 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
11749 * opts.c (COVERAGE_SANITIZER_OPT): Define.
11750 (coverage_sanitizer_opts): New array.
11751 (get_closest_sanitizer_option): Add OPTS argument, handle also
11752 OPT_fsanitize_coverage_.
11753 (parse_sanitizer_options): Adjusted to also handle
11754 OPT_fsanitize_coverage_.
11755 (common_handle_option): Add OPT_fsanitize_coverage_.
11756 * sancov.c (instrument_comparison, instrument_switch): New function.
11757 (sancov_pass): Add trace-cmp support.
11758 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
11759 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
11760 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
11761 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
11762 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
11763 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
11764 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
11765 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
11766 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
11767 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
11769 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
11771 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
11772 error. Only quit immediately if parsing is complete.
11773 (BEGIN): Initialize fatal_err and parse_done.
11774 (begin fpu, end fpu): Check number of arguments.
11775 (begin arch, end arch): Likewise.
11776 (begin cpu, end cpu): Likewise.
11777 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
11778 (optalias): Likewise.
11780 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
11782 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
11783 * config/arm/arm-isa.h: Delete. Move definitions to ...
11784 * arm-cpus.in: ... here. Use new feature and fgroup values.
11785 * config/arm/arm.c (arm_option_override): Use lower case for feature
11787 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
11788 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
11789 * config/arm/parsecpu.awk (END): Add new command 'isa'.
11791 (print_isa_bits_for): New function.
11792 (gen_isa): New function.
11793 (gen_comm_data): Use print_isa_bits_for.
11794 (define feature): New keyword.
11795 (define fgroup): New keyword.
11796 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
11797 (arm-isa.h): Add rule to generate file.
11798 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
11799 case for feature bit names.
11801 2017-09-06 Richard Biener <rguenther@suse.de>
11803 * tree-ssa-pre.c (NECESSARY): Remove.
11804 (create_expression_by_pieces): Do not touch pass-local flags.
11805 (insert_into_preds_of_block): Likewise.
11806 (do_pre_regular_insertion): Likewise.
11807 (eliminate_insert): Likewise.
11808 (eliminate_dom_walker::before_dom_children): Likewise.
11809 (fini_eliminate): Do not look at inserted_exprs.
11810 (mark_operand_necessary): Remove.
11811 (remove_dead_inserted_code): Replace with simple work-list
11812 algorithm based on inserted_exprs and SSA uses.
11813 (pass_pre::execute): Re-order fini_eliminate and
11814 remove_dead_inserted_code.
11816 2017-09-06 Olivier Hainque <hainque@adacore.com>
11818 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
11819 for VxWorks 7. Adjust surrounding comments.
11821 2017-09-06 Richard Biener <rguenther@suse.de>
11823 * gimple-ssa-strength-reduction.c
11824 (find_candidates_dom_walker::before_dom_children): Also allow
11827 2017-09-06 Richard Biener <rguenther@suse.de>
11829 PR tree-optimization/82108
11830 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
11831 for gap in the non-permutation SLP case.
11833 2017-09-06 Martin Jambor <mjambor@suse.cz>
11835 PR tree-optimization/82078
11836 * tree-sra.c (sort_and_splice_var_accesses): Move call to
11837 add_access_to_work_queue...
11838 (build_accesses_from_assign): ...here.
11839 (propagate_all_subaccesses): Make sure racc is the group
11840 representative, if there is one.
11842 2017-09-06 Jakub Jelinek <jakub@redhat.com>
11844 PR middle-end/82095
11845 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
11848 2017-09-06 Richard Biener <rguenther@suse.de>
11850 * gimple-ssa-strength-reduction.c
11851 (find_candidates_dom_walker::before_doom_children): Use a
11852 type and not a mode check.
11854 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
11857 * config/arm/predicates.md (arm_general_adddi_operand): Create new
11859 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
11861 2017-09-05 Jeff Law <law@redhat.com>
11863 PR tree-optimization/64910
11864 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
11865 cases where we have 3 or more operands.
11867 2017-09-05 Jakub Jelinek <jakub@redhat.com>
11869 PR middle-end/81768
11870 * omp-low.c (lower_omp_for): Recompute tree invariant if
11871 gimple_omp_for_initial/final is ADDR_EXPR.
11873 PR middle-end/81768
11874 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
11875 into gimple val before gimplification fo the COND_EXPR.
11877 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
11879 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
11880 REGION_COPY argument.
11881 (thread_through_all_blocks): Remove unused argument to
11882 duplicate_thread_path.
11884 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11885 Alan Hayward <alan.hayward@arm.com>
11886 David Sherwood <david.sherwood@arm.com>
11888 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
11889 Take a scalar_mode rather than a machine_mode.
11890 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11891 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
11892 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11893 (aarch64_gen_adjusted_ldpstp): Likewise.
11894 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
11896 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11897 Alan Hayward <alan.hayward@arm.com>
11898 David Sherwood <david.sherwood@arm.com>
11900 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
11901 Take a scalar_int_mode instead of a machine_mode.
11902 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
11903 (aarch64_output_scalar_simd_mov_immediate): Likewise.
11904 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
11905 (aarch64_simd_attr_length_rglist): Delete.
11906 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
11907 a scalar_int_mode instead of a machine_mode.
11908 (aarch64_add_offset): Likewise.
11909 (aarch64_internal_mov_immediate): Likewise
11910 (aarch64_add_constant_internal): Likewise.
11911 (aarch64_add_constant): Likewise.
11912 (aarch64_movw_imm): Likewise.
11913 (aarch64_rtx_arith_op_extract_p): Likewise.
11914 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
11915 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
11916 Remove assert that the mode isn't a vector.
11917 (aarch64_output_scalar_simd_mov_immediate): Likewise.
11918 (aarch64_expand_mov_immediate): Update calls after above changes.
11919 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
11920 (aarch64_and_bitmask_imm): Check for scalar integer modes.
11921 (aarch64_move_imm): Likewise.
11922 (aarch64_can_const_movi_rtx_p): Likewise.
11923 (aarch64_strip_extend): Likewise.
11924 (aarch64_extr_rtx_p): Likewise.
11925 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
11926 a CONST_INT when the mode parameter is VOIDmode.
11927 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
11929 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11931 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
11932 * stor-layout.c (bitwise_mode_for_mode): Likewise.
11933 (bitwise_type_for_mode): Update accordingly.
11935 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11937 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
11938 * stor-layout.c (mode_for_size_tree): Likewise.
11939 (mode_for_array): Update accordingly.
11940 (layout_decl): Likewise.
11941 (compute_record_mode): Likewise. Only set the mode once.
11943 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11945 * target.def (get_mask_mode): Change return type to opt_mode.
11947 * doc/tm.texi: Regenerate.
11948 * targhooks.h (default_get_mask_mode): Return an opt_mode.
11949 * targhooks.c (default_get_mask_mode): Likewise.
11950 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
11951 * optabs-query.c (can_vec_mask_load_store_p): Update use of
11952 targetm.get_mask_mode.
11953 * tree.c (build_truth_vector_type): Likewise.
11955 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11957 * machmode.h (mode_for_vector): Return an opt_mode.
11958 * stor-layout.c (mode_for_vector): Likewise.
11959 (mode_for_int_vector): Update accordingly.
11960 (layout_type): Likewise.
11961 * config/i386/i386.c (emit_memmov): Likewise.
11962 (ix86_expand_set_or_movmem): Likewise.
11963 (ix86_expand_vector_init): Likewise.
11964 (ix86_get_mask_mode): Likewise.
11965 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
11967 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
11968 * expmed.c (extract_bit_field_1): Likewise.
11969 * expr.c (expand_expr_real_2): Likewise.
11970 * optabs-query.c (can_vec_perm_p): Likewise.
11971 (can_vec_mask_load_store_p): Likewise.
11972 * optabs.c (expand_vec_perm): Likewise.
11973 * targhooks.c (default_get_mask_mode): Likewise.
11974 * tree-vect-stmts.c (vectorizable_store): Likewise.
11975 (vectorizable_load): Likewise.
11976 (get_vectype_for_scalar_type_and_size): Likewise.
11978 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11980 * machmode.h (mode_for_int_vector): New function.
11981 * stor-layout.c (mode_for_int_vector): Likewise.
11982 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
11983 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
11984 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11985 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11986 (s390_expand_vcond): Likewise.
11988 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11990 * machmode.h (opt_machine_mode): New type.
11991 (opt_mode<T>): Allow construction from anything that can be
11993 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
11994 (mode_for_size): Return an opt_machine_mode.
11995 * stor-layout.c (mode_for_size): Likewise.
11996 (mode_for_size_tree): Update call accordingly.
11997 (bitwise_mode_for_mode): Likewise.
11998 (make_fract_type): Likewise.
11999 (make_accum_type): Likewise.
12000 * caller-save.c (replace_reg_with_saved_mem): Update call
12002 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12003 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12004 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12005 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12006 * expmed.c (extract_bit_field_1): Likewise.
12007 * reload.c (get_secondary_mem): Likewise.
12008 * varasm.c (assemble_integer): Likewise.
12009 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
12012 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12014 * machmode.h (decimal_float_mode_for_size): New function.
12015 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
12016 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
12017 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
12018 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
12019 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
12020 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
12022 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12024 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
12025 (get_builtin_sync_mode): Likewise.
12026 (expand_ifn_atomic_compare_exchange): Likewise.
12027 (expand_builtin_atomic_clear): Likewise.
12028 (expand_builtin_atomic_test_and_set): Likewise.
12029 (fold_builtin_atomic_always_lock_free): Likewise.
12030 * calls.c (compute_argument_addresses): Likewise.
12031 (emit_library_call_value_1): Likewise.
12032 (store_one_arg): Likewise.
12033 * combine.c (combine_instructions): Likewise.
12034 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
12035 * config/arm/arm.c (arm_function_value): Likewise.
12036 (aapcs_allocate_return_reg): Likewise.
12037 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
12038 * config/i386/i386.c (construct_container): Likewise.
12039 (ix86_gimplify_va_arg): Likewise.
12040 (ix86_expand_sse_cmp): Likewise.
12041 (emit_memmov): Likewise.
12042 (emit_memset): Likewise.
12043 (expand_small_movmem_or_setmem): Likewise.
12044 (ix86_expand_pextr): Likewise.
12045 (ix86_expand_pinsr): Likewise.
12046 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
12047 * config/microblaze/microblaze.c (microblaze_block_move_straight):
12049 * config/mips/mips.c (mips_function_value_1) Likewise.
12050 (mips_block_move_straight): Likewise.
12051 (mips_expand_ins_as_unaligned_store): Likewise.
12052 * config/powerpcspe/powerpcspe.c
12053 (rs6000_darwin64_record_arg_advance_flush): Likewise.
12054 (rs6000_darwin64_record_arg_flush): Likewise.
12055 * config/rs6000/rs6000.c
12056 (rs6000_darwin64_record_arg_advance_flush): Likewise.
12057 (rs6000_darwin64_record_arg_flush): Likewise.
12058 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
12059 (sparc_function_value_1): Likewise.
12060 * config/spu/spu.c (adjust_operand): Likewise.
12061 (spu_emit_branch_or_set): Likewise.
12062 (arith_immediate_p): Likewise.
12063 * emit-rtl.c (gen_lowpart_common): Likewise.
12064 * expr.c (expand_expr_real_1): Likewise.
12065 * function.c (assign_parm_setup_block): Likewise.
12066 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
12067 * reload1.c (alter_reg): Likewise.
12068 * stor-layout.c (mode_for_vector): Likewise.
12069 (layout_type): Likewise.
12071 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12073 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
12074 (spu_convert_move): Likewise.
12075 * lower-subreg.c (resolve_simple_move): Likewise.
12077 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12080 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
12081 define_insn to a define_expand.
12082 (altivec_vsum2sws_direct): New define_insn.
12083 (altivec_vsumsws): Convert from a define_insn to a define_expand.
12085 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
12087 * config/arm/arm.c (arm_option_params_internal): Improve setting of
12090 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
12095 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
12096 realign stack if stack alignment needed is less than incoming
12099 2017-09-05 Marek Polacek <polacek@redhat.com>
12102 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
12105 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
12107 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
12109 2017-09-05 Richard Biener <rguenther@suse.de>
12111 PR tree-optimization/82084
12112 * fold-const.c (can_native_encode_string_p): Handle wide characters.
12114 2017-09-05 Richard Biener <rguenther@suse.de>
12116 PR tree-optimization/82102
12117 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
12119 2017-09-05 Martin Liska <mliska@suse.cz>
12121 PR tree-optimization/82032
12122 * tree-cfg.c (generate_range_test): New function.
12123 * tree-cfg.h (generate_range_test): Declared here.
12124 * tree-cfgcleanup.c (convert_single_case_switch): New function.
12125 (cleanup_control_expr_graph): Use it.
12126 * tree-switch-conversion.c (try_switch_expansion): Remove
12128 (emit_case_nodes): Use generate_range_test.
12130 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
12133 * config/i386/i386.md (*<btsc><mode>_mask): Add
12134 TARGET_USE_BT to insn constraint.
12135 (*btr<mode>_mask): Ditto.
12137 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
12139 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
12140 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
12142 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
12145 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
12146 TARGET_NEON and TARGET_IWMMXT.
12147 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
12148 TARGET_NEON and TARGET_IWMMXT.
12149 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
12151 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
12153 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
12154 (ix86_rewrite_tls_address): Ditto.
12155 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
12156 (ix86_rewrite_tls_address_1): Ditto.
12157 (ix86_rewrite_tls_address): Ditto.
12158 * config/i386/predicates.md (tls_address_pattern): New predicate.
12159 * config/i386/i386.md (TLS address splitter): New splitter.
12161 2017-09-04 Richard Biener <rguenther@suse.de>
12163 PR tree-optimization/82084
12164 * fold-const.h (can_native_encode_string_p): Declare.
12165 * fold-const.c (can_native_encode_string_p): Factor out from ...
12166 (native_encode_string): ... here.
12167 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
12168 vectorizing stores from constants we later cannot handle.
12170 2017-09-04 Marek Polacek <polacek@redhat.com>
12173 * doc/invoke.texi: Update -Wtautological-compare documentation.
12175 2017-09-04 Jeff Law <law@redhat.com>
12177 PR tree-optimization/64910
12178 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
12179 swap the first and last operand if the last is a constant.
12181 2017-09-04 Marek Polacek <polacek@redhat.com>
12184 * convert.c (do_narrow): When sanitizing signed integer overflows,
12185 bail out for signed types.
12186 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
12188 2017-09-04 Richard Biener <rguenther@suse.de>
12190 PR tree-optimization/82060
12191 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
12192 Move devirtualization after stmt folding and before EH/AB/noreturn
12193 cleanup to get the stmt refs canonicalized. Use a bool instead
12194 of gimple_modified_p since that doesn't work for NOPs. Schedule
12195 NOPs generated by folding for removal.
12197 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12198 Alan Hayward <alan.hayward@arm.com>
12199 David Sherwood <david.sherwood@arm.com>
12201 * coretypes.h (pad_direction): New enum.
12202 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
12203 (FUNCTION_ARG_PADDING): Likewise.
12204 * target.def (function_arg_padding): New hook.
12205 * targhooks.h (default_function_arg_padding): Declare.
12206 * targhooks.c (default_function_arg_padding): New function.
12207 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
12208 (TARGET_FUNCTION_ARG_PADDING): ...this.
12209 * doc/tm.texi: Regenerate.
12210 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
12211 instead of direction.
12212 (compute_argument_addresses): Likewise.
12213 (load_register_parameters): Likewise.
12214 (emit_library_call_value_1): Likewise.
12215 (store_one_arg): Use targetm.calls.function_arg_padding instead
12216 of FUNCTION_ARG_PADDING.
12217 (must_pass_in_stack_var_size_or_pad): Likewise.
12218 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
12219 (emit_group_store): Likewise.
12220 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
12221 instead of FUNCTION_ARG_PADDING.
12222 (emit_push_insn): Likewise, and propagate enum change throughout
12224 * function.h (direction): Delete.
12225 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
12227 * function.c (assign_parm_find_stack_rtl): Likewise.
12228 (assign_parm_setup_block_p): Likewise.
12229 (assign_parm_setup_block): Likewise.
12230 (gimplify_parameters): Likewise.
12231 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
12232 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
12234 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
12235 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12236 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
12237 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
12238 (aarch64_function_arg_padding): ...this new function.
12239 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
12240 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12241 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
12242 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12243 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
12244 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12245 (arm_pad_arg_upward): Replace with...
12246 (arm_function_arg_padding): ...this new function.
12247 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
12249 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
12250 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
12251 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12252 (ia64_hpux_function_arg_padding): Replace with...
12253 (ia64_function_arg_padding): ...this new function. Use pad_direction
12254 instead of direction. Check for TARGET_HPUX.
12255 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
12256 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12257 (iq2000_function_arg_padding): New function.
12258 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
12259 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
12260 (mips_function_arg_padding): ...this new function.
12261 (mips_pad_reg_upward): Update accordingly.
12262 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12263 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
12264 targetm.calls.function_arg_padding.
12265 (FUNCTION_ARG_PADDING): Delete.
12266 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12267 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
12268 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
12269 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
12270 (nios2_block_reg_padding): Return pad_direction instead of direction.
12271 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
12272 instead of direction.
12273 (nios2_function_arg_padding): Likewise. Make static.
12274 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12275 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
12276 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
12277 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
12278 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12279 (pa_function_arg_padding): Make static. Return pad_direction instead
12281 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
12282 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
12283 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
12284 instead of direction. Use targetm.calls.function_arg_padding.
12285 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
12286 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
12287 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
12288 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
12289 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
12291 (function_arg_padding): Rename to...
12292 (rs6000_function_arg_padding): ...this. Make static. Return
12293 pad_direction instead of direction.
12294 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
12295 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
12296 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
12297 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
12298 instead of direction. Use targetm.calls.function_arg_padding.
12299 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
12300 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
12301 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
12302 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
12303 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12304 (function_arg_padding): Rename to...
12305 (rs6000_function_arg_padding): ...this. Make static. Return
12306 pad_direction instead of direction.
12307 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
12308 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
12309 * config/s390/s390.c (s390_function_arg_padding): New function.
12310 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12311 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
12312 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
12313 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12314 (function_arg_padding): Rename to...
12315 (sparc_function_arg_padding): ...this. Make static. Return
12316 pad_direction instead of direction.
12317 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
12318 * config/spu/spu.c (spu_function_arg_padding): New function.
12319 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12320 * system.h (FUNCTION_ARG_PADDING): Poison.
12322 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12323 Alan Hayward <alan.hayward@arm.com>
12324 David Sherwood <david.sherwood@arm.com>
12326 * target.def (modes_tieable_p): New hook.
12327 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
12328 (TARGET_MODES_TIEABLE_P): ...this.
12329 * doc/tm.texi.in: Regenerate.
12330 * hooks.h (hook_bool_mode_mode_true): Declare.
12331 * hooks.c (hook_bool_mode_mode_true): New function.
12332 * combine.c (subst): Use targetm.modes_tieable_p instead of
12334 * dse.c (find_shift_sequence): Likewise.
12335 * expmed.c (extract_low_bits): Likewise.
12336 * lower-subreg.c: Include target.h.
12337 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
12339 * rtlanal.c (rtx_cost): Likewise.
12340 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
12341 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
12342 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
12343 (TARGET_MODES_TIEABLE_P): Redefine.
12344 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
12345 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
12346 (TARGET_MODES_TIEABLE_P): Redefine.
12347 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
12348 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
12349 (arc_modes_tieable_p): New function.
12350 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
12351 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
12352 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
12353 (arm_modes_tieable_p): Make static.
12354 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
12355 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
12356 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
12357 (TARGET_MODES_TIEABLE_P): Redefine.
12358 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
12359 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
12360 (TARGET_MODES_TIEABLE_P): Redefine.
12361 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
12362 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
12363 (cr16_modes_tieable_p): New function.
12364 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
12365 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
12366 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
12367 (TRULY_NOOP_TRUNCATION): Update comment.
12368 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
12369 (TRULY_NOOP_TRUNCATION): Update comment.
12370 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
12371 (frv_modes_tieable_p): New function.
12372 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
12373 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
12374 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
12375 (TARGET_MODES_TIEABLE_P): Redefine.
12376 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
12377 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
12378 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
12379 (TARGET_MODES_TIEABLE_P): Redefine.
12380 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
12381 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
12382 (ia64_modes_tieable_p): New function.
12383 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
12384 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
12385 (iq2000_modes_tieable_p): New function.
12386 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
12387 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
12388 (lm32_modes_tieable_p): New function.
12389 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
12390 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
12391 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
12392 (TARGET_MODES_TIEABLE_P): Redefine.
12393 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
12394 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
12395 (m32r_modes_tieable_p): New function.
12396 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
12397 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
12398 (m68k_modes_tieable_p): New function.
12399 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
12400 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
12401 (mcore_modes_tieable_p): New function.
12402 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
12403 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
12405 (TARGET_MODES_TIEABLE_P): Redefine.
12406 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
12407 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
12408 * config/mips/mips.c (mips_modes_tieable_p): Make static.
12409 (TARGET_MODES_TIEABLE_P): Redefine.
12410 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
12411 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
12412 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
12413 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
12414 (mn10300_modes_tieable_p): ...this and make static.
12415 (TARGET_MODES_TIEABLE_P): Redefine.
12416 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
12417 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
12418 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
12419 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
12420 (msp430_modes_tieable_p): Make static.
12421 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
12422 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
12423 (TARGET_MODES_TIEABLE_P): Redefine.
12424 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
12425 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
12426 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
12427 (TARGET_MODES_TIEABLE_P): Redefine.
12428 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
12429 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
12430 * config/pa/pa.c (pa_modes_tieable_p): Make static.
12431 (TARGET_MODES_TIEABLE_P): Redefine.
12432 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
12433 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
12434 (pdp11_modes_tieable_p): New function.
12435 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
12436 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
12437 (rs6000_modes_tieable_p): New function.
12438 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
12439 * config/powerpcspe/powerpcspe.md: Update comment.
12440 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
12441 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
12442 (TARGET_MODES_TIEABLE_P): Redefine.
12443 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
12444 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
12445 (rl78_modes_tieable_p): New function.
12446 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
12447 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
12448 (rs6000_modes_tieable_p): New function.
12449 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
12450 * config/rs6000/rs6000.md: Update comment.
12451 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
12452 * config/rx/rx.c (rx_modes_tieable_p): New function.
12453 (TARGET_MODES_TIEABLE_P): Redefine.
12454 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
12455 * config/s390/s390.c (s390_modes_tieable_p): New function.
12456 (TARGET_MODES_TIEABLE_P): Redefine.
12457 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
12458 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
12459 (sh_modes_tieable_p): New function.
12460 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
12461 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
12462 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
12463 (sparc_modes_tieable_p): Make static.
12464 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
12465 * config/spu/spu.c (spu_modes_tieable_p): New function.
12466 (TARGET_MODES_TIEABLE_P): Redefine.
12467 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
12468 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
12469 (TARGET_MODES_TIEABLE_P): Redefine.
12470 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
12471 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
12472 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
12473 * config/v850/v850.c (v850_modes_tieable_p): New function.
12474 (TARGET_MODES_TIEABLE_P): Redefine.
12475 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
12476 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
12477 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
12478 (visium_modes_tieable_p): New function.
12479 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
12480 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
12481 (xtensa_modes_tieable_p): New function.
12482 * system.h (MODES_TIEABLE_P): Poison.
12484 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12485 Alan Hayward <alan.hayward@arm.com>
12486 David Sherwood <david.sherwood@arm.com>
12488 * target.def (hard_regno_mode_ok): New hook.
12489 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
12490 (TARGET_HARD_REGNO_MODE_OK): ...this.
12491 * doc/tm.texi.in: Regenerate.
12492 * hooks.h (hook_bool_uint_mode_true): Declare.
12493 * hooks.c (hook_bool_uint_mode_true): New function.
12494 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
12495 HARD_REGNO_MODE_OK.
12496 * genpreds.c (write_insn_preds_c): Add an include of target.h.
12497 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
12498 instead of HARD_REGNO_MODE_OK.
12499 * caller-save.c: Include target.h.
12500 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
12501 HARD_REGNO_MODE_OK.
12502 * combine.c (can_combine_p): Likewise.
12503 (combinable_i3pat): Likewise.
12504 (can_change_dest_mode): Likewise.
12505 * expr.c (init_expr_target): Likewise.
12506 (convert_move): Likewise.
12507 (convert_modes): Likewise.
12508 * ira.c (setup_prohibited_class_mode_regs): Likewise.
12509 (setup_prohibited_mode_move_regs): Likewise.
12510 * ira.h (target_ira): Likewise.
12511 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12512 * lra-constraints.c (process_alt_operands): Likewise.
12513 (split_reg): Likewise.
12514 * recog.c (peep2_find_free_register): Likewise.
12515 * ree.c (combine_reaching_defs): Likewise.
12516 * regcprop.c (maybe_mode_change): Likewise.
12517 * reginfo.c (init_reg_sets_1): Likewise.
12518 (choose_hard_reg_mode): Likewise.
12519 (simplifiable_subregs): Likewise.
12520 * regrename.c (check_new_reg_p): Likewise.
12521 * reload.c (find_valid_class): Likewise.
12522 (find_valid_class_1): Likewise.
12523 (reload_inner_reg_of_subreg): Likewise.
12524 (push_reload): Likewise.
12525 (combine_reloads): Likewise.
12526 (find_dummy_reload): Likewise.
12527 (find_reloads): Likewise.
12528 * reload1.c (find_reg): Likewise.
12529 (set_reload_reg): Likewise.
12530 (allocate_reload_reg): Likewise.
12531 (choose_reload_regs): Likewise.
12532 (reload_adjust_reg_for_temp): Likewise.
12533 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
12534 (simplify_subreg_regno): Likewise.
12535 * sel-sched.c (init_regs_for_mode): Likewise.
12536 * varasm.c (make_decl_rtl): Likewise.
12537 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
12538 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
12539 HARD_REGNO_MODE_OK.
12540 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
12541 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
12542 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12543 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
12544 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
12545 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12546 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
12547 (arc_mode_class): Delete.
12548 (HARD_REGNO_MODE_OK): Delete.
12549 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12550 (arc_hard_regno_mode_ok): Rename old array to...
12551 (arc_hard_regno_mode_ok_modes): ...this.
12552 (arc_conditional_register_usage): Update accordingly.
12553 (arc_mode_class): Make static.
12554 (arc_hard_regno_mode_ok): New function.
12555 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
12556 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
12557 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12558 (arm_hard_regno_mode_ok): Make static.
12559 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
12560 HARD_REGNO_MODE_OK.
12561 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
12562 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
12563 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
12565 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12566 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
12567 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
12568 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
12569 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
12570 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12571 * config/bfin/predicates.md (valid_reg_operand): Use
12572 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
12573 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
12574 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
12575 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12576 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
12577 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
12578 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12579 (cr16_hard_regno_mode_ok): Make static and return a bool.
12580 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
12581 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12582 (cris_hard_regno_mode_ok): New function.
12583 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
12584 (epiphany_mode_class): Delete.
12585 (HARD_REGNO_MODE_OK): Delete.
12586 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
12587 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12588 (hard_regno_mode_ok): Rename to...
12589 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
12590 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
12591 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
12592 HARD_REGNO_MODE_OK.
12593 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
12594 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
12595 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12596 (frv_hard_regno_mode_ok): Make static and return a bool.
12597 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
12598 HARD_REGNO_MODE_OK.
12599 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
12600 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
12601 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
12602 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
12604 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12605 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
12606 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
12607 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
12609 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12610 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
12611 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12612 (ia64_hard_regno_mode_ok): New function.
12613 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
12614 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12615 (iq2000_hard_regno_mode_ok): New function.
12616 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
12617 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12618 (lm32_hard_regno_mode_ok): New function.
12619 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
12620 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
12621 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
12622 instead of HARD_REGNO_MODE_OK.
12623 (m32c_hard_regno_ok): Rename to...
12624 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
12625 (m32c_cannot_change_mode_class): Update accordingly.
12626 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12627 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
12628 (m32r_mode_class): Delete.
12629 (HARD_REGNO_MODE_OK): Delete.
12630 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12631 (m32r_hard_regno_mode_ok): Rename to...
12632 (m32r_hard_regno_modes): ...this.
12633 (m32r_mode_class): Make static.
12634 (m32r_hard_regno_mode_ok): New function.
12635 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
12636 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
12637 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12638 (m68k_hard_regno_mode_ok): Make static.
12639 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
12640 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12641 (mcore_hard_regno_mode_ok): New function.
12642 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
12643 (HARD_REGNO_MODE_OK): Delete.
12644 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
12646 (microblaze_hard_regno_mode_ok_p): ...this and make static.
12647 (microblaze_hard_regno_mode_ok): New function.
12648 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12649 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
12650 (mips_hard_regno_mode_ok): Delete.
12651 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
12652 (mips_hard_regno_mode_ok_p): ...this and make static.
12653 (mips_hard_regno_mode_ok_p): Rename to...
12654 (mips_hard_regno_mode_ok_uncached): ...this.
12655 (mips_hard_regno_mode_ok): New function.
12656 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
12657 of HARD_REGNO_MODE_OK.
12658 (mips_option_override): Update after above name changes.
12659 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12660 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
12661 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
12662 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
12663 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
12664 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12665 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
12666 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
12667 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
12668 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12669 (msp430_hard_regno_mode_ok): Make static and return a bool.
12670 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
12671 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
12672 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
12674 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12675 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
12676 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
12677 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
12678 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
12679 (PA_HARD_REGNO_MODE_OK): ...this
12680 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
12681 (PA_HARD_REGNO_MODE_OK): ...this.
12682 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12683 (pa_hard_regno_mode_ok): New function.
12684 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
12685 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12686 (pdp11_hard_regno_mode_ok): New function.
12687 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
12688 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
12690 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
12692 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12693 (rs6000_hard_regno_mode_ok): Rename to...
12694 (rs6000_hard_regno_mode_ok_uncached): ...this.
12695 (rs6000_init_hard_regno_mode_ok): Update accordingly.
12696 (rs6000_hard_regno_mode_ok): New function.
12697 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
12698 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
12699 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
12700 (riscv_hard_regno_mode_ok): ...this and make static.
12701 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12702 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
12703 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
12704 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12705 (rl78_hard_regno_mode_ok): Make static and return bool.
12706 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
12707 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
12709 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
12710 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12711 (rs6000_hard_regno_mode_ok): Rename to...
12712 (rs6000_hard_regno_mode_ok_uncached): ...this.
12713 (rs6000_init_hard_regno_mode_ok): Update accordingly.
12714 (rs6000_hard_regno_mode_ok): New function.
12715 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
12716 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
12717 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12718 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
12719 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
12720 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
12721 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12722 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
12723 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
12724 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12725 (sh_hard_regno_mode_ok): Make static.
12726 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
12727 instead of HARD_REGNO_MODE_OK.
12728 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
12729 (sparc_mode_class): Delete.
12730 (HARD_REGNO_MODE_OK): Delete.
12731 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12732 (hard_regno_mode_classes): Make static.
12733 (sparc_mode_class): Likewise.
12734 (sparc_hard_regno_mode_ok): New function.
12735 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
12736 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
12737 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
12739 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12740 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
12741 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
12742 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
12743 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
12744 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12745 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
12746 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
12747 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12748 (visium_hard_regno_mode_ok): New function.
12749 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
12750 instead of HARD_REGNO_MODE_OK.
12751 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
12752 (HARD_REGNO_MODE_OK): Delete.
12753 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
12754 (xtensa_hard_regno_mode_ok_p): ...this and make static.
12755 (xtensa_option_override): Update accordingly.
12756 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12757 (xtensa_hard_regno_mode_ok): New function.
12758 * system.h (HARD_REGNO_MODE_OK): Poison.
12760 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12761 Alan Hayward <alan.hayward@arm.com>
12762 David Sherwood <david.sherwood@arm.com>
12764 * target.def (hard_regno_call_part_clobbered): New hook.
12765 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
12766 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
12767 * doc/tm.texi: Regenerate.
12768 * hooks.h (hook_bool_uint_mode_false): Declare.
12769 * hooks.c (hook_bool_uint_mode_false): New function.
12770 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12771 * cselib.c (cselib_process_insn): Use
12772 targetm.hard_regno_call_part_clobbered instead of
12773 HARD_REGNO_CALL_PART_CLOBBERED.
12774 * ira-conflicts.c (ira_build_conflicts): Likewise.
12775 * ira-costs.c (ira_tune_allocno_costs): Likewise.
12776 * lra-constraints.c (need_for_call_save_p): Likewise.
12777 * lra-lives.c: Include target.h.
12778 (check_pseudos_live_through_calls): Use
12779 targetm.hard_regno_call_part_clobbered instead of
12780 HARD_REGNO_CALL_PART_CLOBBERED.
12781 * regcprop.c: Include target.h.
12782 (copyprop_hardreg_forward_1): Use
12783 targetm.hard_regno_call_part_clobbered instead of
12784 HARD_REGNO_CALL_PART_CLOBBERED.
12785 * reginfo.c (choose_hard_reg_mode): Likewise.
12786 * regrename.c (check_new_reg_p): Likewise.
12787 * reload.c (find_equiv_reg): Likewise.
12788 * reload1.c (emit_reload_insns): Likewise.
12789 * sched-deps.c (deps_analyze_insn): Likewise.
12790 * sel-sched.c (init_regs_for_mode): Likewise.
12791 (mark_unavailable_hard_regs): Likewise.
12792 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
12793 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12794 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
12796 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12797 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12798 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
12800 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
12802 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12803 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12804 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
12806 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12807 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12808 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
12810 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12811 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
12813 * config/powerpcspe/powerpcspe.c
12814 (rs6000_hard_regno_call_part_clobbered): New function.
12815 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12816 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12817 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
12819 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12820 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12821 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
12823 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12824 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12825 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
12827 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12828 Alan Hayward <alan.hayward@arm.com>
12829 David Sherwood <david.sherwood@arm.com>
12831 * rtl.h (subreg_memory_offset): Declare.
12832 * emit-rtl.c (subreg_memory_offset): New function.
12833 * expmed.c (store_bit_field_1): Use it.
12834 * expr.c (undefined_operand_subword_p): Likewise.
12835 * simplify-rtx.c (simplify_subreg): Likewise.
12837 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
12839 PR rtl-optimization/57448
12842 * optabs.c (expand_atomic_load): Place compiler memory barriers if
12843 using atomic_load pattern.
12844 (expand_atomic_store): Likewise.
12846 2017-09-04 Jakub Jelinek <jakub@redhat.com>
12849 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
12850 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
12851 handling. Use replace_call_with_value with NULL instead of
12852 gsi_replace, unlink_stmt_vdef and release_defs.
12854 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
12857 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
12859 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12862 * rtl.h (emit_library_call_value_1): Declare.
12863 (emit_library_call): Replace declaration with a series of overloads.
12864 Remove the parameter count argument.
12865 (emit_library_call_value): Likewise.
12866 * calls.c (emit_library_call_value_1): Make global. Replace varargs
12867 with an "rtx_mode_t *".
12868 (emit_library_call_value): Delete.
12869 (emit_library_call): Likewise.
12870 * asan.c (asan_emit_stack_protection): Update calls accordingly.
12871 (asan_emit_allocas_unpoison): Likewise.
12872 * builtins.c (expand_builtin_powi): Likewise.
12873 (expand_asan_emit_allocas_unpoison): Likewise.
12874 * cfgexpand.c (expand_main_function): Likewise.
12875 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
12876 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
12877 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
12878 * config/arm/arm.c (arm_trampoline_init): Likewise.
12879 (arm_call_tls_get_addr): Likewise.
12880 (arm_expand_divmod_libfunc): Likewise.
12881 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
12882 (smulsi3_highpart): Likewise.
12883 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
12884 (c6x_expand_compare): Likewise.
12885 (c6x_expand_movmem): Likewise.
12886 * config/frv/frv.c (frv_trampoline_init): Likewise.
12887 * config/i386/i386.c (ix86_trampoline_init): Likewise.
12888 (ix86_expand_divmod_libfunc): Likewise.
12889 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
12890 (ia64_expand_compare): Likewise.
12891 (ia64_profile_hook): Likewise.
12892 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
12893 (nonlocal_goto): Likewise.
12894 (restore_stack_nonlocal): Likewise.
12895 * config/m32r/m32r.c (block_move_call): Likewise.
12896 (m32r_trampoline_init): Likewise.
12897 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
12898 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
12899 (m68k_call_m68k_read_tp): Likewise.
12900 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
12901 (microblaze_expand_divide): Likewise.
12902 * config/mips/mips.h (mips_args): Likewise.
12903 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
12904 (MIPS_ICACHE_SYNC): Likewise.
12905 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
12906 (nios2_trampoline_init): Likewise.
12907 * config/pa/pa.c (hppa_tls_call): Likewise.
12908 (pa_trampoline_init): Likewise.
12909 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
12910 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
12911 (expand_strn_compare): Likewise.
12912 (rs6000_generate_compare): Likewise.
12913 (rs6000_expand_float128_convert): Likewise.
12914 (output_profile_hook): Likewise.
12915 (rs6000_trampoline_init): Likewise.
12916 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
12917 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
12918 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
12919 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
12920 (rs6000_generate_compare): Likewise.
12921 (rs6000_expand_float128_convert): Likewise.
12922 (output_profile_hook): Likewise.
12923 (rs6000_trampoline_init): Likewise.
12924 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
12925 * config/sh/sh.c (sh_trampoline_init): Likewise.
12926 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
12927 (sparc_emit_float_lib_cmp): Likewise.
12928 (sparc32_initialize_trampoline): Likewise.
12929 (sparc64_initialize_trampoline): Likewise.
12930 (sparc_profile_hook): Likewise.
12931 * config/spu/spu.c (ea_load_store): Likewise.
12932 * config/spu/spu.md (floatunssidf2): Likewise.
12933 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
12934 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
12935 * config/visium/visium.c (expand_block_move_4): Likewise.
12936 (expand_block_move_2): Likewise.
12937 (expand_block_move_1): Likewise.
12938 (expand_block_set_4): Likewise.
12939 (expand_block_set_2): Likewise.
12940 (expand_block_set_1): Likewise.
12941 (visium_trampoline_init): Likewise.
12942 (visium_profile_hook): Likewise.
12943 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
12944 (xtensa_setup_frame_addresses): Likewise.
12945 (xtensa_trampoline_init): Likewise.
12946 * except.c (sjlj_emit_function_enter): Likewise.
12947 (sjlj_emit_function_exit): Likewise.
12948 * explow.c (allocate_dynamic_stack_space): Likewise.
12949 (probe_stack_range): Likewise.
12950 * expr.c (convert_mode_scalar): Likewise.
12951 * optabs.c (expand_binop): Likewise.
12952 (expand_twoval_binop_libfunc): Likewise.
12953 (expand_unop): Likewise.
12954 (prepare_cmp_insn): Likewise.
12955 (prepare_float_lib_cmp): Likewise.
12956 (expand_float): Likewise.
12957 (expand_fix): Likewise.
12958 (expand_fixed_convert): Likewise.
12959 (maybe_emit_sync_lock_test_and_set): Likewise.
12960 (expand_atomic_compare_and_swap): Likewise.
12961 (expand_mem_thread_fence): Likewise.
12962 (expand_atomic_fetch_op): Likewise.
12964 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
12966 * doc/generic.texi (OpenACC): Adjust URL.
12967 * doc/invoke.texi (C Dialect Options): Ditto.
12969 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
12971 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
12972 predicate for operand 1. Add (m,<S>) constraint.
12973 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
12974 Prevent memory operand 1 with register operand 2.
12976 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
12978 PR rtl-optimization/82024
12979 * combine.c (try_combine): If the combination result is a PARALLEL,
12980 and we only need to retain the SET in there that would be placed
12981 at I2, check that we can place that at I3 instead, before doing so.
12983 2017-09-01 Jakub Jelinek <jakub@redhat.com>
12986 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
12988 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
12989 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
12992 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
12993 Jeff Law <law@redhat.com>
12995 * varasm.c (bss_initializer_p): Do not put constants into .bss
12996 (categorize_decl_for_section): Handle bss_initializer_p returning
12997 false when DECL_INITIAL is NULL.
12999 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13002 * config/s390/s390.c (s390_can_inline_p): New function.
13004 2017-09-01 Jeff Law <law@redhat.com>
13006 PR tree-optimization/82052
13007 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
13008 Always initialize the returned slot after a hash table miss
13009 when INSERT is true.
13011 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
13013 * config/s390/s390.md (mem_signal_fence): Remove.
13014 * doc/md.texi (mem_signal_fence): Remove.
13015 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
13017 * target-insns.def (mem_signal_fence): Remove.
13019 2017-09-01 Jakub Jelinek <jakub@redhat.com>
13022 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
13025 * asan.c (create_odr_indicator): Strip name encoding from assembler
13026 name before appending it after __odr_asan_.
13028 2017-09-01 Martin Liska <mliska@suse.cz>
13030 PR tree-optimization/82059
13031 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
13032 frequency only when an edge is redirected.
13034 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13036 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
13037 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
13038 (arc_conditional_register_usage): Remove ARC600 lp_count
13040 (arc_file_start): Emit Tag_ARC_CPU_variation.
13041 (arc_can_use_doloop_p): New conditions to use ZOLs.
13042 (hwloop_fail): New function.
13043 (hwloop_optimize): Likewise.
13044 (hwloop_pattern_reg): Likewise.
13045 (arc_doloop_hooks): New struct, to be used with reorg_loops.
13046 (arc_reorg_loops): New function, calls reorg_loops.
13047 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
13048 (arc600_corereg_hazard): Remove ZOL checking, case handled by
13050 (arc_loop_hazard): Remove function, functionality moved into
13052 (arc_hazard): Remove arc_loop_hazard call.
13053 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
13054 into hwloop_optimize.
13055 (arc_label_align): Remove ZOL handling.
13056 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
13057 * config/arc/arc.md (doloop_begin): Remove pattern.
13058 (doloop_begin_i): Likewise.
13059 (doloop_end_i): Likewise.
13060 (doloop_fallback): Likewise.
13061 (doloop_fallback_m): Likewise.
13062 (doloop_end): Reimplement expand.
13063 (arc_lp): New pattern for LP instruction.
13064 (loop_end): New pattern.
13065 (loop_fail): Likewise.
13066 (decrement_and_branch_until_zero): Likewise.
13067 * config/arc/arc.opt (mlpc-width): New option.
13068 * doc/invoke.texi (mlpc-width): Document option.
13070 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13072 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
13073 (arc_ccfsm_advance): Fix checking for delay slots.
13074 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
13076 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13078 * config/arc/arc.md (movqi_insn): Add stores to save constant long
13080 (movhi_insn): Update store instruction constraint which are saving
13081 6-bit short immediates.
13082 (movsi_insn): Consider also short scaled load operations.
13083 (zero_extendhisi2_i): Use Usd constraint instead of T.
13084 (extendhisi2_i): Add q constraint.
13085 (arc_clzsi2): Add type and length attributes.
13086 (arc_ctzsi2): Likewise.
13087 * config/arc/constraints.md (Usc): Update constraint, the
13088 assembler can parse two relocations for a single instruction.
13090 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13092 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
13093 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
13095 2017-08-31 Olivier Hainque <hainque@adacore.com>
13097 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
13098 match as powerpc-wrs-vxworks*.
13100 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
13102 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
13103 register constraint for by-element operand.
13104 (aarch64_mls_elt_merge<mode>): Likewise.
13106 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13108 * config/arc/arc.c (arc_can_follow_jump): Check for short
13111 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13113 * config.gcc: Use g.opt for arc.
13114 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
13115 functionality moved to ...
13116 (legitimate_scaled_address_p): New function, ...here.
13117 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
13118 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
13119 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
13121 (arc_override_options): Handle G option.
13122 (arc_output_pic_addr_const): Correct function definition.
13123 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
13124 (arc_decl_anon_ns_mem_p): Delete.
13125 (arc_in_small_data_p): Overhaul this function to take into
13126 consideration the value given via G option.
13127 (arc_rewrite_small_data_1): Renamed and corrected old
13128 arc_rewrite_small_data function.
13129 (arc_rewrite_small_data): New function.
13130 (small_data_pattern): Don't use pic_offset_table_rtx.
13131 * config/arc/arc.h (CC1_SPEC): Recognize G option.
13132 * config/arc/simdext.md (movmisalignv2hi): Use
13133 prepare_move_operands function.
13135 (movmisalign*): Likewise.
13136 * doc/invoke.texi (ARC options): Document -G option.
13138 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13140 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
13142 * config/arc/arc.c (arc_print_operand): Output scalled address for
13143 sdata whenever is possible.
13144 (arc_in_small_data_p): Allow sdata for 64bit datum when double
13145 load/stores are available.
13146 (compact_sda_memory_operand): Check for the alignment required by
13147 code density instructions.
13148 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
13150 * config/arc/constraints.md (Usd): Update constraint.
13151 (Us0): New constraint.
13152 (Usc): Update constraint.
13154 2017-08-31 Richard Biener <rguenther@suse.de>
13156 PR middle-end/82054
13157 * dwarf2out.c (dwarf2out_early_global_decl): Process each
13158 function only once.
13160 2017-08-31 Tamar Christina <tamar.christina@arm.com>
13162 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
13163 Resize type_signature.
13165 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13166 Alan Hayward <alan.hayward@arm.com>
13167 David Sherwood <david.sherwood@arm.com>
13169 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
13170 subregs whose inner modes can be stored in GPRs.
13171 (aarch64_classify_index): Likewise.
13173 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13174 Alan Hayward <alan.hayward@arm.com>
13175 David Sherwood <david.sherwood@arm.com>
13177 * config/aarch64/iterators.md (V_cmp_result): Rename to...
13178 (V_INT_EQUIV): ...this.
13179 (v_cmp_result): Rename to...
13180 (v_int_equiv): ...this.
13181 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
13182 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
13183 (copysign<mode>3): Likewise.
13184 (aarch64_simd_bsl<mode>_internal): Likewise.
13185 (aarch64_simd_bsl<mode>): Likewise.
13186 (vec_cmp<mode><mode>): Likewise.
13187 (vcond<mode><mode>): Likewise.
13188 (vcond<v_cmp_mixed><mode>): Likewise.
13189 (vcondu<mode><v_cmp_mixed>): Likewise.
13190 (aarch64_cm<optab><mode>): Likewise.
13191 (aarch64_cmtst<mode>): Likewise.
13192 (aarch64_fac<optab><mode>): Likewise.
13193 (vec_perm_const<mode>): Likewise.
13194 (vcond_mask_<mode><v_cmp_result>): Rename to...
13195 (vcond_mask_<mode><v_int_equiv>): ...this.
13196 (vec_cmp<mode><v_cmp_result>): Rename to...
13197 (vec_cmp<mode><v_int_equiv>): ...this.
13199 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13200 Alan Hayward <alan.hayward@arm.com>
13201 David Sherwood <david.sherwood@arm.com>
13203 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
13205 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
13206 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
13207 (UNSPEC_LD4_DREG): New unspecs.
13208 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
13209 (aarch64_ld2<mode>_dreg_be): Replace with...
13210 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
13212 (aarch64_ld3<mode>_dreg_le)
13213 (aarch64_ld3<mode>_dreg_be): Replace with...
13214 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
13216 (aarch64_ld4<mode>_dreg_le)
13217 (aarch64_ld4<mode>_dreg_be): Replace with...
13218 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
13221 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13223 PR tree-optimization/81987
13224 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
13225 insert an initializer in a location not dominated by the stride
13228 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
13230 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
13231 on the entire header of the finally block in the fallthru case.
13233 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
13235 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
13237 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
13239 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
13240 rs6000_emit_move_from_cr and call renamed function.
13241 (rs6000_emit_prologue): Call renamed functions.
13242 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
13243 movesi_from_cr, remove volatile CRs.
13245 2017-08-30 Jon Beniston <jon@beniston.com>
13246 Richard Biener <rguenther@suse.de>
13248 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
13249 of VECTOR_MODE_P check.
13250 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
13251 element vector types.
13253 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13255 * df.h (df_read_modify_subreg_p): Remove in favor of...
13256 * rtl.h (read_modify_subreg_p): ...this new function. Take a
13257 const_rtx instead of an rtx.
13258 * cprop.c (local_cprop_find_used_regs): Update accordingly.
13259 * df-problems.c (df_word_lr_mark_ref): Likewise.
13260 * ira-lives.c (mark_pseudo_reg_live): Likewise.
13261 (mark_pseudo_reg_dead): Likewise.
13262 (mark_ref_dead): Likewise.
13263 * reginfo.c (init_subregs_of_mode): Likewise.
13264 * sched-deps.c (sched_analyze_1): Likewise.
13265 * df-scan.c (df_def_record_1): Likewise.
13266 (df_uses_record): Likewise.
13267 (df_read_modify_subreg_p): Remove in favor of...
13268 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
13269 const_rtx instead of an rtx.
13271 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13272 Alan Hayward <alan.hayward@arm.com>
13273 David Sherwood <david.sherwood@arm.com>
13275 * rtl.h (partial_subreg_p): New function.
13276 * caller-save.c (save_call_clobbered_regs): Use it.
13277 * calls.c (expand_call): Likewise.
13278 * combine.c (combinable_i3pat): Likewise.
13279 (simplify_set): Likewise.
13280 (make_extraction): Likewise.
13281 (make_compound_operation_int): Likewise.
13282 (gen_lowpart_or_truncate): Likewise.
13283 (force_to_mode): Likewise.
13284 (make_field_assignment): Likewise.
13285 (reg_truncated_to_mode): Likewise.
13286 (record_truncated_value): Likewise.
13287 (move_deaths): Likewise.
13288 * cse.c (record_jump_cond): Likewise.
13289 (cse_insn): Likewise.
13290 * cselib.c (cselib_lookup_1): Likewise.
13291 * expmed.c (extract_bit_field_using_extv): Likewise.
13292 * function.c (assign_parm_setup_reg): Likewise.
13293 * ifcvt.c (noce_convert_multiple_sets): Likewise.
13294 * ira-build.c (create_insn_allocnos): Likewise.
13295 * lra-coalesce.c (merge_pseudos): Likewise.
13296 * lra-constraints.c (match_reload): Likewise.
13297 (simplify_operand_subreg): Likewise.
13298 (curr_insn_transform): Likewise.
13299 * lra-lives.c (process_bb_lives): Likewise.
13300 * lra.c (new_insn_reg): Likewise.
13301 (lra_substitute_pseudo): Likewise.
13302 * regcprop.c (mode_change_ok): Likewise.
13303 (maybe_mode_change): Likewise.
13304 (copyprop_hardreg_forward_1): Likewise.
13305 * reload.c (push_reload): Likewise.
13306 (find_reloads): Likewise.
13307 (find_reloads_subreg_address): Likewise.
13308 * reload1.c (alter_reg): Likewise.
13309 (eliminate_regs_1): Likewise.
13310 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13312 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
13314 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
13315 back to if statements, including unpack.
13317 2017-08-30 Martin Liska <mliska@suse.cz>
13319 PR inline-asm/82001
13320 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
13322 (func_checker::compare_asm_inputs_outputs): ... this function.
13323 (func_checker::compare_gimple_asm): Use the function to compare
13324 also ASM constrains.
13325 * ipa-icf-gimple.h: Rename the function.
13327 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13328 Alan Hayward <alan.hayward@arm.com>
13329 David Sherwood <david.sherwood@arm.com>
13331 * coretypes.h (complex_mode): New type.
13332 * gdbhooks.py (build_pretty_printer): Handle it.
13333 * machmode.h (complex_mode): New class.
13334 (complex_mode::includes_p): New function.
13335 (is_complex_int_mode): Likewise.
13336 (is_complex_float_mode): Likewise.
13337 * genmodes.c (get_mode_class): Handle complex mode classes.
13338 * function.c (expand_function_end): Use is_complex_int_mode.
13340 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13341 Alan Hayward <alan.hayward@arm.com>
13342 David Sherwood <david.sherwood@arm.com>
13344 * coretypes.h (scalar_mode_pod): New typedef.
13345 * gdbhooks.py (build_pretty_printer): Handle it.
13346 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
13347 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
13348 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
13349 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
13350 as_a <scalar_mode>.
13352 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13353 Alan Hayward <alan.hayward@arm.com>
13354 David Sherwood <david.sherwood@arm.com>
13356 * machmode.h (mode_for_vector): Take a scalar_mode instead
13358 * stor-layout.c (mode_for_vector): Likewise.
13359 * explow.c (promote_mode): Use as_a <scalar_mode>.
13360 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
13362 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13363 Alan Hayward <alan.hayward@arm.com>
13364 David Sherwood <david.sherwood@arm.com>
13366 * target.def (preferred_simd_mode): Take a scalar_mode
13367 instead of a machine_mode.
13368 * targhooks.h (default_preferred_simd_mode): Likewise.
13369 * targhooks.c (default_preferred_simd_mode): Likewise.
13370 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
13371 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
13372 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
13373 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
13374 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
13375 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
13376 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
13377 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
13379 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
13380 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
13381 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
13382 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
13383 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
13384 * doc/tm.texi: Regenerate.
13385 * optabs-query.c (can_vec_mask_load_store_p): Return false for
13388 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13389 Alan Hayward <alan.hayward@arm.com>
13390 David Sherwood <david.sherwood@arm.com>
13392 * target.def (scalar_mode_supported_p): Take a scalar_mode
13393 instead of a machine_mode.
13394 * targhooks.h (default_scalar_mode_supported_p): Likewise.
13395 * targhooks.c (default_scalar_mode_supported_p): Likewise.
13396 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
13397 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13398 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
13399 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
13400 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
13401 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
13402 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
13403 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
13404 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
13405 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
13406 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
13407 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
13409 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
13410 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
13411 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13412 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13413 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
13415 * doc/tm.texi: Regenerate.
13417 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13418 Alan Hayward <alan.hayward@arm.com>
13419 David Sherwood <david.sherwood@arm.com>
13421 * coretypes.h (opt_scalar_mode): New typedef.
13422 * gdbhooks.py (build_pretty_printers): Handle it.
13423 * machmode.h (mode_iterator::get_2xwider): Add overload for
13425 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
13427 * expr.c (convert_mode_scalar): Likewise.
13428 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13429 * optabs.c (expand_float): Likewise.
13430 (expand_fix): Likewise.
13431 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13433 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13434 Alan Hayward <alan.hayward@arm.com>
13435 David Sherwood <david.sherwood@arm.com>
13437 * optabs.c (expand_float): Explicitly check for scalars before
13438 using a branching expansion.
13439 (expand_fix): Likewise.
13441 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13442 Alan Hayward <alan.hayward@arm.com>
13443 David Sherwood <david.sherwood@arm.com>
13445 * expr.c (convert_mode): Split scalar handling out into...
13446 (convert_mode_scalar): ...this new function. Treat the modes
13449 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13450 Alan Hayward <alan.hayward@arm.com>
13451 David Sherwood <david.sherwood@arm.com>
13453 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
13455 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
13457 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13458 Alan Hayward <alan.hayward@arm.com>
13459 David Sherwood <david.sherwood@arm.com>
13461 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
13462 rather than a machine_mode.
13463 (fixed_from_string): Likewise.
13464 (fixed_convert): Likewise.
13465 (fixed_convert_from_int): Likewise.
13466 (fixed_convert_from_real): Likewise.
13467 (real_convert_from_fixed): Likewise.
13468 * fixed-value.c (fixed_from_double_int): Likewise.
13469 (fixed_from_string): Likewise.
13470 (fixed_convert): Likewise.
13471 (fixed_convert_from_int): Likewise.
13472 (fixed_convert_from_real): Likewise.
13473 (real_convert_from_fixed): Likewise.
13474 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
13476 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13477 Alan Hayward <alan.hayward@arm.com>
13478 David Sherwood <david.sherwood@arm.com>
13480 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
13481 of separate mode class checks. Do not allow vector modes here.
13482 (immed_wide_int_const): Use as_a <scalar_mode>.
13483 * explow.c (trunc_int_for_mode): Likewise.
13484 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
13485 (wi::shwi): Likewise.
13486 (wi::min_value): Likewise.
13487 (wi::max_value): Likewise.
13488 * dwarf2out.c (loc_descriptor): Likewise.
13489 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
13490 for CONST_WIDE_INT.
13492 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13493 Alan Hayward <alan.hayward@arm.com>
13494 David Sherwood <david.sherwood@arm.com>
13496 * tree.h (SCALAR_TYPE_MODE): New macro.
13497 * expr.c (expand_expr_addr_expr_1): Use it.
13498 (expand_expr_real_2): Likewise.
13499 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
13500 (fold_convert_const_fixed_from_int): Likewise.
13501 (fold_convert_const_fixed_from_real): Likewise.
13502 (native_encode_fixed): Likewise
13503 (native_encode_complex): Likewise
13504 (native_encode_vector): Likewise.
13505 (native_interpret_fixed): Likewise.
13506 (native_interpret_real): Likewise.
13507 (native_interpret_complex): Likewise.
13508 (native_interpret_vector): Likewise.
13509 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
13510 (simd_clone_adjust_argument_types): Likewise.
13511 (simd_clone_init_simd_arrays): Likewise.
13512 (simd_clone_adjust): Likewise.
13513 * stor-layout.c (layout_type): Likewise.
13514 * tree.c (build_minus_one_cst): Likewise.
13515 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
13516 * tree-inline.c (estimate_move_cost): Likewise.
13517 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
13518 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
13519 (vectorizable_reduction): Likewise.
13520 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
13521 (vect_recog_mixed_size_cond_pattern): Likewise.
13522 (check_bool_pattern): Likewise.
13523 (adjust_bool_pattern): Likewise.
13524 (search_type_for_mask_1): Likewise.
13525 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
13526 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13527 (vectorizable_load): Likewise.
13528 (vectorizable_store): Likewise.
13529 * ubsan.c (ubsan_encode_value): Likewise.
13530 * varasm.c (output_constant): Likewise.
13532 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13533 Alan Hayward <alan.hayward@arm.com>
13534 David Sherwood <david.sherwood@arm.com>
13536 * coretypes.h (scalar_mode): New class.
13537 * machmode.h (scalar_mode): Likewise.
13538 (scalar_mode::includes_p): New function.
13539 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
13540 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
13541 * genmodes.c (get_mode_class): Handle remaining scalar modes.
13542 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
13543 * expmed.c (store_bit_field_1): Likewise.
13544 (extract_bit_field_1): Likewise.
13545 * expr.c (write_complex_part): Likewise.
13546 (read_complex_part): Likewise.
13547 (emit_move_complex_push): Likewise.
13548 (expand_expr_real_2): Likewise.
13549 * function.c (assign_parm_setup_reg): Likewise.
13550 (assign_parms_unsplit_complex): Likewise.
13551 * optabs.c (expand_binop): Likewise.
13552 * rtlanal.c (subreg_get_info): Likewise.
13553 * simplify-rtx.c (simplify_immed_subreg): Likewise.
13554 * varasm.c (output_constant_pool_2): Likewise.
13556 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13557 Alan Hayward <alan.hayward@arm.com>
13558 David Sherwood <david.sherwood@arm.com>
13560 * expmed.c (extract_high_half): Use scalar_int_mode and remove
13562 (expmed_mult_highpart_optab): Likewise.
13563 (expmed_mult_highpart): Likewise.
13565 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13566 Alan Hayward <alan.hayward@arm.com>
13567 David Sherwood <david.sherwood@arm.com>
13569 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
13570 instead of a machine_mode.
13571 (builtin_memset_read_str): Likewise.
13572 * builtins.c (c_readstr): Likewise.
13573 (builtin_memcpy_read_str): Likewise.
13574 (builtin_strncpy_read_str): Likewise.
13575 (builtin_memset_read_str): Likewise.
13576 (builtin_memset_gen_str): Likewise.
13577 (expand_builtin_signbit): Use scalar_int_mode for local variables.
13578 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
13579 instead of a machine_mode.
13580 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
13582 (make_extraction): Likewise.
13583 (try_widen_shift_mode): Take and return scalar_int_modes instead
13585 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
13586 a scalar_int_mode instead of a machine_mode.
13587 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
13588 (avr_addr_space_pointer_mode): Likewise.
13589 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
13590 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
13591 (msp430_unwind_word_mode): Likewise.
13592 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
13593 (spu_addr_space_pointer_mode): Likewise.
13594 (spu_addr_space_address_mode): Likewise.
13595 (spu_libgcc_cmp_return_mode): Likewise.
13596 (spu_libgcc_shift_count_mode): Likewise.
13597 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
13598 (rl78_addr_space_pointer_mode): Likewise.
13599 (fl78_unwind_word_mode): Likewise.
13600 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
13602 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
13603 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
13604 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
13605 (mips_valid_pointer_mode): Likewise.
13606 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
13607 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
13608 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
13610 (ft32_addr_space_address_mode): Likewise.
13611 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
13612 scalar_int_mode instead of a machine_mode.
13613 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
13615 (m32c_addr_space_address_mode): Likewise.
13616 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
13617 (rs6000_eh_return_filter_mode): Likewise.
13618 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
13619 (rs6000_eh_return_filter_mode): Likewise.
13620 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
13621 (s390_libgcc_shift_count_mode): Likewise.
13622 (s390_unwind_word_mode): Likewise.
13623 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
13625 * target.def (mode_rep_extended): Likewise.
13626 (valid_pointer_mode): Likewise.
13627 (addr_space.valid_pointer_mode): Likewise.
13628 (eh_return_filter_mode): Return a scalar_int_mode rather than
13630 (libgcc_cmp_return_mode): Likewise.
13631 (libgcc_shift_count_mode): Likewise.
13632 (unwind_word_mode): Likewise.
13633 (addr_space.pointer_mode): Likewise.
13634 (addr_space.address_mode): Likewise.
13635 * doc/tm.texi: Regenerate.
13636 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
13638 (do_jump): Use scalar_int_mode for local variables.
13639 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
13640 rather than a machine_mode.
13641 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
13642 (scompare_loc_descriptor_wide): Likewise.
13643 (scompare_loc_descriptor_narrow): Likewise.
13644 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
13646 * except.c (sjlj_emit_dispatch_table): Likewise.
13647 (expand_builtin_eh_copy_values): Likewise.
13648 * explow.c (convert_memory_address_addr_space_1): Likewise.
13649 Take a scalar_int_mode rather than a machine_mode.
13650 (convert_memory_address_addr_space): Take a scalar_int_mode rather
13651 than a machine_mode.
13652 (memory_address_addr_space): Use scalar_int_mode for local variables.
13653 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
13654 rather than a machine_mode.
13655 * expmed.c (mask_rtx): Likewise.
13656 (init_expmed_one_conv): Likewise.
13657 (expand_mult_highpart_adjust): Likewise.
13658 (extract_high_half): Likewise.
13659 (expmed_mult_highpart_optab): Likewise.
13660 (expmed_mult_highpart): Likewise.
13661 (expand_smod_pow2): Likewise.
13662 (expand_sdiv_pow2): Likewise.
13663 (emit_store_flag_int): Likewise.
13664 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
13666 (extract_low_bits): Likewise.
13667 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
13669 * expr.c (pieces_addr::adjust): Likewise.
13670 (can_store_by_pieces): Likewise.
13671 (store_by_pieces): Likewise.
13672 (clear_by_pieces_1): Likewise.
13673 (expand_expr_addr_expr_1): Likewise.
13674 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
13675 (expand_expr_real_1): Likewise.
13676 (try_casesi): Likewise.
13677 * final.c (shorten_branches): Likewise.
13678 * fold-const.c (fold_convert_const_int_from_fixed): Change the
13679 type of "mode" to machine_mode.
13680 * internal-fn.c (expand_arith_overflow_result_store): Take a
13681 scalar_int_mode rather than a machine_mode.
13682 (expand_mul_overflow): Use scalar_int_mode for local variables.
13683 * loop-doloop.c (doloop_modify): Likewise.
13684 (doloop_optimize): Likewise.
13685 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
13686 than a machine_mode.
13687 (expand_doubleword_shift_condmove): Likewise.
13688 (expand_doubleword_shift): Likewise.
13689 (expand_doubleword_clz): Likewise.
13690 (expand_doubleword_popcount): Likewise.
13691 (expand_doubleword_parity): Likewise.
13692 (expand_absneg_bit): Use scalar_int_mode for local variables.
13693 (prepare_float_lib_cmp): Likewise.
13694 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
13695 rather than a machine_mode.
13696 (convert_memory_address_addr_space): Likewise.
13697 (get_mode_bounds): Likewise.
13698 (get_address_mode): Return a scalar_int_mode rather than a
13700 * rtlanal.c (get_address_mode): Likewise.
13701 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
13702 than a machine_mode.
13703 * targhooks.c (default_mode_rep_extended): Likewise.
13704 (default_valid_pointer_mode): Likewise.
13705 (default_addr_space_valid_pointer_mode): Likewise.
13706 (default_eh_return_filter_mode): Return a scalar_int_mode rather
13707 than a machine_mode.
13708 (default_libgcc_cmp_return_mode): Likewise.
13709 (default_libgcc_shift_count_mode): Likewise.
13710 (default_unwind_word_mode): Likewise.
13711 (default_addr_space_pointer_mode): Likewise.
13712 (default_addr_space_address_mode): Likewise.
13713 * targhooks.h (default_eh_return_filter_mode): Likewise.
13714 (default_libgcc_cmp_return_mode): Likewise.
13715 (default_libgcc_shift_count_mode): Likewise.
13716 (default_unwind_word_mode): Likewise.
13717 (default_addr_space_pointer_mode): Likewise.
13718 (default_addr_space_address_mode): Likewise.
13719 (default_mode_rep_extended): Take a scalar_int_mode rather than
13721 (default_valid_pointer_mode): Likewise.
13722 (default_addr_space_valid_pointer_mode): Likewise.
13723 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
13725 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
13726 rather than a machine_mode.
13727 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
13728 for local variables.
13729 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13730 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
13731 than a machine_mode.
13733 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13734 Alan Hayward <alan.hayward@arm.com>
13735 David Sherwood <david.sherwood@arm.com>
13737 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
13738 the mode argument to scalar_int_mode.
13739 (do_jump_by_parts_zero_rtx): Likewise.
13740 (do_jump_by_parts_equality_rtx): Likewise.
13741 (do_jump_by_parts_greater): Take a mode argument.
13742 (do_jump_by_parts_equality): Likewise.
13743 (do_jump_1): Update calls accordingly.
13745 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13746 Alan Hayward <alan.hayward@arm.com>
13747 David Sherwood <david.sherwood@arm.com>
13749 * is-a.h (safe_dyn_cast): New function.
13750 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
13751 (jump_table_for_label): Likewise.
13752 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
13753 instead of an rtx_insn *.
13754 (shorten_branches): Use dyn_cast instead of LABEL_P and
13755 JUMP_TABLE_DATA_P. Use jump_table_for_label and
13756 rtx_jump_table_data::get_data_mode.
13757 (final_scan_insn): Likewise.
13759 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13760 Alan Hayward <alan.hayward@arm.com>
13761 David Sherwood <david.sherwood@arm.com>
13763 * combine.c (try_combine): Use is_a <scalar_int_mode> when
13764 trying to combine a full-register integer set with a subreg
13767 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13768 Alan Hayward <alan.hayward@arm.com>
13769 David Sherwood <david.sherwood@arm.com>
13771 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
13772 that is always either address_mode or pointer_mode.
13774 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13775 Alan Hayward <alan.hayward@arm.com>
13776 David Sherwood <david.sherwood@arm.com>
13778 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
13779 when the two are known to be equal.
13781 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13782 Alan Hayward <alan.hayward@arm.com>
13783 David Sherwood <david.sherwood@arm.com>
13785 * simplify-rtx.c (simplify_const_unary_operation): Use
13786 is_a <scalar_int_mode> instead of checking for a nonzero
13787 precision. Forcibly convert op_mode to a scalar_int_mode
13788 in that case. More clearly differentiate the operand and
13789 result modes and use the former when deciding what the value
13790 of a count-bits operation should be. Use is_int_mode instead
13791 of checking for a MODE_INT. Remove redundant check for whether
13792 this mode has a zero precision.
13794 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13795 Alan Hayward <alan.hayward@arm.com>
13796 David Sherwood <david.sherwood@arm.com>
13798 * optabs.c (widen_leading): Change the type of the mode argument
13799 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
13800 (widen_bswap): Likewise.
13801 (expand_parity): Likewise.
13802 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
13803 (expand_ffs): Likewise.
13804 (epand_unop): Check for scalar integer modes before calling the
13807 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13808 Alan Hayward <alan.hayward@arm.com>
13809 David Sherwood <david.sherwood@arm.com>
13811 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
13813 (expand_expr_real_1): Update call accordingly.
13815 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13816 Alan Hayward <alan.hayward@arm.com>
13817 David Sherwood <david.sherwood@arm.com>
13819 * expmed.c (store_bit_field_using_insv): Add op0_mode and
13820 value_mode arguments. Use scalar_int_mode internally.
13821 (store_bit_field_1): Rename the new integer mode from imode
13822 to op0_mode and use it instead of GET_MODE (op0). Update calls
13823 to store_split_bit_field, store_bit_field_using_insv and
13824 store_fixed_bit_field.
13825 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
13826 Use scalar_int_mode internally. Use a bit count rather than a mode
13827 when calculating the largest bit size for get_best_mode.
13828 Update calls to store_split_bit_field and store_fixed_bit_field_1.
13829 (store_fixed_bit_field_1): Add mode and value_mode arguments.
13830 Remove assertion that OP0 has a scalar integer mode.
13831 (store_split_bit_field): Add op0_mode and value_mode arguments.
13832 Update calls to extract_fixed_bit_field.
13833 (extract_bit_field_using_extv): Add an op0_mode argument.
13834 Use scalar_int_mode internally.
13835 (extract_bit_field_1): Rename the new integer mode from imode to
13836 op0_mode and use it instead of GET_MODE (op0). Update calls to
13837 extract_split_bit_field, extract_bit_field_using_extv and
13838 extract_fixed_bit_field.
13839 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
13840 to extract_split_bit_field and extract_fixed_bit_field_1.
13841 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
13842 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
13843 on the target mode.
13844 (extract_split_bit_field): Add an op0_mode argument. Update call
13845 to extract_fixed_bit_field.
13847 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13848 Alan Hayward <alan.hayward@arm.com>
13849 David Sherwood <david.sherwood@arm.com>
13851 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
13852 * explow.c (hard_function_value): Likewise.
13853 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
13854 convert_to_mode call outside the loop.
13855 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
13856 for the mode iterator. Require the mode specified by max_pieces
13858 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
13860 (copy_blkmode_to_reg): Likewise.
13861 (set_storage_via_setmem): Likewise.
13862 * optabs.c (prepare_cmp_insn): Likewise.
13863 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13864 * stor-layout.c (finish_bitfield_representative): Likewise.
13866 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13867 Alan Hayward <alan.hayward@arm.com>
13868 David Sherwood <david.sherwood@arm.com>
13870 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
13871 * expr.c (convert_move): Use them.
13872 (convert_modes): Likewise.
13873 (store_expr_with_bounds): Likewise.
13875 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13876 Alan Hayward <alan.hayward@arm.com>
13877 David Sherwood <david.sherwood@arm.com>
13879 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
13880 parameter for the mode of "x". Remove the "known_x", "known_mode"
13881 and "known_ret" arguments. Change the type of the mode argument
13882 to scalar_int_mode.
13883 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
13884 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
13885 (reg_num_sign_bit_copies_for_combine): Likewise.
13886 * rtlanal.c (nonzero_bits1): Likewise.
13887 (num_sign_bit_copies1): Likewise.
13888 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
13889 (reg_num_sign_bit_copies_general): Likewise.
13890 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
13891 (reg_nonzero_bits_general): Likewise.
13893 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13894 Alan Hayward <alan.hayward@arm.com>
13895 David Sherwood <david.sherwood@arm.com>
13897 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
13898 than in subroutines. Return 1 for non-integer modes.
13899 (cached_num_sign_bit_copies): Change the type of the mode parameter
13900 to scalar_int_mode.
13901 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
13902 classes. Handle CONST_INT_P first and then check whether X also
13903 has a scalar integer mode. Check the same thing for inner registers
13904 of a SUBREG and for values that are being extended or truncated.
13906 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13907 Alan Hayward <alan.hayward@arm.com>
13908 David Sherwood <david.sherwood@arm.com>
13910 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
13911 in subroutines. Return the mode mask for non-integer modes.
13912 (cached_nonzero_bits): Change the type of the mode parameter
13913 to scalar_int_mode.
13914 (nonzero_bits1): Likewise. Remove early exit for other mode
13915 classes. Handle CONST_INT_P first and then check whether X
13916 also has a scalar integer mode.
13918 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13919 Alan Hayward <alan.hayward@arm.com>
13920 David Sherwood <david.sherwood@arm.com>
13922 * expr.c (widest_int_mode_for_size): Make the comment match the code.
13923 Return a scalar_int_mode and assert that the size is greater than
13925 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
13926 (op_by_pieces_d::op_by_pieces_d): Likewise.
13927 (op_by_pieces_d::run): Likewise.
13928 (can_store_by_pieces): Likewise.
13930 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13931 Alan Hayward <alan.hayward@arm.com>
13932 David Sherwood <david.sherwood@arm.com>
13934 * combine.c (extract_left_shift): Add a mode argument and update
13936 (make_compound_operation_int): Change the type of the mode parameter
13937 to scalar_int_mode and update the call to extract_left_shift.
13939 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13940 Alan Hayward <alan.hayward@arm.com>
13941 David Sherwood <david.sherwood@arm.com>
13943 * combine.c (simplify_and_const_int): Change the type of the mode
13944 parameter to scalar_int_mode.
13945 (simplify_and_const_int_1): Likewise. Update recursive call.
13947 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13948 Alan Hayward <alan.hayward@arm.com>
13949 David Sherwood <david.sherwood@arm.com>
13951 * combine.c (simplify_compare_const): Check that the mode is a
13952 scalar_int_mode (rather than VOIDmode) before testing its
13954 (simplify_comparison): Move COMPARISON_P handling out of the
13955 loop and restrict the latter part of the loop to scalar_int_modes.
13956 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
13957 and when considering SUBREG_REGs. Use is_int_mode instead of
13958 checking GET_MODE_CLASS against MODE_INT.
13960 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13961 Alan Hayward <alan.hayward@arm.com>
13962 David Sherwood <david.sherwood@arm.com>
13964 * combine.c (try_widen_shift_mode): Move check for equal modes to...
13965 (simplify_shift_const_1): ...here. Use scalar_int_mode for
13966 shift_unit_mode and for modes involved in scalar shifts.
13968 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13969 Alan Hayward <alan.hayward@arm.com>
13970 David Sherwood <david.sherwood@arm.com>
13972 * combine.c (force_int_to_mode): New function, split out from...
13973 (force_to_mode): ...here. Keep xmode up-to-date and use it
13974 instead of GET_MODE (x).
13976 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13977 Alan Hayward <alan.hayward@arm.com>
13978 David Sherwood <david.sherwood@arm.com>
13980 * optabs-query.h (extraction_insn::struct_mode): Change type to
13981 opt_scalar_int_mode and update comment.
13982 (extraction_insn::field_mode): Change type to scalar_int_mode.
13983 (extraction_insn::pos_mode): Likewise.
13984 * combine.c (make_extraction): Update accordingly.
13985 * optabs-query.c (get_traditional_extraction_insn): Likewise.
13986 (get_optab_extraction_insn): Likewise.
13987 * recog.c (simplify_while_replacing): Likewise.
13988 * expmed.c (narrow_bit_field_mem): Change the type of the mode
13989 parameter to opt_scalar_int_mode.
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 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
13996 to a scalar_int_mode instead of a machine_mode.
13997 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
13998 (get_best_mode): Return a boolean and use a pointer argument to store
13999 the selected mode. Replace the limit mode parameter with a bit limit.
14000 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
14001 for the values returned by bit_field_mode_iterator::next_mode.
14002 (store_bit_field): Update call to get_best_mode.
14003 (store_fixed_bit_field): Likewise.
14004 (extract_fixed_bit_field): Likewise.
14005 * expr.c (optimize_bitfield_assignment_op): Likewise.
14006 * fold-const.c (optimize_bit_field_compare): Likewise.
14007 (fold_truth_andor_1): Likewise.
14008 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
14009 Update for new type of m_mode.
14010 (get_best_mode): As above.
14012 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14013 Alan Hayward <alan.hayward@arm.com>
14014 David Sherwood <david.sherwood@arm.com>
14016 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
14017 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
14018 (store_bit_field): Check is_a <scalar_int_mode> before calling
14019 strict_volatile_bitfield_p.
14020 (extract_bit_field): Likewise.
14022 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14023 Alan Hayward <alan.hayward@arm.com>
14024 David Sherwood <david.sherwood@arm.com>
14026 * target.def (cstore_mode): Return a scalar_int_mode.
14027 * doc/tm.texi: Regenerate.
14028 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
14029 * targhooks.h (default_cstore_mode): Likewise.
14030 * targhooks.c (default_cstore_mode): Likewise, using a forced
14032 * expmed.c (emit_cstore): Expect the target of the cstore to be
14035 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14036 Alan Hayward <alan.hayward@arm.com>
14037 David Sherwood <david.sherwood@arm.com>
14039 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
14041 (niter_desc): Likewise mode.
14042 (iv_analyze): Add a mode parameter.
14044 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
14045 and change its type to scalar_int_mode.
14046 * loop-iv.c: Update commentary at head of file.
14047 (iv_constant): Pass the mode paraeter before the rtx it describes
14048 and change its type to scalar_int_mode. Remove VOIDmode handling.
14049 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
14050 (iv_extend): Likewise.
14051 (shorten_into_mode): Likewise.
14052 (iv_add): Use scalar_int_mode.
14053 (iv_mult): Likewise.
14054 (iv_shift): Likewise.
14055 (canonicalize_iv_subregs): Likewise.
14056 (get_biv_step_1): Pass the outer_mode parameter before the rtx
14057 it describes and change its mode to scalar_int_mode. Also change
14058 the type of the returned inner_mode to scalar_int_mode.
14059 (get_biv_step): Likewise, turning outer_mode from a pointer
14060 into a direct parameter. Update call to get_biv_step_1.
14061 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
14062 iv_constant and get_biv_step.
14063 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
14064 and change its type to scalar_int_mode. Don't initialise iv->mode
14065 to VOIDmode and remove later checks for its still being VOIDmode.
14066 Update calls to iv_analyze_op and iv_analyze_expr. Check
14067 is_a <scalar_int_mode> when changing the mode under consideration.
14068 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
14069 Update call to iv_analyze_expr.
14070 (iv_analyze_op): Add a mode parameter. Reject subregs whose
14071 inner register is not also a scalar_int_mode. Update call to
14073 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
14074 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
14075 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
14076 separate mode class checks. Update calls to iv_analyze. Remove
14077 fix-up of VOIDmodes after iv_analyze_biv.
14078 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
14079 don't have a scalar_int_mode. Update call to biv_p.
14081 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14082 Alan Hayward <alan.hayward@arm.com>
14083 David Sherwood <david.sherwood@arm.com>
14085 * cfgexpand.c (convert_debug_memory_address): Use
14086 as_a <scalar_int_mode>.
14087 * combine.c (expand_compound_operation): Likewise.
14088 (make_extraction): Likewise.
14089 (change_zero_ext): Likewise.
14090 (simplify_comparison): Likewise.
14091 * cse.c (cse_insn): Likewise.
14092 * dwarf2out.c (minmax_loc_descriptor): Likewise.
14093 (mem_loc_descriptor): Likewise.
14094 (loc_descriptor): Likewise.
14095 * expmed.c (init_expmed_one_mode): Likewise.
14096 (synth_mult): Likewise.
14097 (emit_store_flag_1): Likewise.
14098 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
14099 of a comparison with size.
14100 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
14101 (reduce_to_bit_field_precision): Likewise.
14102 * function.c (expand_function_end): Likewise.
14103 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
14104 * loop-doloop.c (doloop_modify): Likewise.
14105 * optabs.c (expand_binop): Likewise.
14106 (expand_unop): Likewise.
14107 (expand_copysign_absneg): Likewise.
14108 (prepare_cmp_insn): Likewise.
14109 (maybe_legitimize_operand): Likewise.
14110 * recog.c (const_scalar_int_operand): Likewise.
14111 * rtlanal.c (get_address_mode): Likewise.
14112 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14113 (simplify_cond_clz_ctz): Likewise.
14114 * tree-nested.c (get_nl_goto_field): Likewise.
14115 * tree.c (build_vector_type_for_mode): Likewise.
14116 * var-tracking.c (use_narrower_mode): Likewise.
14118 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14119 Alan Hayward <alan.hayward@arm.com>
14120 David Sherwood <david.sherwood@arm.com>
14122 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
14123 * builtins.c (expand_builtin_signbit): Use it.
14124 * cfgexpand.c (expand_debug_expr): Likewise.
14125 * dojump.c (do_jump): Likewise.
14126 (do_compare_and_jump): Likewise.
14127 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
14128 * expmed.c (make_tree): Likewise.
14129 * expr.c (expand_expr_real_2): Likewise.
14130 (expand_expr_real_1): Likewise.
14131 (try_casesi): Likewise.
14132 * fold-const-call.c (fold_const_call_ss): Likewise.
14133 * fold-const.c (unextend): Likewise.
14134 (extract_muldiv_1): Likewise.
14135 (fold_single_bit_test): Likewise.
14136 (native_encode_int): Likewise.
14137 (native_encode_string): Likewise.
14138 (native_interpret_int): Likewise.
14139 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
14140 * internal-fn.c (expand_addsub_overflow): Likewise.
14141 (expand_neg_overflow): Likewise.
14142 (expand_mul_overflow): Likewise.
14143 (expand_arith_overflow): Likewise.
14144 * match.pd: Likewise.
14145 * stor-layout.c (layout_type): Likewise.
14146 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
14147 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
14148 * tree-ssanames.c (get_range_info): Likewise.
14149 * tree-switch-conversion.c (array_value_type) Likewise.
14150 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
14151 (vect_recog_divmod_pattern): Likewise.
14152 (vect_recog_mixed_size_cond_pattern): Likewise.
14153 * tree-vrp.c (extract_range_basic): Likewise.
14154 (simplify_float_conversion_using_ranges): Likewise.
14155 * tree.c (int_fits_type_p): Likewise.
14156 * ubsan.c (instrument_bool_enum_load): Likewise.
14157 * varasm.c (mergeable_string_section): Likewise.
14158 (narrowing_initializer_constant_valid_p): Likewise.
14159 (output_constant): Likewise.
14161 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14162 Alan Hayward <alan.hayward@arm.com>
14163 David Sherwood <david.sherwood@arm.com>
14165 * machmode.h (NARROWEST_INT_MODE): New macro.
14166 * expr.c (alignment_for_piecewise_move): Use it instead of
14167 GET_CLASS_NARROWEST_MODE (MODE_INT).
14168 (push_block): Likewise.
14169 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
14171 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14173 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14174 Alan Hayward <alan.hayward@arm.com>
14175 David Sherwood <david.sherwood@arm.com>
14177 * postreload.c (move2add_valid_value_p): Change the type of the
14178 mode parameter to scalar_int_mode.
14179 (move2add_use_add2_insn): Add a mode parameter and use it instead
14181 (move2add_use_add3_insn): Likewise.
14182 (reload_cse_move2add): Update accordingly.
14184 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14185 Alan Hayward <alan.hayward@arm.com>
14186 David Sherwood <david.sherwood@arm.com>
14188 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
14190 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
14191 * optabs.c (expand_unop): Likewise.
14193 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14194 Alan Hayward <alan.hayward@arm.com>
14195 David Sherwood <david.sherwood@arm.com>
14197 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
14198 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
14199 (popcount_loc_descriptor): Likewise.
14200 (bswap_loc_descriptor): Likewise.
14201 (rotate_loc_descriptor): Likewise.
14202 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
14203 calling the functions above.
14205 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14206 Alan Hayward <alan.hayward@arm.com>
14207 David Sherwood <david.sherwood@arm.com>
14209 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
14211 (try_combine): Likewise.
14212 (simplify_if_then_else): Likewise.
14213 * cse.c (cse_insn): Likewise.
14214 * dwarf2out.c (mem_loc_descriptor): Likewise.
14215 * emit-rtl.c (gen_lowpart_common): Likewise.
14216 * simplify-rtx.c (simplify_truncation): Likewise.
14217 (simplify_binary_operation_1): Likewise.
14218 (simplify_const_relational_operation): Likewise.
14219 (simplify_ternary_operation): Likewise.
14220 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
14222 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14223 Alan Hayward <alan.hayward@arm.com>
14224 David Sherwood <david.sherwood@arm.com>
14226 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
14227 * reload.c (push_reload): Likewise.
14228 (find_reloads): Likewise.
14230 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14231 Alan Hayward <alan.hayward@arm.com>
14232 David Sherwood <david.sherwood@arm.com>
14234 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
14235 (make_compound_operation_int): Likewise.
14236 (change_zero_ext): Likewise.
14237 * expr.c (convert_move): Likewise.
14238 (convert_modes): Likewise.
14239 * fwprop.c (forward_propagate_subreg): Likewise.
14240 * loop-iv.c (get_biv_step_1): Likewise.
14241 * optabs.c (widen_operand): Likewise.
14242 * postreload.c (move2add_valid_value_p): Likewise.
14243 * recog.c (simplify_while_replacing): Likewise.
14244 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14245 (simplify_binary_operation_1): Likewise. Remove redundant
14246 mode equality check.
14248 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14249 Alan Hayward <alan.hayward@arm.com>
14250 David Sherwood <david.sherwood@arm.com>
14252 * combine.c (combine_simplify_rtx): Add checks for
14253 is_a <scalar_int_mode>.
14254 (simplify_if_then_else): Likewise.
14255 (make_field_assignment): Likewise.
14256 (simplify_comparison): Likewise.
14257 * ifcvt.c (noce_try_bitop): Likewise.
14258 * loop-invariant.c (canonicalize_address_mult): Likewise.
14259 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14261 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14262 Alan Hayward <alan.hayward@arm.com>
14263 David Sherwood <david.sherwood@arm.com>
14265 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
14266 is_a <scalar_int_mode> instead of != BLKmode.
14268 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14269 Alan Hayward <alan.hayward@arm.com>
14270 David Sherwood <david.sherwood@arm.com>
14272 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
14273 instead of != VOIDmode.
14274 * combine.c (if_then_else_cond): Likewise.
14275 (change_zero_ext): Likewise.
14276 * dwarf2out.c (mem_loc_descriptor): Likewise.
14277 (loc_descriptor): Likewise.
14278 * rtlanal.c (canonicalize_condition): Likewise.
14279 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
14281 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14282 Alan Hayward <alan.hayward@arm.com>
14283 David Sherwood <david.sherwood@arm.com>
14285 * simplify-rtx.c (simplify_binary_operation_1): Use
14286 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
14288 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14289 Alan Hayward <alan.hayward@arm.com>
14290 David Sherwood <david.sherwood@arm.com>
14292 * wide-int.h (int_traits<unsigned char>) New class.
14293 (int_traits<unsigned short>) Likewise.
14294 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
14295 Use GET_MODE_UNIT_PRECISION and remove redundant test for
14297 * combine.c (set_nonzero_bits_and_sign_copies): Use
14298 is_a <scalar_int_mode>.
14299 (find_split_point): Likewise.
14300 (combine_simplify_rtx): Likewise.
14301 (simplify_logical): Likewise.
14302 (expand_compound_operation): Likewise.
14303 (expand_field_assignment): Likewise.
14304 (make_compound_operation): Likewise.
14305 (extended_count): Likewise.
14306 (change_zero_ext): Likewise.
14307 (simplify_comparison): Likewise.
14308 * dwarf2out.c (scompare_loc_descriptor): Likewise.
14309 (ucompare_loc_descriptor): Likewise.
14310 (minmax_loc_descriptor): Likewise.
14311 (mem_loc_descriptor): Likewise.
14312 (loc_descriptor): Likewise.
14313 * expmed.c (init_expmed_one_mode): Likewise.
14314 * lra-constraints.c (lra_constraint_offset): Likewise.
14315 * optabs.c (prepare_libcall_arg): Likewise.
14316 * postreload.c (move2add_note_store): Likewise.
14317 * reload.c (operands_match_p): Likewise.
14318 * rtl.h (load_extend_op): Likewise.
14319 * rtlhooks.c (gen_lowpart_general): Likewise.
14320 * simplify-rtx.c (simplify_truncation): Likewise.
14321 (simplify_unary_operation_1): Likewise.
14322 (simplify_binary_operation_1): Likewise.
14323 (simplify_const_binary_operation): Likewise.
14324 (simplify_const_relational_operation): Likewise.
14325 (simplify_subreg): Likewise.
14326 * stor-layout.c (bitwise_mode_for_mode): Likewise.
14327 * var-tracking.c (adjust_mems): Likewise.
14328 (prepare_call_arguments): Likewise.
14330 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14331 Alan Hayward <alan.hayward@arm.com>
14332 David Sherwood <david.sherwood@arm.com>
14334 * machmode.h (is_int_mode): New fuction.
14335 * combine.c (find_split_point): Use it.
14336 (combine_simplify_rtx): Likewise.
14337 (simplify_if_then_else): Likewise.
14338 (simplify_set): Likewise.
14339 (simplify_shift_const_1): Likewise.
14340 (simplify_comparison): Likewise.
14341 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
14342 * cse.c (notreg_cost): Likewise.
14343 (cse_insn): Likewise.
14344 * cselib.c (cselib_lookup_1): Likewise.
14345 * dojump.c (do_jump_1): Likewise.
14346 (do_compare_rtx_and_jump): Likewise.
14347 * dse.c (get_call_args): Likewise.
14348 * dwarf2out.c (rtl_for_decl_init): Likewise.
14349 (native_encode_initializer): Likewise.
14350 * expmed.c (emit_store_flag_1): Likewise.
14351 (emit_store_flag): Likewise.
14352 * expr.c (convert_modes): Likewise.
14353 (store_field): Likewise.
14354 (expand_expr_real_1): Likewise.
14355 * fold-const.c (fold_read_from_constant_string): Likewise.
14356 * gimple-ssa-sprintf.c (get_format_string): Likewise.
14357 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
14358 * optabs.c (expand_binop): Likewise.
14359 (expand_unop): Likewise.
14360 (expand_abs_nojump): Likewise.
14361 (expand_one_cmpl_abs_nojump): Likewise.
14362 * simplify-rtx.c (mode_signbit_p): Likewise.
14363 (val_signbit_p): Likewise.
14364 (val_signbit_known_set_p): Likewise.
14365 (val_signbit_known_clear_p): Likewise.
14366 (simplify_relational_operation_1): Likewise.
14367 * tree.c (vector_type_mode): Likewise.
14369 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14370 Alan Hayward <alan.hayward@arm.com>
14371 David Sherwood <david.sherwood@arm.com>
14373 * machmode.h (smallest_mode_for_size): Fix formatting.
14374 (smallest_int_mode_for_size): New function.
14375 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
14376 instead of smallest_mode_for_size.
14377 * combine.c (make_extraction): Likewise.
14378 * config/arc/arc.c (arc_expand_movmem): Likewise.
14379 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
14380 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
14381 * config/s390/s390.c (s390_expand_insv): Likewise.
14382 * config/sparc/sparc.c (assign_int_registers): Likewise.
14383 * config/spu/spu.c (spu_function_value): Likewise.
14384 (spu_function_arg): Likewise.
14385 * coverage.c (get_gcov_type): Likewise.
14386 (get_gcov_unsigned_t): Likewise.
14387 * dse.c (find_shift_sequence): Likewise.
14388 * expmed.c (store_bit_field_1): Likewise.
14389 * expr.c (convert_move): Likewise.
14390 (store_field): Likewise.
14391 * internal-fn.c (expand_arith_overflow): Likewise.
14392 * optabs-query.c (get_best_extraction_insn): Likewise.
14393 * optabs.c (expand_twoval_binop_libfunc): Likewise.
14394 * stor-layout.c (layout_type): Likewise.
14395 (initialize_sizetypes): Likewise.
14396 * targhooks.c (default_get_mask_mode): Likewise.
14397 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
14399 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14400 Alan Hayward <alan.hayward@arm.com>
14401 David Sherwood <david.sherwood@arm.com>
14403 * machmode.h (opt_mode::else_blk): New function.
14404 (int_mode_for_mode): Declare.
14405 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
14406 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
14408 * cfgexpand.c (expand_debug_expr): Likewise.
14409 * combine.c (gen_lowpart_or_truncate): Likewise.
14410 (gen_lowpart_for_combine): Likewise.
14411 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
14412 * config/avr/avr.c (avr_to_int_mode): Likewise.
14413 (avr_out_plus_1): Likewise.
14414 (avr_out_plus): Likewise.
14415 (avr_out_round): Likewise.
14416 * config/i386/i386.c (ix86_split_to_parts): Likewise.
14417 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
14418 (s390_expand_vcond): Likewise.
14419 * config/spu/spu.c (spu_split_immediate): Likewise.
14420 (spu_expand_mov): Likewise.
14421 * dse.c (get_stored_val): Likewise.
14422 * expmed.c (store_bit_field_1): Likewise.
14423 (convert_extracted_bit_field): Use int_mode_for_mode instead of
14425 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
14426 (extract_low_bits): Likewise.
14427 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
14428 handling rather than repeating the check.
14429 (emit_group_store): Likewise.
14430 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
14431 * optabs.c (expand_absneg_bit): Likewise.
14432 (expand_copysign_absneg): Likewise.
14433 (expand_copysign_bit): Likewise.
14434 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
14435 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
14436 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
14437 * var-tracking.c (prepare_call_arguments): Likewise.
14438 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
14439 int_mode_for_mode instead of mode_for_size.
14440 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
14442 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14443 Alan Hayward <alan.hayward@arm.com>
14444 David Sherwood <david.sherwood@arm.com>
14446 * machmode.h (int_mode_for_size): New function.
14447 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
14448 instead of mode_for_size.
14449 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
14451 * combine.c (expand_field_assignment): Use int_mode_for_size
14452 instead of mode_for_size.
14453 (make_extraction): Likewise.
14454 (simplify_shift_const_1): Likewise.
14455 (simplify_comparison): Likewise.
14456 * dojump.c (do_jump): Likewise.
14457 * dwarf2out.c (mem_loc_descriptor): Likewise.
14458 * emit-rtl.c (init_derived_machine_modes): Likewise.
14459 * expmed.c (flip_storage_order): Likewise.
14460 (convert_extracted_bit_field): Likewise.
14461 * expr.c (copy_blkmode_from_reg): Likewise.
14462 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
14463 * internal-fn.c (expand_mul_overflow): Likewise.
14464 * lower-subreg.c (simple_move): Likewise.
14465 * optabs-libfuncs.c (init_optabs): Likewise.
14466 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14467 * tree.c (vector_type_mode): Likewise.
14468 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
14469 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
14470 * tree-vect-generic.c (expand_vector_parallel): Likewise.
14471 * tree-vect-stmts.c (vectorizable_load): Likewise.
14472 (vectorizable_store): Likewise.
14474 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14475 Alan Hayward <alan.hayward@arm.com>
14476 David Sherwood <david.sherwood@arm.com>
14478 * coretypes.h (pod_mode): New type.
14479 (scalar_int_mode_pod): New typedef.
14480 * machmode.h (pod_mode): New class.
14481 (int_n_data_t::m): Change type to scalar_int_mode_pod.
14482 * genmodes.c (emit_mode_int_n): Update accordingly.
14483 * lower-subreg.h (target_lower_subreg): Change type to
14484 scalar_int_mode_pod.
14485 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
14486 scalar_int_mode_pod.
14488 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14489 Alan Hayward <alan.hayward@arm.com>
14490 David Sherwood <david.sherwood@arm.com>
14492 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
14493 machine_mode to scalar_int_mode.
14494 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
14495 (rs6000_option_override_internal): Remove cast to int.
14496 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
14497 machine_mode to scalar_int_mode.
14498 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
14499 (rs6000_option_override_internal): Remove cast to int.
14500 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
14501 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
14503 * config/s390/s390.c (s390_expand_builtin): Likewise.
14504 * coretypes.h (scalar_int_mode): New type.
14505 (opt_scalar_int_mode): New typedef.
14506 * machmode.h (scalar_int_mode): New class.
14507 (scalar_int_mode::includes_p): New function.
14508 (byte_mode): Change type to scalar_int_mode.
14509 (word_mode): Likewise.
14510 (ptr_mode): Likewise.
14511 * emit-rtl.c (byte_mode): Likewise.
14512 (word_mode): Likewise.
14513 (ptr_mode): Likewise.
14514 (init_derived_machine_modes): Update accordingly.
14515 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
14516 and MODE_PARTIAL_INT.
14517 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
14518 opt_scalar_int_mode.
14520 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14521 Alan Hayward <alan.hayward@arm.com>
14522 David Sherwood <david.sherwood@arm.com>
14524 * target.def (libgcc_floating_mode_supported_p): Take a
14526 * doc/tm.texi: Regenerate.
14527 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
14529 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
14530 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
14533 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14534 Alan Hayward <alan.hayward@arm.com>
14535 David Sherwood <david.sherwood@arm.com>
14537 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
14538 * doc/tm.texi: Regenerate.
14539 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
14540 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
14541 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
14542 * targhooks.h (default_floatn_mode): Likewise.
14543 * targhooks.c (default_floatn_mode): Likewise.
14544 * tree.c (build_common_tree_nodes): Update accordingly.
14546 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14547 Alan Hayward <alan.hayward@arm.com>
14548 David Sherwood <david.sherwood@arm.com>
14550 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
14551 (mode_iterator::iterate_p): Likewise.
14552 (mode_iterator::get_wider): Likewise.
14553 * expr.c (init_expr_target): Use opt_scalar_float_mode.
14555 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14556 Alan Hayward <alan.hayward@arm.com>
14557 David Sherwood <david.sherwood@arm.com>
14559 * coretypes.h (opt_scalar_float_mode): New typedef.
14560 * machmode.h (float_mode_for_size): New function.
14561 * emit-rtl.c (double_mode): Delete.
14562 (init_emit_once): Use float_mode_for_size.
14563 * stor-layout.c (layout_type): Likewise.
14564 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
14566 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14567 Alan Hayward <alan.hayward@arm.com>
14568 David Sherwood <david.sherwood@arm.com>
14570 * output.h (assemble_real): Take a scalar_float_mode.
14571 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
14572 * config/arm/arm.md (consttable_4): Likewise.
14573 (consttable_8): Likewise.
14574 (consttable_16): Likewise.
14575 * config/mips/mips.md (consttable_float): Likewise.
14576 * config/s390/s390.c (s390_output_pool_entry): Likewise.
14577 * varasm.c (assemble_real): Take a scalar_float_mode.
14578 (output_constant_pool_2): Update accordingly.
14579 (output_constant): Likewise.
14581 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14582 Alan Hayward <alan.hayward@arm.com>
14583 David Sherwood <david.sherwood@arm.com>
14585 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
14586 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
14587 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
14588 (native_encode_real): Likewise.
14589 (native_interpret_real): Likewise.
14590 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
14591 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14593 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14594 Alan Hayward <alan.hayward@arm.com>
14595 David Sherwood <david.sherwood@arm.com>
14597 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
14598 <scalar_float_mode>. Simplify.
14599 (gen_extend_conv_libfunc): Likewise.
14601 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14602 Alan Hayward <alan.hayward@arm.com>
14603 David Sherwood <david.sherwood@arm.com>
14605 * coretypes.h (scalar_float_mode): New type.
14606 * machmode.h (mode_traits::from_int): Use machine_mode if
14607 USE_ENUM_MODES is defined.
14608 (is_a): New function.
14610 (dyn_cast): Likewise.
14611 (scalar_float_mode): New class.
14612 (scalar_float_mode::includes_p): New function.
14613 (is_float_mode): Likewise.
14614 * gdbhooks.py (MachineModePrinter): New class.
14615 (build_pretty_printer): Use it for scalar_float_mode.
14616 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
14617 (format_helper::format_helper): Turn into a template.
14618 * genmodes.c (get_mode_class): New function.
14619 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
14620 or machine_mode if none.
14621 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
14622 as_a <scalar_float_mode>.
14623 * dwarf2out.c (mem_loc_descriptor): Likewise.
14624 (insert_float): Likewise.
14625 (add_const_value_attribute): Likewise.
14626 * simplify-rtx.c (simplify_immed_subreg): Likewise.
14627 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
14628 (expand_unop): Update accordingly.
14629 (expand_abs_nojump): Likewise.
14630 (expand_copysign_absneg): Take a scalar_float_mode.
14631 (expand_copysign_bit): Likewise.
14632 (expand_copysign): Update accordingly.
14634 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14635 Alan Hayward <alan.hayward@arm.com>
14636 David Sherwood <david.sherwood@arm.com>
14638 * coretypes.h (opt_mode): New class.
14639 * machmode.h (opt_mode): Likewise.
14640 (opt_mode::else_void): New function.
14641 (opt_mode::require): Likewise.
14642 (opt_mode::exists): Likewise.
14643 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
14644 (GET_MODE_2XWIDER_MODE): Likewise.
14645 (mode_iterator::get_wider): Update accordingly.
14646 (mode_iterator::get_2xwider): Likewise.
14647 (mode_iterator::get_known_wider): Likewise, turning into a template.
14648 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
14649 forcing a wider mode to exist.
14650 * config/cr16/cr16.h (LONG_REG_P): Likewise.
14651 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
14652 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
14653 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
14654 * lower-subreg.c (init_lower_subreg): Likewise.
14655 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
14656 on the final iteration.
14657 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
14658 a wider mode exists before asking for a move pattern.
14659 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
14660 forcing a wider mode to exist.
14661 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
14662 returning false if no such mode exists.
14663 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
14664 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
14665 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
14666 Avoid checking for a MODE_INT if we already know the mode is not a
14668 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
14669 forcing a wider mode to exist.
14670 (expmed_mult_highpart_optab): Likewise.
14671 (expmed_mult_highpart): Likewise.
14672 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
14674 * lto-streamer-in.c (lto_input_mode_table): Likewise.
14675 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
14676 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
14677 * internal-fn.c (expand_mul_overflow): Update use of
14678 GET_MODE_2XWIDER_MODE.
14679 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14680 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
14681 GET_MODE_WIDER_MODE.
14682 (convert_plusminus_to_widen): Likewise.
14683 * tree-switch-conversion.c (array_value_type): Likewise.
14684 * var-tracking.c (emit_note_insn_var_location): Likewise.
14685 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14686 Return false inside rather than outside the loop if no wider mode
14688 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
14689 and GET_MODE_2XWIDER_MODE
14690 (can_compare_p): Use else_void.
14691 * gdbhooks.py (OptMachineModePrinter): New class.
14692 (build_pretty_printer): Use it for opt_mode.
14694 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14695 Alan Hayward <alan.hayward@arm.com>
14696 David Sherwood <david.sherwood@arm.com>
14698 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
14699 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
14701 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14702 Alan Hayward <alan.hayward@arm.com>
14703 David Sherwood <david.sherwood@arm.com>
14705 * machmode.h (mode_traits): New structure.
14706 (get_narrowest_mode): New function.
14707 (mode_iterator::start): Likewise.
14708 (mode_iterator::iterate_p): Likewise.
14709 (mode_iterator::get_wider): Likewise.
14710 (mode_iterator::get_known_wider): Likewise.
14711 (mode_iterator::get_2xwider): Likewise.
14712 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
14713 (FOR_EACH_MODE): Likewise.
14714 (FOR_EACH_MODE_FROM): Likewise.
14715 (FOR_EACH_MODE_UNTIL): Likewise.
14716 (FOR_EACH_WIDER_MODE): Likewise.
14717 (FOR_EACH_2XWIDER_MODE): Likewise.
14718 * builtins.c (expand_builtin_strlen): Use new mode iterators.
14719 * combine.c (simplify_comparison): Likewise
14720 * config/i386/i386.c (type_natural_mode): Likewise.
14721 * cse.c (cse_insn): Likewise.
14722 * dse.c (find_shift_sequence): Likewise.
14723 * emit-rtl.c (init_derived_machine_modes): Likewise.
14724 (init_emit_once): Likewise.
14725 * explow.c (hard_function_value): Likewise.
14726 * expmed.c (extract_fixed_bit_field_1): Likewise.
14727 (extract_bit_field_1): Likewise.
14728 (expand_divmod): Likewise.
14729 (emit_store_flag_1): Likewise.
14730 * expr.c (init_expr_target): Likewise.
14731 (convert_move): Likewise.
14732 (alignment_for_piecewise_move): Likewise.
14733 (widest_int_mode_for_size): Likewise.
14734 (emit_block_move_via_movmem): Likewise.
14735 (copy_blkmode_to_reg): Likewise.
14736 (set_storage_via_setmem): Likewise.
14737 (compress_float_constant): Likewise.
14738 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14739 * optabs-query.c (get_best_extraction_insn): Likewise.
14740 * optabs.c (expand_binop): Likewise.
14741 (expand_twoval_unop): Likewise.
14742 (expand_twoval_binop): Likewise.
14743 (widen_leading): Likewise.
14744 (widen_bswap): Likewise.
14745 (expand_parity): Likewise.
14746 (expand_unop): Likewise.
14747 (prepare_cmp_insn): Likewise.
14748 (prepare_float_lib_cmp): Likewise.
14749 (expand_float): Likewise.
14750 (expand_fix): Likewise.
14751 (expand_sfix_optab): Likewise.
14752 * postreload.c (move2add_use_add2_insn): Likewise.
14753 * reg-stack.c (reg_to_stack): Likewise.
14754 * reginfo.c (choose_hard_reg_mode): Likewise.
14755 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
14756 * stor-layout.c (mode_for_size): Likewise.
14757 (smallest_mode_for_size): Likewise.
14758 (mode_for_vector): Likewise.
14759 (finish_bitfield_representative): Likewise.
14760 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
14761 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
14762 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14763 * var-tracking.c (prepare_call_arguments): Likewise.
14765 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14766 Alan Hayward <alan.hayward@arm.com>
14767 David Sherwood <david.sherwood@arm.com>
14769 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
14770 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
14771 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
14772 * machmode.h (mode_size): Move earlier in file.
14773 (mode_precision): Likewise.
14774 (mode_inner): Likewise.
14775 (mode_nunits): Likewise.
14776 (mode_unit_size): Likewise.
14777 (unit_unit_precision): Likewise.
14778 (mode_wider): Likewise.
14779 (mode_2xwider): Likewise.
14780 (machine_mode): New class.
14781 (mode_to_bytes): New function.
14782 (mode_to_bits): Likewise.
14783 (mode_to_precision): Likewise.
14784 (mode_to_inner): Likewise.
14785 (mode_to_unit_size): Likewise.
14786 (mode_to_unit_precision): Likewise.
14787 (mode_to_nunits): Likewise.
14788 (GET_MODE_SIZE): Use mode_to_bytes.
14789 (GET_MODE_BITSIZE): Use mode_to_bits.
14790 (GET_MODE_PRECISION): Use mode_to_precision.
14791 (GET_MODE_INNER): Use mode_to_inner.
14792 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
14793 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
14794 (GET_MODE_NUNITS): Use mode_to_nunits.
14795 * system.h (ALWAYS_INLINE): New macro.
14796 * config/powerpcspe/powerpcspe-c.c
14797 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
14798 int for arg1_mode and arg2_mode.
14800 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14801 Alan Hayward <alan.hayward@arm.com>
14802 David Sherwood <david.sherwood@arm.com>
14804 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
14805 Prefix mode names with E_ in case statements.
14806 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14807 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
14808 (aarch64_split_simd_move): Likewise.
14809 (aarch64_gen_storewb_pair): Likewise.
14810 (aarch64_gen_loadwb_pair): Likewise.
14811 (aarch64_gen_store_pair): Likewise.
14812 (aarch64_gen_load_pair): Likewise.
14813 (aarch64_get_condition_code_1): Likewise.
14814 (aarch64_constant_pool_reload_icode): Likewise.
14815 (get_rsqrte_type): Likewise.
14816 (get_rsqrts_type): Likewise.
14817 (get_recpe_type): Likewise.
14818 (get_recps_type): Likewise.
14819 (aarch64_gimplify_va_arg_expr): Likewise.
14820 (aarch64_simd_container_mode): Likewise.
14821 (aarch64_emit_load_exclusive): Likewise.
14822 (aarch64_emit_store_exclusive): Likewise.
14823 (aarch64_expand_compare_and_swap): Likewise.
14824 (aarch64_gen_atomic_cas): Likewise.
14825 (aarch64_emit_bic): Likewise.
14826 (aarch64_emit_atomic_swap): Likewise.
14827 (aarch64_emit_atomic_load_op): Likewise.
14828 (aarch64_evpc_trn): Likewise.
14829 (aarch64_evpc_uzp): Likewise.
14830 (aarch64_evpc_zip): Likewise.
14831 (aarch64_evpc_ext): Likewise.
14832 (aarch64_evpc_rev): Likewise.
14833 (aarch64_evpc_dup): Likewise.
14834 (aarch64_gen_ccmp_first): Likewise.
14835 (aarch64_gen_ccmp_next): Likewise.
14836 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
14837 (alpha_emit_xfloating_libcall): Likewise.
14838 (emit_insxl): Likewise.
14839 (alpha_arg_type): Likewise.
14840 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
14841 (arc_preferred_simd_mode): Likewise.
14842 (arc_secondary_reload): Likewise.
14843 (get_arc_condition_code): Likewise.
14844 (arc_print_operand): Likewise.
14845 (arc_legitimate_constant_p): Likewise.
14846 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14847 * config/arc/arc.md (casesi_load): Likewise.
14848 (casesi_compact_jump): Likewise.
14849 * config/arc/predicates.md (proper_comparison_operator): Likewise.
14850 (cc_use_register): Likewise.
14851 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14852 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
14853 (arm_init_iwmmxt_builtins): Likewise.
14854 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
14855 (neon_expand_vector_init): Likewise.
14856 (arm_attr_length_move_neon): Likewise.
14857 (maybe_get_arm_condition_code): Likewise.
14858 (arm_emit_vector_const): Likewise.
14859 (arm_preferred_simd_mode): Likewise.
14860 (arm_output_iwmmxt_tinsr): Likewise.
14861 (thumb1_output_casesi): Likewise.
14862 (thumb2_output_casesi): Likewise.
14863 (arm_emit_load_exclusive): Likewise.
14864 (arm_emit_store_exclusive): Likewise.
14865 (arm_expand_compare_and_swap): Likewise.
14866 (arm_evpc_neon_vuzp): Likewise.
14867 (arm_evpc_neon_vzip): Likewise.
14868 (arm_evpc_neon_vrev): Likewise.
14869 (arm_evpc_neon_vtrn): Likewise.
14870 (arm_evpc_neon_vext): Likewise.
14871 (arm_validize_comparison): Likewise.
14872 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
14873 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
14874 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
14875 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
14876 (c6x_preferred_simd_mode): Likewise.
14877 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
14878 (epiphany_rtx_costs): Likewise.
14879 * config/epiphany/predicates.md (proper_comparison_operator):
14881 * config/frv/frv.c (condexec_memory_operand): Likewise.
14882 (frv_emit_move): Likewise.
14883 (output_move_single): Likewise.
14884 (output_condmove_single): Likewise.
14885 (frv_hard_regno_mode_ok): Likewise.
14886 (frv_matching_accg_mode): Likewise.
14887 * config/h8300/h8300.c (split_adds_subs): Likewise.
14888 (h8300_rtx_costs): Likewise.
14889 (h8300_print_operand): Likewise.
14890 (compute_mov_length): Likewise.
14891 (output_logical_op): Likewise.
14892 (compute_logical_op_length): Likewise.
14893 (compute_logical_op_cc): Likewise.
14894 (h8300_shift_needs_scratch_p): Likewise.
14895 (output_a_shift): Likewise.
14896 (compute_a_shift_length): Likewise.
14897 (compute_a_shift_cc): Likewise.
14898 (expand_a_rotate): Likewise.
14899 (output_a_rotate): Likewise.
14900 * config/i386/i386.c (classify_argument): Likewise.
14901 (function_arg_advance_32): Likewise.
14902 (function_arg_32): Likewise.
14903 (function_arg_64): Likewise.
14904 (function_value_64): Likewise.
14905 (ix86_gimplify_va_arg): Likewise.
14906 (ix86_legitimate_constant_p): Likewise.
14907 (put_condition_code): Likewise.
14908 (split_double_mode): Likewise.
14909 (ix86_avx256_split_vector_move_misalign): Likewise.
14910 (ix86_expand_vector_logical_operator): Likewise.
14911 (ix86_split_idivmod): Likewise.
14912 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
14913 (ix86_build_const_vector): Likewise.
14914 (ix86_build_signbit_mask): Likewise.
14915 (ix86_match_ccmode): Likewise.
14916 (ix86_cc_modes_compatible): Likewise.
14917 (ix86_expand_branch): Likewise.
14918 (ix86_expand_sse_cmp): Likewise.
14919 (ix86_expand_sse_movcc): Likewise.
14920 (ix86_expand_int_sse_cmp): Likewise.
14921 (ix86_expand_vec_perm_vpermi2): Likewise.
14922 (ix86_expand_vec_perm): Likewise.
14923 (ix86_expand_sse_unpack): Likewise.
14924 (ix86_expand_int_addcc): Likewise.
14925 (ix86_split_to_parts): Likewise.
14926 (ix86_vectorize_builtin_gather): Likewise.
14927 (ix86_vectorize_builtin_scatter): Likewise.
14928 (avx_vpermilp_parallel): Likewise.
14929 (inline_memory_move_cost): Likewise.
14930 (ix86_tieable_integer_mode_p): Likewise.
14931 (x86_maybe_negate_const_int): Likewise.
14932 (ix86_expand_vector_init_duplicate): Likewise.
14933 (ix86_expand_vector_init_one_nonzero): Likewise.
14934 (ix86_expand_vector_init_one_var): Likewise.
14935 (ix86_expand_vector_init_concat): Likewise.
14936 (ix86_expand_vector_init_interleave): Likewise.
14937 (ix86_expand_vector_init_general): Likewise.
14938 (ix86_expand_vector_set): Likewise.
14939 (ix86_expand_vector_extract): Likewise.
14940 (emit_reduc_half): Likewise.
14941 (ix86_emit_i387_round): Likewise.
14942 (ix86_mangle_type): Likewise.
14943 (ix86_expand_round_sse4): Likewise.
14944 (expand_vec_perm_blend): Likewise.
14945 (canonicalize_vector_int_perm): Likewise.
14946 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
14947 (expand_vec_perm_1): Likewise.
14948 (expand_vec_perm_interleave3): Likewise.
14949 (expand_vec_perm_even_odd_pack): Likewise.
14950 (expand_vec_perm_even_odd_1): Likewise.
14951 (expand_vec_perm_broadcast_1): Likewise.
14952 (ix86_vectorize_vec_perm_const_ok): Likewise.
14953 (ix86_expand_vecop_qihi): Likewise.
14954 (ix86_expand_mul_widen_hilo): Likewise.
14955 (ix86_expand_sse2_abs): Likewise.
14956 (ix86_expand_pextr): Likewise.
14957 (ix86_expand_pinsr): Likewise.
14958 (ix86_preferred_simd_mode): Likewise.
14959 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
14960 * config/i386/sse.md (*andnot<mode>3): Likewise.
14961 (<mask_codefor><code><mode>3<mask_name>): Likewise.
14962 (*<code><mode>3): Likewise.
14963 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
14964 (ia64_expand_atomic_op): Likewise.
14965 (ia64_arg_type): Likewise.
14966 (ia64_mode_to_int): Likewise.
14967 (ia64_scalar_mode_supported_p): Likewise.
14968 (ia64_vector_mode_supported_p): Likewise.
14969 (expand_vec_perm_broadcast): Likewise.
14970 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
14971 (iq2000_function_arg_advance): Likewise.
14972 (iq2000_function_arg): Likewise.
14973 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
14974 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
14975 (m68k_libcall_value): Likewise.
14976 (m68k_function_value): Likewise.
14977 (sched_attr_op_type): Likewise.
14978 * config/mcore/mcore.c (mcore_output_move): Likewise.
14979 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
14981 (microblaze_function_arg): Likewise.
14982 * config/mips/mips.c (mips16_build_call_stub): Likewise.
14983 (mips_print_operand): Likewise.
14984 (mips_mode_ok_for_mov_fmt_p): Likewise.
14985 (mips_vector_mode_supported_p): Likewise.
14986 (mips_preferred_simd_mode): Likewise.
14987 (mips_expand_vpc_loongson_even_odd): Likewise.
14988 (mips_expand_vec_unpack): Likewise.
14989 (mips_expand_vi_broadcast): Likewise.
14990 (mips_expand_vector_init): Likewise.
14991 (mips_expand_vec_reduc): Likewise.
14992 (mips_expand_msa_cmp): Likewise.
14993 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
14994 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
14995 (cc_flags_for_mode): Likewise.
14996 * config/msp430/msp430.c (msp430_print_operand): Likewise.
14997 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
14998 (nds32_output_casesi_pc_relative): Likewise.
14999 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15000 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
15001 (nvptx_gen_unpack): Likewise.
15002 (nvptx_gen_pack): Likewise.
15003 (nvptx_gen_shuffle): Likewise.
15004 (nvptx_gen_wcast): Likewise.
15005 (nvptx_preferred_simd_mode): Likewise.
15006 * config/pa/pa.c (pa_secondary_reload): Likewise.
15007 * config/pa/predicates.md (base14_operand): Likewise.
15008 * config/powerpcspe/powerpcspe-c.c
15009 (altivec_resolve_overloaded_builtin): Likewise.
15010 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
15012 (rs6000_preferred_simd_mode): Likewise.
15013 (output_vec_const_move): Likewise.
15014 (rs6000_expand_vector_extract): Likewise.
15015 (rs6000_split_vec_extract_var): Likewise.
15016 (reg_offset_addressing_ok_p): Likewise.
15017 (rs6000_legitimate_offset_address_p): Likewise.
15018 (rs6000_legitimize_address): Likewise.
15019 (rs6000_emit_set_const): Likewise.
15020 (rs6000_const_vec): Likewise.
15021 (rs6000_emit_move): Likewise.
15022 (spe_build_register_parallel): Likewise.
15023 (rs6000_darwin64_record_arg_recurse): Likewise.
15024 (swap_selector_for_mode): Likewise.
15025 (spe_init_builtins): Likewise.
15026 (paired_init_builtins): Likewise.
15027 (altivec_init_builtins): Likewise.
15028 (do_load_for_compare): Likewise.
15029 (rs6000_generate_compare): Likewise.
15030 (rs6000_expand_float128_convert): Likewise.
15031 (emit_load_locked): Likewise.
15032 (emit_store_conditional): Likewise.
15033 (rs6000_output_function_epilogue): Likewise.
15034 (rs6000_handle_altivec_attribute): Likewise.
15035 (rs6000_function_value): Likewise.
15036 (emit_fusion_gpr_load): Likewise.
15037 (emit_fusion_p9_load): Likewise.
15038 (emit_fusion_p9_store): Likewise.
15039 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
15040 (fusion_gpr_mem_load): Likewise.
15041 (fusion_addis_mem_combo_load): Likewise.
15042 (fusion_addis_mem_combo_store): Likewise.
15043 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
15044 (fusion_gpr_mem_load): Likewise.
15045 (fusion_addis_mem_combo_load): Likewise.
15046 (fusion_addis_mem_combo_store): Likewise.
15047 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15049 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
15050 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15051 (rs6000_preferred_simd_mode): Likewise.
15052 (output_vec_const_move): Likewise.
15053 (rs6000_expand_vector_extract): Likewise.
15054 (rs6000_split_vec_extract_var): Likewise.
15055 (reg_offset_addressing_ok_p): Likewise.
15056 (rs6000_legitimate_offset_address_p): Likewise.
15057 (rs6000_legitimize_address): Likewise.
15058 (rs6000_emit_set_const): Likewise.
15059 (rs6000_const_vec): Likewise.
15060 (rs6000_emit_move): Likewise.
15061 (rs6000_darwin64_record_arg_recurse): Likewise.
15062 (swap_selector_for_mode): Likewise.
15063 (paired_init_builtins): Likewise.
15064 (altivec_init_builtins): Likewise.
15065 (rs6000_expand_float128_convert): Likewise.
15066 (emit_load_locked): Likewise.
15067 (emit_store_conditional): Likewise.
15068 (rs6000_output_function_epilogue): Likewise.
15069 (rs6000_handle_altivec_attribute): Likewise.
15070 (rs6000_function_value): Likewise.
15071 (emit_fusion_gpr_load): Likewise.
15072 (emit_fusion_p9_load): Likewise.
15073 (emit_fusion_p9_store): Likewise.
15074 * config/rx/rx.c (rx_gen_move_template): Likewise.
15075 (flags_from_mode): Likewise.
15076 * config/s390/predicates.md (s390_alc_comparison): Likewise.
15077 (s390_slb_comparison): Likewise.
15078 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
15079 (s390_vector_mode_supported_p): Likewise.
15080 (s390_cc_modes_compatible): Likewise.
15081 (s390_match_ccmode_set): Likewise.
15082 (s390_canonicalize_comparison): Likewise.
15083 (s390_emit_compare_and_swap): Likewise.
15084 (s390_branch_condition_mask): Likewise.
15085 (s390_rtx_costs): Likewise.
15086 (s390_secondary_reload): Likewise.
15087 (__SECONDARY_RELOAD_CASE): Likewise.
15088 (s390_expand_cs): Likewise.
15089 (s390_preferred_simd_mode): Likewise.
15090 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
15091 * config/sh/sh.c (sh_print_operand): Likewise.
15092 (dump_table): Likewise.
15093 (sh_secondary_reload): Likewise.
15094 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15095 * config/sh/sh.md (casesi_worker_1): Likewise.
15096 (casesi_worker_2): Likewise.
15097 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
15098 (fcc_comparison_operator): Likewise.
15099 * config/sparc/sparc.c (sparc_expand_move): Likewise.
15100 (emit_soft_tfmode_cvt): Likewise.
15101 (sparc_preferred_simd_mode): Likewise.
15102 (output_cbranch): Likewise.
15103 (sparc_print_operand): Likewise.
15104 (sparc_expand_vec_perm_bmask): Likewise.
15105 (vector_init_bshuffle): Likewise.
15106 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
15107 (spu_vector_mode_supported_p): Likewise.
15108 (spu_expand_insv): Likewise.
15109 (spu_emit_branch_or_set): Likewise.
15110 (spu_handle_vector_attribute): Likewise.
15111 (spu_builtin_splats): Likewise.
15112 (spu_builtin_extract): Likewise.
15113 (spu_builtin_promote): Likewise.
15114 (spu_expand_sign_extend): Likewise.
15115 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
15116 (tilegx_simd_int): Likewise.
15117 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
15118 (tilepro_simd_int): Likewise.
15119 * config/v850/v850.c (const_double_split): Likewise.
15120 (v850_print_operand): Likewise.
15121 (ep_memory_offset): Likewise.
15122 * config/vax/vax.c (vax_rtx_costs): Likewise.
15123 (vax_output_int_move): Likewise.
15124 (vax_output_int_add): Likewise.
15125 (vax_output_int_subtract): Likewise.
15126 * config/visium/predicates.md (visium_branch_operator): Likewise.
15127 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
15128 (visium_print_operand_address): Likewise.
15129 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15130 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
15131 (xtensa_expand_conditional_branch): Likewise.
15132 (xtensa_copy_incoming_a7): Likewise.
15133 (xtensa_output_literal): Likewise.
15134 * dfp.c (decimal_real_maxval): Likewise.
15135 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
15137 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15138 Alan Hayward <alan.hayward@arm.com>
15139 David Sherwood <david.sherwood@arm.com>
15141 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
15142 (mode_nunits_inline): Likewise.
15143 (mode_inner_inline): Likewise.
15144 (mode_unit_size_inline): Likewise.
15145 (mode_unit_precision_inline): Likewise.
15146 (emit_insn_modes_h): Likewise. Also emit a #define of the
15148 (emit_mode_wider): Add an E_ prefix to mode names.
15149 (emit_mode_complex): Likewise.
15150 (emit_mode_inner): Likewise.
15151 (emit_mode_adjustments): Likewise.
15152 (emit_mode_int_n): Likewise.
15153 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
15154 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
15155 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
15156 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
15157 (CRC32_BUILTIN, ENTRY): Likewise.
15158 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
15159 (aarch64_pop_regs): Likewise.
15160 (aarch64_process_components): Likewise.
15161 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
15162 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
15163 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
15164 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
15165 * config/arm/arm.c (arm_init_libfuncs): Likewise.
15166 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
15168 * config/i386/i386-builtin.def (pcmpestr): Likewise.
15169 (pcmpistr): Likewise.
15170 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
15171 * config/mmix/mmix.c (mmix_output_condition): Likewise.
15172 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
15174 * config/rl78/rl78.c (mduc_regs): Likewise.
15175 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
15176 (htm_expand_builtin): Likewise.
15177 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
15178 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
15179 * config/xtensa/xtensa.c (print_operand): Likewise.
15180 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
15181 (NUM_MODE_VECTOR_INT): Likewise.
15182 * genoutput.c (null_operand): Likewise.
15183 (output_operand_data): Likewise.
15184 * genrecog.c (print_parameter_value): Likewise.
15185 * lra.c (debug_operand_data): Likewise.
15187 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15189 * dwarf2out.c (scompare_loc_descriptor_wide)
15190 (scompare_loc_descriptor_narrow): New functions, split out from...
15191 (scompare_loc_descriptor): ...here.
15192 * expmed.c (emit_store_flag_int): New function, split out from...
15193 (emit_store_flag): ...here.
15195 2017-08-30 Richard Biener <rguenther@suse.de>
15197 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
15198 (dwarf2out_early_finish): Move setting of AT_pubnames from
15199 early debug output to early finish.
15201 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15203 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
15204 and -mdata-region to the assembler.
15206 2017-08-30 Richard Biener <rguenther@suse.de>
15208 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
15210 (gen_subprogram_die): Add DW_AT_object_pointer only early.
15211 (dwarf2out_early_global_decl): Only generate a DIE for the
15212 abstract origin if it doesn't already exist or is a declaration DIE.
15213 (resolve_addr): Do not add the linkage name twice when
15214 generating a stub DIE for the DW_TAG_GNU_call_site target.
15216 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15218 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15219 Use machine_mode rather than int for arg1_mode.
15221 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
15224 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
15225 that the second argument of the built-in functions to unpack
15226 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
15227 switch statement instead a lot of if statements.
15228 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
15229 Allow 64-bit values to be in Altivec registers as well as
15230 traditional floating point registers.
15231 (pack<mode>, FMOVE128_VSX iterator): Likewise.
15233 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
15235 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
15236 MAX_REGS_PER_ADDRESS == 1.
15238 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
15240 * config/i386/i386.opt (flag_fentry): Do not init to -1.
15241 * config/i386/i386.c (ix86_option_override_internal): Simplify
15242 setting of opts->x_flag_entry.
15244 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15245 Jakub Jelinek <jakub@redhat.com>
15246 Richard Biener <rguenther@suse.de>
15248 PR tree-optimization/81503
15249 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
15250 folded constant fits in the target type; reorder tests for clarity.
15252 2017-08-29 Martin Liska <mliska@suse.cz>
15254 * passes.def: Include pass_lower_switch.
15255 * stmt.c (dump_case_nodes): Remove and move to
15256 tree-switch-conversion.
15257 (case_values_threshold): Likewise.
15258 (expand_switch_as_decision_tree_p): Likewise.
15259 (emit_case_decision_tree): Likewise.
15260 (expand_case): Likewise.
15261 (balance_case_nodes): Likewise.
15262 (node_has_low_bound): Likewise.
15263 (node_has_high_bound): Likewise.
15264 (node_is_bounded): Likewise.
15265 (emit_case_nodes): Likewise.
15266 (struct simple_case_node): New struct.
15267 (add_case_node): Remove.
15268 (emit_case_dispatch_table): Use vector instead of case_list.
15269 (reset_out_edges_aux): Remove.
15270 (compute_cases_per_edge): Likewise.
15271 (expand_case): Build list of simple_case_node.
15272 (expand_sjlj_dispatch_table): Use it.
15273 * tree-switch-conversion.c (struct case_node): Moved from
15274 stmt.c and adjusted.
15275 (emit_case_nodes): Likewise.
15276 (node_has_low_bound): Likewise.
15277 (node_has_high_bound): Likewise.
15278 (node_is_bounded): Likewise.
15279 (case_values_threshold): Likewise.
15280 (reset_out_edges_aux): Likewise.
15281 (compute_cases_per_edge): Likewise.
15282 (add_case_node): Likewise.
15283 (dump_case_nodes): Likewise.
15284 (balance_case_nodes): Likewise.
15285 (expand_switch_as_decision_tree_p): Likewise.
15286 (emit_jump): Likewise.
15287 (emit_case_decision_tree): Likewise.
15288 (try_switch_expansion): Likewise.
15289 (do_jump_if_equal): Likewise.
15290 (emit_cmp_and_jump_insns): Likewise.
15291 (fix_phi_operands_for_edge): New function.
15292 (record_phi_operand_mapping): Likewise.
15293 (class pass_lower_switch): New pass.
15294 (pass_lower_switch::execute): New function.
15295 (make_pass_lower_switch): Likewise.
15296 (conditional_probability):
15297 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
15298 * tree-pass.h: Add make_pass_lower_switch.
15300 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15303 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
15306 2017-08-29 Richard Biener <rguenther@suse.de>
15308 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
15309 we do not add a DW_AT_inline attribute twice.
15310 (gen_subprogram_die): Remove code setting DW_AT_inline on
15311 DECL_ABSTRACT_P nodes.
15313 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
15315 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
15316 calls to internal functions.
15317 (gimplify_modify_expr): Likewise.
15318 * tree-call-cdce.c (use_internal_fn): Likewise.
15319 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
15320 (convert_to_divmod): Set the nothrow flag.
15321 * tree-if-conv.c (predicate_mem_writes): Likewise.
15322 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15323 (vectorizable_call): Likewise.
15324 (vectorizable_store): Likewise.
15325 (vectorizable_load): Likewise.
15326 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
15327 (vect_recog_mask_conversion_pattern): Likewise.
15329 2017-08-29 Martin Liska <mliska@suse.cz>
15332 * gcc.c (driver_handle_option): Add new argument.
15333 * opts-common.c (handle_option): Pass
15334 target_option_override_hook.
15335 * opts-global.c (lang_handle_option): Add new option.
15336 (set_default_handlers): Add new argument.
15337 (decode_options): Likewise.
15338 * opts.c (target_handle_option): Likewise.
15339 (common_handle_option): Call target_option_override_hook.
15340 * opts.h (struct cl_option_handler_func): Add hook for
15341 target option override.
15342 (struct cl_option_handlers): Likewise.
15343 (set_default_handlers): Add new argument.
15344 (decode_options): Likewise.
15345 (common_handle_option): Likewise.
15346 (target_handle_option): Likewise.
15347 * toplev.c (toplev::main): Pass targetm.target_option.override
15350 2017-08-29 Richard Biener <rguenther@suse.de>
15351 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
15353 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
15354 life to the active subtree.
15356 2017-08-28 Jeff Law <law@redhat.com>
15358 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
15359 derive_equivalences.
15360 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
15361 Code moved into....
15362 (edge_info::derive_equivalences): New private member function
15364 * tree-ssa-dom.c (class edge_info): Changed from a struct
15365 to a class. Add ctor/dtor, methods and data members.
15366 (edge_info::edge_info): Renamed from allocate_edge_info.
15367 Initialize additional members.
15368 (edge_info::~edge_info): New.
15369 (free_dom_edge_info): Delete the edge info.
15370 (record_edge_info): Use new class & associated member functions.
15371 Tighten forms for testing for edge equivalences.
15372 (record_temporary_equivalences): Iterate over the simple
15373 equivalences rather than assuming there's only one per edge.
15374 (cprop_into_successor_phis): Iterate over the simple
15375 equivalences rather than assuming there's only one per edge.
15376 (optimize_stmt): Use operand_equal_p rather than pointer
15377 equality for mini-DSE code.
15379 2017-08-28 Nathan Sidwell <nathan@acm.org>
15381 * gcc.c (execute): Fold SIGPIPE handling into switch
15382 statement. Adjust internal error message.
15384 2017-08-28 Richard Biener <rguenther@suse.de>
15387 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
15388 Do nothing for removed DIEs.
15390 2017-08-28 Richard Biener <rguenther@suse.de>
15392 PR tree-optimization/81977
15393 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
15396 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
15399 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
15400 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
15401 using targetm.gen_mem_thread_fence.
15403 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
15406 * config/i386/i386.md (*<btsc><mode>): Change operand 2
15407 predicate to register_operand. Reorder operands.
15408 (*btr<mode>): Ditto.
15409 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
15410 (*btr<mode>_mask): Ditto.
15412 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
15414 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
15415 * config/rs6000/xmmintrin.h: New file.
15416 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
15418 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15421 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
15422 parameter and_insn and return it.
15423 (recombine_lvx_pattern): Insert a copy to ensure availability of
15424 the base register of the copied masking operation at the point of
15425 the instruction replacement.
15426 (recombine_stvx_pattern): Likewise.
15428 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
15430 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
15431 undocumented switches.
15432 (-mpower9-dform-vector): Likewise.
15433 (-mpower9-dform): Likewise.
15434 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
15435 comments to delete references to -mpower9-dform* switches.
15436 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
15437 Delete reference to -mpower9-dform* switches, test for
15438 -mpower9-vector instead.
15439 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
15440 (OTHER_P9_VECTOR_MASKS): Likewise.
15441 (POWERPC_MASKS): Likewise.
15442 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
15443 tests against -mpower9-dform* to -mpower9-vector. Delete code
15444 that checked for -mpower9-dform* consistancy with other options.
15445 Add test for -mpower9-misc to enable other power9 switches.
15446 (rs6000_init_hard_regno_mode_ok): Likewise.
15447 (rs6000_option_override_internal): Likewise.
15448 (rs6000_emit_prologue): Likewise.
15449 (rs6000_emit_epilogue): Likewise.
15450 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
15451 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
15452 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
15454 (emit_fusion_p9_store): Likewise.
15455 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
15456 resetting these macros if the assembler does not support ISA 3.0
15458 (TARGET_P9_DFORM_VECTOR): Likewise.
15459 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
15460 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
15462 2017-08-25 Alan Modra <amodra@gmail.com>
15464 PR rtl-optimization/81747
15465 * cse.c (cse_extended_basic_block): Don't attempt to record
15466 equivalences for degenerate conditional branches that branch
15467 to their fall-through.
15469 2017-08-24 Martin Sebor <msebor@redhat.com>
15471 PR middle-end/81908
15472 * gimple-fold.c (size_must_be_zero_p): New function.
15473 (gimple_fold_builtin_memory_op): Call it.
15475 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
15477 * config/rs6000/mm_malloc.h: New file.
15479 2017-08-24 Bin Cheng <bin.cheng@arm.com>
15481 PR tree-optimization/81913
15482 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
15483 analysis when either IVs in condition can wrap.
15485 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
15487 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
15488 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
15490 2017-08-24 Richard Biener <rguenther@suse.de>
15493 * targhooks.c (default_target_can_inline_p): Properly
15494 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
15495 is present and always compare.
15496 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
15497 infer -mfpmath=sse from TARGET_SSE_P.
15498 (ix86_can_inline_p): Properly use target_option_default_node when
15499 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
15501 2017-08-24 Richard Biener <rguenther@suse.de>
15504 * dwarf2out.c (output_die): Handle flag_generate_offload like
15506 (output_comp_unit): Likewise.
15507 (gen_array_type_die): Likewise.
15508 (dwarf2out_early_finish): Likewise.
15509 (note_variable_value_in_expr): Likewise.
15510 (dwarf2out_finish): Likewise. Adjust assert.
15511 * cgraphunit.c (symbol_table::compile): Move setting of
15512 flag_generate_offload earlier ...
15513 (symbol_table::finalize_compilation_unit): ... here, before
15514 early debug finalization.
15516 2017-08-24 Richard Biener <rguenther@suse.de>
15518 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
15519 and ipa-fnsummary.h.
15520 (ix86_can_inline_p): When ix86_fpmath flags do not match
15521 check whether the callee uses FP math at all.
15523 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
15525 PR middle-end/81931
15526 * tree-ssanames.c (get_nonzero_bits): Use element_precision
15527 instead of TYPE_PRECISION.
15529 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
15530 Alan Hayward <alan.hayward@arm.com>
15531 David Sherwood <david.sherwood@arm.com>
15533 * combine.c (make_extraction): Use subreg_offset_from_lsb.
15535 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
15537 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
15539 (ix86_frame::stack_realign_allocate): New field.
15540 (struct machine_frame_state): Modify comments.
15541 (machine_frame_state::sp_realigned_fp_end): New field.
15542 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
15543 layout calculation.
15544 (sp_valid_at): Add assertion to assure no attempt to access invalid
15545 offset of a realigned stack.
15546 (fp_valid_at): Likewise.
15547 (choose_baseaddr): Modify comments.
15548 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
15549 ix86_expand_prologue.
15550 (ix86_expand_prologue): Modify stack realignment and allocation.
15551 (ix86_expand_epilogue): Modify comments.
15552 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
15553 avx2_runtime, avx512f, and avx512f_runtime.
15555 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
15557 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
15558 (mstackrealign): Do not init to -1.
15559 * config/i386/i386.c (ix86_option_override_internal):
15560 Check opts_set, not opts when setting default value of
15561 opts->x_ix86_force_align_arg_pointer.
15563 2017-08-23 Richard Biener <rguenther@suse.de>
15565 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
15566 lang_hooks.decl_printable_name.
15567 * print-rtl-function.c (print_rtx_function): Likewise.
15568 * tree-pretty-print.c (dump_function_header): Likewise.
15570 2017-08-23 Richard Biener <rguenther@suse.de>
15573 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
15574 -g0 at compile-time.
15576 2017-08-23 Tamar Christina <tamar.christina@arm.com>
15578 PR middle-end/19706
15579 * doc/sourcebuild.texi (Other hardware attributes):
15582 2017-08-23 Tamar Christina <tamar.christina@arm.com>
15584 PR middle-end/19706
15585 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
15586 Fix single-use check.
15588 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
15590 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
15592 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
15594 * doc/install.texi: Modify to add more details on running selected
15597 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
15599 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
15600 is combined with -mabi=ms.
15601 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
15604 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15606 PR tree-optimization/81488
15607 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
15608 and cached_basis fields.
15609 (MAX_SPREAD): New constant.
15610 (alloc_cand_and_find_basis): Initialize new fields.
15611 (clear_visited): New function.
15612 (create_phi_basis_1): Rename from create_phi_basis, set visited
15613 and cached_basis fields.
15614 (create_phi_basis): New wrapper function.
15615 (phi_add_costs_1): Rename from phi_add_costs, add spread
15616 parameter, set visited field, short-circuit when limits reached.
15617 (phi_add_costs): New wrapper function.
15618 (record_phi_increments_1): Rename from record_phi_increments, set
15620 (record_phi_increments): New wrapper function.
15621 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
15622 (phi_incr_cost): New wrapper function.
15623 (all_phi_incrs_profitable_1): Rename from
15624 all_phi_incrs_profitable, set visited field.
15625 (all_phi_incrs_profitable): New wrapper function.
15627 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
15628 Alan Hayward <alan.hayward@arm.com>
15629 David Sherwood <david.sherwood@arm.com>
15631 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
15632 that takes the outer and inner modes.
15633 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
15634 comparison as the canonical test for a paradoxical subreg.
15635 * combine.c (simplify_set): Use paradoxical_subreg_p.
15636 (make_extraction): Likewise.
15637 (force_to_mode): Likewise.
15638 (rtx_equal_for_field_assignment_p): Likewise.
15639 (gen_lowpart_for_combine): Likewise.
15640 (simplify_comparison): Likewise.
15641 * cse.c (equiv_constant): Likewise.
15642 * expmed.c (store_bit_field_1): Likewise.
15643 * final.c (alter_subreg): Likewise.
15644 * fwprop.c (propagate_rtx): Likewise.
15645 (forward_propagate_subreg): Likewise.
15646 * ira-conflicts.c (ira_build_conflicts): Likewise.
15647 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
15648 * lra-constraints.c (curr_insn_transform): Likewise.
15649 (split_reg): Likewise.
15650 * lra-eliminations.c (move_plus_up): Likewise.
15651 (lra_eliminate_regs_1): Likewise.
15652 * recog.c (general_operand): Likewise.
15653 * ree.c (combine_reaching_defs): Likewise.
15654 * reload.c (push_reload): Likewise.
15655 (find_reloads): Likewise.
15656 * reload1.c (elimination_effects): Likewise.
15657 (compute_reload_subreg_offset): Likewise.
15658 (choose_reload_regs): Likewise.
15659 * rtlanal.c (subreg_lsb_1): Likewise.
15660 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15661 (simplify_subreg): Likewise.
15662 * var-tracking.c (track_loc_p): Likewise.
15663 * emit-rtl.c (byte_lowpart_offset): Likewise.
15664 (paradoxical_subreg_p): Delete out-of-line definition.
15666 2017-08-22 Jeff Law <law@redhat.com>
15668 PR tree-optimization/81741
15669 PR tree-optimization/71947
15670 * tree-ssa-dom.c: Include tree-inline.h.
15671 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
15672 equivalences if one is more expensive to compute than the other.
15673 * tree-ssa-scopedtables.h (class const_or_copies): Make
15674 record_const_or_copy_raw method private.
15675 (class avail_exprs_stack): New method simplify_binary_operation.
15676 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
15677 avail_exprs_stack::simplify_binary_operation as needed.
15678 (avail_exprs_stack::simplify_binary_operation): New function.
15680 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
15682 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
15683 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
15684 (DOT_SYMBOLS): Likewise.
15685 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
15686 (RELOCATABLE_NEEDS_FIXUP): Likewise.
15687 (RS6000_ABI_NAME): Likewise.
15688 (TARGET_CMODEL): Likewise.
15689 (TOC_SECTION_ASM_OP): Likewise.
15690 (SET_CMODEL): New macro.
15691 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
15693 2017-08-22 Richard Biener <rguenther@suse.de>
15695 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
15696 to allow for free-lang-data replacements similar to verify_type_variant.
15698 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
15699 Alan Hayward <alan.hayward@arm.com>
15700 David Sherwood <david.sherwood@arm.com>
15702 * config/aarch64/aarch64.md (casesi): Use DImode rather than
15703 VOIDmode for the LABEL_REF.
15705 2017-08-22 Richard Biener <rguenther@suse.de>
15707 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
15709 2017-08-22 Richard Biener <rguenther@suse.de>
15711 * common.opt (feliminate-dwarf2-dups): Ignore.
15712 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
15713 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
15714 same_die_p_wrap, compute_section_prefix,
15715 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
15716 (comdat_symbol_id, comdat_symbol_number): Likewise.
15717 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
15719 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
15720 (output_die): Mark unreachable path unreachable.
15721 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
15722 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
15723 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
15724 (dwarf2out_early_finish): Likewise.
15726 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
15728 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
15730 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
15733 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
15734 not VAR_P. Filter attribute warnings with OPT_Wattributes.
15735 (avr_attribute_table) <io, io_low, address>: Initialize
15736 .decl_required with true.
15738 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
15740 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
15741 undocumented debugging options.
15742 (-mvsx-scalar-double): Likewise.
15743 (-mallow-df-permute): Likewise.
15744 (-mvectorize-builtins): Likewise.
15745 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
15746 (rs6000_builtin_vectorized_function): Likewise.
15747 (rs6000_builtin_md_vectorized_function): Likewise.
15748 (rs6000_opt_vars): Likewise.
15750 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
15753 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
15754 (*btrq_imm): Rename from *btrq.
15755 (*btcq_imm): Rename from *btcq.
15756 (btsc): New code attribute.
15757 (*<btsc><mode>): New insn pattern.
15758 (*btr<mode>): Ditto.
15759 (*<btsc><mode>_mask): New insn_and_split pattern.
15760 (*btr<mode>_mask): Ditto.
15762 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15763 Alan Hayward <alan.hayward@arm.com>
15764 David Sherwood <david.sherwood@arm.com>
15766 * function.c (pad_below): Simplify padding calculation.
15768 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15769 Alan Hayward <alan.hayward@arm.com>
15770 David Sherwood <david.sherwood@arm.com>
15772 * target.def (function_prologue): Remove frame size argument.
15773 (function_epilogue): Likewise.
15774 * doc/tm.texi: Regenerate.
15775 * final.c (final_start_function): Update call to function_prologue.
15776 (final_end_function): Update call to function_epilogue.
15777 (default_function_pro_epilogue): Remove frame size argument.
15778 * output.h (default_function_pro_epilogue): Likewise.
15779 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
15780 (arm_output_function_prologue): Likewise.
15781 * config/frv/frv.c (frv_function_prologue): Likewise.
15782 (frv_function_epilogue): Likewise.
15783 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
15784 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
15785 (ia64_output_function_epilogue): Likewise.
15786 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
15787 (m32r_output_function_epilogue): Likewise.
15788 * config/microblaze/microblaze.c (microblaze_function_prologue)
15789 (microblaze_function_epilogue): Likewise.
15790 * config/mips/mips.c (mips_output_function_prologue): Likewise.
15791 (mips_output_function_epilogue): Likewise.
15792 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
15793 (mmix_target_asm_function_epilogue): Likewise.
15794 * config/msp430/msp430.c (msp430_start_function): Likewise.
15795 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
15796 (nds32_asm_function_epilogue): Likewise.
15797 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
15798 * config/pa/pa.c (pa_output_function_prologue): Likewise.
15799 (pa_output_function_epilogue): Likewise.
15800 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
15801 (rs6000_output_function_epilogue): Likewise.
15802 * config/rl78/rl78.c (rl78_start_function): Likewise.
15803 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
15804 (rs6000_output_function_epilogue): Likewise.
15805 * config/rx/rx.c (rx_output_function_prologue): Likewise.
15806 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
15807 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
15808 (sparc_asm_function_epilogue): Likewise.
15810 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15812 * tree.h (type_has_mode_precision_p): New function.
15813 * convert.c (convert_to_integer_1): Use it.
15814 * expr.c (expand_expr_real_2): Likewise.
15815 (expand_expr_real_1): Likewise.
15816 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
15817 * match.pd: Likewise.
15818 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
15819 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
15820 * tree-tailcall.c (process_assignment): Likewise.
15821 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15822 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
15823 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
15824 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15825 (vectorizable_assignment): Likewise.
15826 (vectorizable_shift): Likewise.
15827 (vectorizable_operation): Likewise.
15828 * tree-vrp.c (register_edge_assert_for_2): Likewise.
15830 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
15832 * match.pd: Add pow (C, x) simplification.
15834 2017-08-21 Richard Biener <rguenther@suse.de>
15836 PR tree-optimization/81900
15837 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
15838 for blocks with abnormal predecessors.
15839 (compute_antic): Do not set visited flag prematurely.
15841 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
15844 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
15846 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15848 * stor-layout.h (vector_type_mode): Move to...
15849 * tree.h (vector_type_mode): ...here.
15850 * stor-layout.c (vector_type_mode): Move to...
15851 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
15853 2017-08-21 Richard Biener <rguenther@suse.de>
15855 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
15856 register_external_die hooks.
15857 (debug_false_tree_charstarstar_uhwistar): Declare.
15858 (debug_nothing_tree_charstar_uhwi): Likewise.
15859 * debug.c (do_nothing_debug_hooks): Adjust.
15860 (debug_false_tree_charstarstar_uhwistar): New do nothing.
15861 (debug_nothing_tree_charstar_uhwi): Likewise.
15862 * dbxout.c (dbx_debug_hooks): Adjust.
15863 (xcoff_debug_hooks): Likewise.
15864 * sdbout.c (sdb_debug_hooks): Likewise.
15865 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
15866 * dwarf2out.c (macinfo_label_base): New global.
15867 (dwarf2out_register_external_die): New function for the
15868 register_external_die hook.
15869 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
15870 (dwarf2_debug_hooks): Use them.
15871 (dwarf2_lineno_debug_hooks): Adjust.
15872 (struct die_struct): Add with_offset flag.
15873 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
15874 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
15875 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
15876 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
15877 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
15878 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
15879 defining section names for the early LTO debug variants.
15880 (reset_indirect_string): New helper.
15881 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
15882 (print_dw_val): Add support for offsetted symbol references.
15883 (get_ultimate_context): Split out from is_cxx.
15884 (is_cxx): Use get_ultimate_context.
15885 (is_fortran): Add decl overload.
15886 (compute_comp_unit_symbol): Split out worker from
15887 compute_section_prefix.
15888 (compute_section_prefix): Call compute_comp_unit_symbol and
15889 set comdat_type_p here.
15890 (output_die): Skip DIE symbol output for the LTO added one.
15891 Handle DIE symbol references with offset.
15892 (output_comp_unit): Guard section name mangling properly.
15893 For LTO debug sections emit a symbol at the section beginning
15894 which we use to refer to its DIEs.
15895 (add_abstract_origin_attribute): For DIEs registered via
15896 dwarf2out_register_external_die directly refer to the early
15897 DIE rather than indirectly through the shadow one we created.
15898 Remove obsolete call to dwarf2out_abstract_function for
15899 non-function/block origins.
15900 (gen_array_type_die): When generating early LTO debug do
15901 not emit DW_AT_string_length.
15902 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
15903 late when in LTO. As suggested place a gcc_unreachable for
15904 the DECL_ABSTRACT_P case.
15905 (gen_subprogram_die): Avoid another specification DIE
15906 for early built declarations/definitions for the late LTO case.
15907 (gen_variable_die): Add type references for late duplicated VLA dies
15909 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
15910 we have the abstract instance already.
15911 (process_scope_var): Adjust decl DIE contexts in LTO which
15912 first puts them in limbo.
15913 (gen_decl_die): Do not generate type DIEs late apart from
15914 types for VLAs or for decls we do not yet have a DIE. Do not
15915 call dwarf2out_abstract_function late.
15916 (dwarf2out_early_global_decl): Make sure to create DIEs
15917 for abstract instances of a decl first.
15918 (dwarf2out_late_global_decl): Adjust comment.
15919 (output_macinfo_op): With multiple macro sections use
15920 macinfo_label_base to distinguish labels.
15921 (output_macinfo): Likewise. Update macinfo_label_base.
15922 Pass in the line info label.
15923 (note_variable_value_in_expr): When generating LTO resolve
15924 all variable values here by generating DIEs as needed.
15925 (init_sections_and_labels): Add early LTO debug flag parameter
15926 and generate different sections and names if set. Add generation
15927 counter for the labels so we can have multiple of them.
15928 (reset_dies): Helper to allow DIEs to be output multiple times.
15929 (dwarf2out_finish): When outputting DIEs to the fat part of an
15930 LTO object first reset DIEs.
15931 (dwarf2out_early_finish): Output early DIEs when generating LTO.
15932 (modified_type_die): Check for decl_ultimate_origin being self
15934 (gen_type_die_with_usage): Likewise.
15935 (gen_typedef_die): Allow decl_ultimate_origin being self.
15936 (set_decl_abstract_flags): Remove.
15937 (set_block_abstract_flags): Likewise.
15938 (dwarf2out_abstract_function): Treat the early generated DIEs
15939 as the abstract copy and only add DW_AT_inline and
15940 DW_AT_artificial here and call set_decl_origin_self.
15941 If the DIE has an abstract origin don't do anything.
15942 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
15943 if we have none yet (Go fails to build one, PR78628).
15944 (variably_modified_type_p): Prevent endless recursion for Ada
15945 cyclic pointer types.
15946 * lto-streamer-in.c: Include debug.h.
15947 (dref_queue): New global.
15948 (lto_read_tree_1): Stream in DIE references.
15949 (lto_input_tree): Register DIE references.
15950 (input_function): Stream DECL_DEBUG_ARGS.
15951 * lto-streamer-out.c: Include debug.h.
15952 (lto_write_tree_1): Output DIE references.
15953 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
15954 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
15955 (output_function): Stream DECL_DEBUG_ARGS.
15956 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
15957 Stream DECL_ABSTRACT_ORIGIN.
15958 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
15959 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
15960 DECL_CONTEXT for file-scope decls.
15961 * lto-streamer.h (struct dref_entry): Declare.
15962 (dref_queue): Likewise.
15963 * cfgexpand.c (pass_expand::execute): Do not call the
15964 outlining_inline_function hook here.
15965 * lto-wrapper.c (debug_obj): New global.
15966 (tool_cleanup): Unlink it if required.
15967 (debug_objcopy): New function.
15968 (run_gcc): Handle early debug sections in the IL files by
15969 extracting them to separate files, partially linkin them and
15970 feeding the result back as result to the linker.
15971 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
15972 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
15973 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
15974 sections into a separate segment.
15975 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
15977 (darwin_asm_dwarf_section): Likewise.
15978 (darwin_asm_output_dwarf_offset): Likewise.
15979 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
15981 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15982 Alan Hayward <alan.hayward@arm.com>
15983 David Sherwood <david.sherwood@arm.com>
15985 * read-md.h (md_reader::record_potential_iterator_use): Replace
15986 pointer argument with an rtx and an index.
15987 * read-rtl.c (iterator_group::apply_iterator): Likewise.
15988 (apply_mode_iterator): Likewise.
15989 (apply_code_iterator): Likewise.
15990 (apply_int_iterator): Likewise.
15991 (apply_subst_iterator): Likewise.
15992 (record_iterator_use): Likewise.
15993 (record_attribute_use): Likewise.
15994 (md_reader::record_potential_iterator_use): Likewise. Update calls
15995 to record_iterator_use and apply_iterator.
15996 (iterator_use): Replace ptr with x and index.
15997 (attribute_use): Likewise.
15998 (apply_attribute_uses): Update calls to apply_iterator.
15999 (apply_iterators): Likewise. Update initialization of iterator_use.
16000 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
16001 and record_potential_iterator_use.
16002 (rtx_reader::read_rtx_operand): Likewise.
16004 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16005 Alan Hayward <alan.hayward@arm.com>
16006 David Sherwood <david.sherwood@arm.com>
16008 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
16011 2017-08-21 Richard Biener <rguenther@suse.de>
16013 PR middle-end/81884
16014 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
16015 at struct end conservatively when comparing common bases.
16017 2017-08-21 Richard Biener <rguenther@suse.de>
16019 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
16020 (mem_ref_in_stmt): Remove.
16021 (determine_max_movement): Use ref index to get at the reference.
16022 (invariantness_dom_walker::before_dom_children): Deal with
16023 lim data already initialized.
16024 (gather_mem_refs_stmt): Initialize lim data and record ref index.
16026 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
16028 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
16029 (TARGET_ISA_ROUND): Ditto.
16030 (TARGET_ROUND): Ditto.
16031 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
16032 * config/i386/i386.md: Ditto.
16033 * config/i386/sse.md: Ditto.
16034 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
16035 with OPTION_MASK_ISA_SSE4_1.
16037 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
16040 * doc/extend.texi (x86 Built-in Functions): Correct the name of
16041 __builtin_ia32_lzcnt_u16.
16043 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
16046 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
16047 (rs6000_set_current_function): Rewrite function to use it.
16049 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
16052 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
16053 and TYPE_WARN_IF_NOT_ALIGN.
16054 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
16055 (handle_warn_if_not_align): New.
16056 (place_union_field): Call handle_warn_if_not_align.
16057 (place_field): Call handle_warn_if_not_align.
16058 Copy TYPE_WARN_IF_NOT_ALIGN.
16059 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
16060 (layout_type): Likewise.
16061 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
16063 (tree_decl_common): Add warn_if_not_align.
16064 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
16065 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
16066 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
16067 (DECL_WARN_IF_NOT_ALIGN): Likewise.
16068 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
16069 * doc/extend.texi: Document warn_if_not_aligned attribute.
16070 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
16072 2017-08-17 Martin Liska <mliska@suse.cz>
16075 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
16076 (get_data_dependence): Use it as pointer type.
16077 (distribute_loop): Likewise.
16079 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16081 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
16082 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
16083 (p8_vmrgow_<mode>_direct): New define_insn.
16084 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
16085 handle endianness for vmrgew and vmrgow permute patterns.
16087 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
16089 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
16090 * config/rs6000/rs6000-cpus.def: Remove comment.
16091 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
16092 (POWERPC_MASKS): Likewise.
16093 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
16094 use of TARGET_VSX_TIMODE.
16095 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
16096 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
16097 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
16098 (rs6000_option_override_internal): Remove dead code.
16099 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
16100 (rs6000_legitimize_reload_address): Likewise.
16101 (rs6000_legitimate_address_p): Likewise.
16102 (rs6000_opt_masks): Delete "vsx-timode".
16103 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
16104 from function comment.
16105 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
16106 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
16107 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
16109 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
16110 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
16111 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
16112 (VSX_TI): Likewise.
16114 (define_peephole2): Likewise.
16116 2017-08-17 Martin Sebor <msebor@redhat.com>
16119 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
16120 past the end of an array.
16121 (test_pp_format): Add test cases.
16123 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
16125 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
16126 missing ECF_NOTHROW flags.
16128 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
16131 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
16132 operands residing in integer registers.
16133 (*vsx_le_perm_load_<mode>): Likewise.
16134 (*vsx_le_perm_store_<mode>): Likewise.
16135 (define_peephole2): Add peepholes to optimize the above.
16137 2017-08-17 Marek Polacek <polacek@redhat.com>
16139 PR middle-end/81814
16140 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
16141 to mimic what shorten_compare did. Change the return type to bool.
16142 (fold_cond_expr_with_comparison): Update call to
16143 operand_equal_for_comparison_p.
16144 (fold_ternary_loc): Likewise.
16146 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
16148 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
16150 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
16152 2017-08-17 Richard Biener <rguenther@suse.de>
16154 * tree-ssa-structalias.c (solve_graph): When propagating
16155 to successors update the graphs succ edges and avoid duplicate work.
16157 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
16160 * config/i386/i386.c (ix86_option_override_internal): Save target
16161 specific options after ix86_stack_protector_guard_reg was changed.
16163 2017-08-17 Richard Biener <rguenther@suse.de>
16165 PR tree-optimization/81827
16166 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
16167 (new_var_info): Initialize it conservatively.
16168 (get_call_vi): Mark register vars.
16169 (new_scalar_tmp_constraint_exp): Likewise.
16170 (handle_rhs_call): Likewise.
16171 (handle_const_call): Likewise.
16172 (create_function_info_for): Likewise.
16173 (solve_constraints): Sort varinfos to separate register from
16174 non-register vars to pack points-to solution bitmaps during
16177 2017-08-17 Marek Polacek <polacek@redhat.com>
16179 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
16181 2017-08-17 Richard Biener <rguenther@suse.de>
16183 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
16184 to true when overflow is undefined and we saturated the result.
16186 2017-08-17 Alan Modra <amodra@gmail.com>
16189 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
16190 Don't use store multiple if only one reg needs saving.
16191 (interesting_frame_related_regno): New function.
16192 (rs6000_frame_related): Don't emit frame info for regs that
16194 (rs6000_emit_epilogue): Likewise.
16196 2017-08-16 Nathan Sidwell <nathan@acm.org>
16198 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
16199 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
16200 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
16201 * tree.c (free_lang_data_in_type): Use else-if chain. Always
16202 clear TYPE_LANG_1. Remove obsolete member-function stripping.
16203 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
16204 (verify_type): Adjust for TYPE_BINFO move.
16205 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
16206 process TYPE_BINFO directly.
16207 (hash_tree): Likewise.
16208 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
16210 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
16213 2017-08-16 David Malcolm <dmalcolm@redhat.com>
16215 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
16217 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
16220 * config/i386/i386.md (*anddi_1_btr): Change predicates of
16221 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
16222 Add ix86_binary_operator_ok to insn constraint.
16223 (*iordi_1_bts): Ditto.
16224 (*xordi_1_btc): Ditto.
16225 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
16226 Update corresponding peephole2 pattern.
16230 2017-08-16 Bin Cheng <bin.cheng@arm.com>
16232 PR tree-optimization/81832
16233 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
16234 copy loop header which has IFN_LOOP_DIST_ALIAS call.
16236 2017-08-16 Marek Polacek <polacek@redhat.com>
16239 * fold-const.c (fold_indirect_ref_1): Restore original behavior
16240 regarding size_zero_node.
16242 2017-08-16 Martin Liska <mliska@suse.cz>
16245 * config.gcc: Respect previously set extra_objs in case
16248 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
16250 PR tree-optimization/81835
16251 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
16252 the phi SSA_NAME. Check that the condition in a COND_EXPR does
16253 not depend on the phi.
16255 2017-08-16 Alan Modra <amodra@gmail.com>
16257 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
16260 2017-08-16 Alan Modra <amodra@gmail.com>
16262 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
16263 (save_reg_p): ..into this. Update all callers.
16264 (first_reg_to_save): Simplify.
16266 2017-08-16 Alan Modra <amodra@gmail.com>
16268 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
16271 2017-08-15 Joseph Myers <joseph@codesourcery.com>
16275 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
16276 constant count if that count is less than 32.
16278 2017-08-15 Nathan Sidwell <nathan@acm.org>
16280 * gcc.c (execute): Emit friendlier message if inferior is killed
16281 by an external cause.
16283 2017-08-15 Richard Biener <rguenther@suse.de>
16285 PR tree-optimization/81790
16286 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
16287 CONSTRUCTORs from simplifying and VN.
16289 2017-08-14 Martin Sebor <msebor@redhat.com>
16291 * builtin-attrs.def: Add comments.
16293 2017-08-14 Martin Sebor <msebor@redhat.com>
16296 * doc/extend.texi (attribute nonstring): Document new attribute.
16298 2017-08-14 Martin Sebor <msebor@redhat.com>
16301 * tree-diagnostic.c (default_tree_printer): Handle %G.
16302 * gimple-pretty-print.h (percent_G_format): Declare new function.
16303 * gimple-pretty-print.c (percent_G_format): Define.
16304 * tree-pretty-print.c (percent_K_format): Add argument.
16306 2017-08-14 Martin Sebor <msebor@redhat.com>
16308 PR translation/79998
16309 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
16310 Remove a stray space.
16312 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
16315 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
16316 (*iordi_1_bts): Ditto.
16317 (*xordi_1_btc): Ditto.
16319 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16322 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
16323 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16325 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
16326 quoted strings, and make more translator-friendly.
16327 (darwin_rs6000_override_options): Likewise.
16328 (rs6000_option_override_internal): Likewise.
16329 (rs6000_return_in_memory): Fix overlong line.
16330 (init_cmulative_args): Use quoted strings, and make more
16331 translator-friendly.
16332 (rs6000_pass_by_reference): Fix overlong line.
16333 (def_builtin): Use quoted strings.
16334 (altivec_expand_predicate_builtin): Use quoted strings, and make
16335 more translator-friendly.
16336 (htm_expand_builtin): Use quoted strings.
16337 (cpu_expand_builtin): Use quoted strings, and make more
16338 translator-friendly.
16339 (altivec_expand_builtin): Likewise.
16340 (paired_expand_predicate_builtin): Likewise.
16341 (rs6000_invalid_builtin): Likewise.
16342 (builtin_function_type): Use quoted strings.
16343 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
16344 more translator-friendly.
16345 (rs6000_trampoline_init): Likewise.
16346 (rs6000_handle_altivec_attribute): Likewise.
16347 (rs6000_inner_target_options): Use quoted strings.
16348 (rs6000_disable_incompatible_switches): Likewise.
16349 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
16350 strings, and make more translator-friendly.
16351 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
16353 2017-08-14 Bin Cheng <bin.cheng@arm.com>
16355 PR tree-optimization/81799
16356 * tree-loop-distribution.c (version_loop_by_alias_check): Force
16357 cond_expr to simple gimple operand.
16359 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
16361 PR middle-end/46932
16362 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
16364 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
16368 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
16369 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
16370 TARGET_GASISR_PROLOGUES.
16371 * config/avr/avr.c (avr_option_override): Same.
16372 (avr_pass_pre_proep::execute): Same.
16374 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
16377 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
16378 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
16380 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
16382 * config/i386/i386.md (*load_tp_<mode>): Redefine as
16383 define_insn_and_split. Split to a memory load from 0 in
16384 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
16385 using PTR mode iterator.
16386 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
16387 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
16388 (*add_tp_<mode>): Redefine as define_insn_and_split.
16389 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
16390 address space. Merge with *add_tp_x32 using PTR mode iterator.
16391 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
16392 Split to an add with a memory load from 0 in
16393 DEFAULT_TLS_SEG_REG address space.
16395 2017-08-12 Andrew Pinski <apinski@cavium.com>
16397 * config/aarch64/aarch64-option-extensions.def (rdma):
16398 Fix feature string to what Linux prints out in /proc/cpuinfo.
16400 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
16403 * dwarf2out.c (modified_type_die): For C typedef types that have
16404 an ultimate origin, process the ultimate origin instead of the
16406 (gen_typedef_die): Assert that input DECLs have no ultimate
16408 (gen_type_die_with_usage): For typedef variants that have an
16409 ultimate origin, just call gen_decl_die on the original DECL.
16410 (process_scope_var): Avoid creating DIEs for local typedefs and
16411 concrete static variables.
16413 2017-08-12 Alan Modra <amodra@gmail.com>
16417 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
16418 match gnu-user.h startfile.
16419 (ENDFILE_LINUX_SPEC): Similarly.
16421 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
16424 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
16426 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
16428 2017-08-11 Tamar Christina <tamar.christina@arm.com>
16429 * config/aarch64/aarch64.md (mov<mode>): Change.
16430 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
16431 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
16432 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
16434 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
16436 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
16437 for storage order barriers.
16439 2017-08-11 Martin Liska <mliska@suse.cz>
16442 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
16443 variables of void type.
16445 2017-08-11 Martin Liska <mliska@suse.cz>
16447 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
16448 TARGET_SUPPORTS_ALIASES.
16449 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
16450 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
16451 (optimize_weakref): Likewise.
16452 * symtab.c (symtab_node::noninterposable_alias): Likewise.
16453 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
16454 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
16456 2017-08-11 Martin Liska <mliska@suse.cz>
16459 * config/i386/i386.c (make_resolver_func): Do complete
16460 refactoring of the function.
16462 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
16465 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
16466 * config/i386/i386.c (ix86_stack_protect_guard): Use
16467 ix86_stack_protect_guard_symbol_str to generate varible declaration.
16468 * doc/invoke.texi (x86 Options): Document
16469 -mstack-protector-guard-symbol= option.
16471 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
16473 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
16474 * config/i386/i386.c (ix86_split_stack_guard): New function.
16475 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
16476 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
16477 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
16478 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
16479 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
16480 (split_stack_space_check): Call ix86_split_stack_guard.
16482 2017-08-10 Martin Sebor <msebor@redhat.com>
16484 * print-tree.c (print_node): Print location using the established
16486 Replace spaces with colons.
16487 (debug_raw, debug): Ditto.
16489 2017-08-10 Martin Sebor <msebor@redhat.com>
16492 * pretty-print.c (pp_format): Correct the handling of %s precision.
16494 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
16497 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
16499 (ix86_finalize_stack_frame_flags): This. Also clear
16500 frame_pointer_needed if -fno-omit-frame-pointer is used without
16502 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
16503 with ix86_finalize_stack_frame_flags.
16504 (ix86_expand_epilogue): Likewise.
16505 (ix86_expand_split_stack_prologue): Likewise.
16506 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
16508 2017-08-10 Martin Liska <mliska@suse.cz>
16511 * c-attribs.c (handle_target_attribute):
16512 Report warning for an empty string argument of target attribute.
16514 2017-08-09 Jakub Jelinek <jakub@redhat.com>
16517 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
16519 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
16520 or DECL_NONLOCAL labels.
16521 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
16522 or DECL_NONLOCAL labels here.
16524 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
16526 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
16527 to indicate when early gimple folding has been disabled.
16528 (rs6000_gimple_fold_builtin): Add debug content.
16529 (rs6000_invalid_builtin): Fix whitespace.
16530 (rs6000_expand_builtin): Fix whitespace.
16531 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
16533 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
16536 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
16537 SAVE_MULTIPLE if not all the registers that saves, should be saved.
16539 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
16541 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
16542 (qdf24xx): Likewise.
16543 * config/aarch64/aarch64.md: Include falkor.md.
16544 * config/aarch64/falkor.md: New.
16546 2017-08-09 Marek Polacek <polacek@redhat.com>
16549 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
16550 * diagnostic.c (emit_diagnostic): Add a comment.
16551 (emit_diagnostic_valist): New function.
16553 2017-08-09 Marek Polacek <polacek@redhat.com>
16556 * input.c (make_location): New overload.
16557 * input.h (make_location): Declare.
16559 2017-08-08 Alan Modra <amodra@gmail.com>
16560 H.J. Lu <hongjiu.lu@intel.com>
16563 * gcc.c (NO_PIE_SPEC): Delete.
16564 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
16566 (LINK_PIE_SPEC): ..to here.
16567 (LINK_COMMAND_SPEC): Support -no-pie.
16568 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
16569 chain of crtbegin*.o selection, update for PIE_SPEC changes and
16571 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
16572 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
16573 (ENDFILE_CRTEND_SPEC): Similarly.
16575 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
16578 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
16579 (mstack-protector-guard-offset=): Ditto.
16580 * config/i386/i386.c (ix86_option_override): Handle
16581 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
16583 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
16584 ix86_stack_protect_guard_offset variables.
16585 (TARGET_STACK_PROTECT_GUARD): Always define.
16586 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
16587 and -mstack-protector-guard-offset= options.
16589 2017-08-08 Bin Cheng <bin.cheng@arm.com>
16591 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
16592 boundary case for the last candidate.
16594 2017-08-08 Bin Cheng <bin.cheng@arm.com>
16596 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
16597 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
16599 2017-08-08 Tamar Christina <tamar.christina@arm.com>
16601 PR middle-end/19706
16602 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
16603 * config/aarch64/aarch64-builtins.c
16604 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
16605 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
16606 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
16608 2017-08-08 Tamar Christina <tamar.christina@arm.com>
16609 Andrew Pinski <pinskia@gmail.com>
16611 PR middle-end/19706
16612 * internal-fn.def (XORSIGN): New.
16613 * optabs.def (xorsign_optab): New.
16614 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
16615 (convert_expand_mult_copysign): New.
16616 (pass_optimize_widening_mul::execute): Call
16617 convert_expand_mult_copysign.
16619 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16621 PR tree-optimization/81354
16622 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
16623 Insert on edges rather than explicitly creating landing pads.
16624 (analyze_candidates_and_replace): Commit edge inserts.
16626 2017-08-08 Richard Biener <rguenther@suse.de>
16628 PR middle-end/81719
16629 * tree-ssa-loop-niter.c: Include tree-dfa.h.
16630 (expand_simple_operations): Also look through ADDR_EXPRs with
16631 MEM_REF bases treating them as POINTER_PLUS_EXPR.
16633 2017-08-08 Richard Biener <rguenther@suse.de>
16635 PR tree-optimization/81723
16636 * tree-vect-slp.c (struct bst_traits): New hash traits.
16637 (bst_fail): New global.
16638 (vect_build_slp_tree_2): New worker, split out from ...
16639 (vect_build_slp_tree): ... this now wrapping it with using
16640 bst_fail set to cache SLP tree build fails. Properly handle
16642 (vect_analyze_slp_instance): Allocate and free bst_fail.
16644 2017-08-08 Martin Liska <mliska@suse.cz>
16647 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
16648 LABEL_DECLs that can be from a different function.
16650 2017-08-08 Bin Cheng <bin.cheng@arm.com>
16652 PR tree-optimization/81744
16653 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
16654 loop's number of iterations.
16656 2017-08-08 Martin Liska <mliska@suse.cz>
16658 * asan.c: Include header files.
16659 * attribs.c (build_decl_attribute_variant): New function moved
16661 (build_type_attribute_qual_variant): Likewise.
16662 (cmp_attrib_identifiers): Likewise.
16663 (simple_cst_list_equal): Likewise.
16664 (omp_declare_simd_clauses_equal): Likewise.
16665 (attribute_value_equal): Likewise.
16666 (comp_type_attributes): Likewise.
16667 (build_type_attribute_variant): Likewise.
16668 (lookup_ident_attribute): Likewise.
16669 (remove_attribute): Likewise.
16670 (merge_attributes): Likewise.
16671 (merge_type_attributes): Likewise.
16672 (merge_decl_attributes): Likewise.
16673 (merge_dllimport_decl_attributes): Likewise.
16674 (handle_dll_attribute): Likewise.
16675 (attribute_list_equal): Likewise.
16676 (attribute_list_contained): Likewise.
16677 * attribs.h (lookup_attribute): New function moved from tree.[ch].
16678 (lookup_attribute_by_prefix): Likewise.
16679 * bb-reorder.c: Include header files.
16680 * builtins.c: Likewise.
16681 * calls.c: Likewise.
16682 * cfgexpand.c: Likewise.
16683 * cgraph.c: Likewise.
16684 * cgraphunit.c: Likewise.
16685 * convert.c: Likewise.
16686 * dwarf2out.c: Likewise.
16687 * final.c: Likewise.
16688 * fold-const.c: Likewise.
16689 * function.c: Likewise.
16690 * gimple-expr.c: Likewise.
16691 * gimple-fold.c: Likewise.
16692 * gimple-pretty-print.c: Likewise.
16693 * gimple.c: Likewise.
16694 * gimplify.c: Likewise.
16695 * hsa-common.c: Likewise.
16696 * hsa-gen.c: Likewise.
16697 * internal-fn.c: Likewise.
16698 * ipa-chkp.c: Likewise.
16699 * ipa-cp.c: Likewise.
16700 * ipa-devirt.c: Likewise.
16701 * ipa-fnsummary.c: Likewise.
16702 * ipa-inline.c: Likewise.
16703 * ipa-visibility.c: Likewise.
16705 * lto-cgraph.c: Likewise.
16706 * omp-expand.c: Likewise.
16707 * omp-general.c: Likewise.
16708 * omp-low.c: Likewise.
16709 * omp-offload.c: Likewise.
16710 * omp-simd-clone.c: Likewise.
16711 * opts-global.c: Likewise.
16712 * passes.c: Likewise.
16713 * predict.c: Likewise.
16714 * sancov.c: Likewise.
16715 * sanopt.c: Likewise.
16716 * symtab.c: Likewise.
16717 * toplev.c: Likewise.
16718 * trans-mem.c: Likewise.
16719 * tree-chkp.c: Likewise.
16720 * tree-eh.c: Likewise.
16721 * tree-into-ssa.c: Likewise.
16722 * tree-object-size.c: Likewise.
16723 * tree-parloops.c: Likewise.
16724 * tree-profile.c: Likewise.
16725 * tree-ssa-ccp.c: Likewise.
16726 * tree-ssa-live.c: Likewise.
16727 * tree-ssa-loop.c: Likewise.
16728 * tree-ssa-sccvn.c: Likewise.
16729 * tree-ssa-structalias.c: Likewise.
16730 * tree-ssa.c: Likewise.
16731 * tree-streamer-in.c: Likewise.
16732 * tree-vectorizer.c: Likewise.
16733 * tree-vrp.c: Likewise.
16734 * tsan.c: Likewise.
16735 * ubsan.c: Likewise.
16736 * varasm.c: Likewise.
16737 * varpool.c: Likewise.
16738 * tree.c: Remove functions moved to attribs.[ch].
16739 * tree.h: Likewise.
16740 * config/aarch64/aarch64.c: Add attrs.h header file.
16741 * config/alpha/alpha.c: Likewise.
16742 * config/arc/arc.c: Likewise.
16743 * config/arm/arm.c: Likewise.
16744 * config/avr/avr.c: Likewise.
16745 * config/bfin/bfin.c: Likewise.
16746 * config/c6x/c6x.c: Likewise.
16747 * config/cr16/cr16.c: Likewise.
16748 * config/cris/cris.c: Likewise.
16749 * config/darwin.c: Likewise.
16750 * config/epiphany/epiphany.c: Likewise.
16751 * config/fr30/fr30.c: Likewise.
16752 * config/frv/frv.c: Likewise.
16753 * config/ft32/ft32.c: Likewise.
16754 * config/h8300/h8300.c: Likewise.
16755 * config/i386/winnt.c: Likewise.
16756 * config/ia64/ia64.c: Likewise.
16757 * config/iq2000/iq2000.c: Likewise.
16758 * config/lm32/lm32.c: Likewise.
16759 * config/m32c/m32c.c: Likewise.
16760 * config/m32r/m32r.c: Likewise.
16761 * config/m68k/m68k.c: Likewise.
16762 * config/mcore/mcore.c: Likewise.
16763 * config/microblaze/microblaze.c: Likewise.
16764 * config/mips/mips.c: Likewise.
16765 * config/mmix/mmix.c: Likewise.
16766 * config/mn10300/mn10300.c: Likewise.
16767 * config/moxie/moxie.c: Likewise.
16768 * config/msp430/msp430.c: Likewise.
16769 * config/nds32/nds32-isr.c: Likewise.
16770 * config/nds32/nds32.c: Likewise.
16771 * config/nios2/nios2.c: Likewise.
16772 * config/nvptx/nvptx.c: Likewise.
16773 * config/pa/pa.c: Likewise.
16774 * config/pdp11/pdp11.c: Likewise.
16775 * config/powerpcspe/powerpcspe.c: Likewise.
16776 * config/riscv/riscv.c: Likewise.
16777 * config/rl78/rl78.c: Likewise.
16778 * config/rx/rx.c: Likewise.
16779 * config/s390/s390.c: Likewise.
16780 * config/sh/sh.c: Likewise.
16781 * config/sol2.c: Likewise.
16782 * config/sparc/sparc.c: Likewise.
16783 * config/spu/spu.c: Likewise.
16784 * config/stormy16/stormy16.c: Likewise.
16785 * config/tilegx/tilegx.c: Likewise.
16786 * config/tilepro/tilepro.c: Likewise.
16787 * config/v850/v850.c: Likewise.
16788 * config/vax/vax.c: Likewise.
16789 * config/visium/visium.c: Likewise.
16790 * config/xtensa/xtensa.c: Likewise.
16792 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16795 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
16796 constraints since the -mupper-regs-* switches have been
16798 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
16799 into a vector from a double word element that was extracted from
16800 another vector, and eliminate extra XXPERMDI instructions.
16801 (vsx_concat_<mode>_2): Likewise.
16802 (vsx_concat_<mode>_3): Likewise.
16803 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
16804 concat to allow optimizing inserts from previous extracts.
16806 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
16808 * config/i386/i386.c (ix86_stack_protect_guard): Generate
16809 memory reference to a SSP offset in TLS address space.
16810 (ix86_print_operand) <case '@'>: Remove.
16811 (ix86_print_operand_punct_valid_p): Remove '@' code.
16812 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
16813 UNSPEC_SP_TLS_TEST.
16814 (stack_tls_protect_set_<mode>): Remove.
16815 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
16816 (stack_tls_protect_test_<mode>): Remove.
16817 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
16819 2017-08-07 Olivier Hainque <hainque@adacore.com>
16822 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
16824 2017-08-07 Douglas Rupp <rupp@adacore.com>
16826 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
16827 variable was referenced as multidir in command.
16829 2017-08-07 Jakub Jelinek <jakub@redhat.com>
16832 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
16835 2017-08-07 Martin Liska <mliska@suse.cz>
16837 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
16838 * config/rl78/rl78.c: Add include of attribs.h.
16839 * config/sh/sh.c: Likewise.
16840 * config/v850/v850.c: Likewise.
16842 2017-08-07 Tom de Vries <tom@codesourcery.com>
16844 PR middle-end/78266
16845 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
16847 2017-08-07 Martin Liska <mliska@suse.cz>
16849 * config/mips/mips.c: Include attribs.h.
16851 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
16854 * doc/invoke.texi: Document change in behvaior for -Ofast for
16857 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
16859 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
16861 (aarch64_pop_regs): Likewise.
16862 (aarch64_gen_load_pair): Likewise.
16863 (aarch64_save_callee_saves): Likewise.
16864 (aarch64_restore_callee_saves): Likewise.
16866 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
16868 * config/i386/i386.c: Revert the last change.
16870 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
16873 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
16875 (ix86_finalize_stack_frame_flags): This. Also clear
16876 frame_pointer_needed if -fno-omit-frame-pointer is used without
16878 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
16879 with ix86_finalize_stack_frame_flags.
16880 (ix86_expand_epilogue): Likewise.
16881 (ix86_expand_split_stack_prologue): Likewise.
16883 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
16886 * config/i386/i386.c (get_builtin_code_for_version): Set priority
16887 to P_AES for Westmere.
16889 2017-08-07 Jonathan Yong <10walls@gmail.com>
16891 * config/i386/mingw.opt (fset-stack-executable): Removed.
16892 * config/i386/cygming.opt (fset-stack-executable): Moved
16894 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
16896 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
16898 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
16900 2017-08-07 Marek Polacek <polacek@redhat.com>
16902 PR middle-end/81737
16903 * fold-const.c (fold_indirect_ref_1): Check type_domain.
16905 2017-08-07 Martin Liska <mliska@suse.cz>
16907 * attribs.h (canonicalize_attr_name): New function.
16908 (cmp_attribs): Move from c-format.c and adjusted.
16909 (is_attribute_p): Moved from tree.h.
16910 * tree-inline.c: Add new includes.
16911 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
16912 (private_is_attribute_p): Remove.
16913 (private_lookup_attribute): Likewise.
16914 (private_lookup_attribute_by_prefix): Simplify.
16915 (remove_attribute): Use is_attribute_p.
16916 * tree.h: Remove removed declarations.
16918 2017-08-07 Jakub Jelinek <jakub@redhat.com>
16920 PR middle-end/81698
16921 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
16922 instead of computing it in the function. Formatting fix.
16923 (expand_case): Don't rely on default_edge being the first edge,
16924 clear it if removing it, pass default_edge to
16925 emit_case_dispatch_table.
16926 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
16929 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
16931 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
16932 insn in the function, emit NOP after the insn.
16934 2017-08-06 Tom de Vries <tom@codesourcery.com>
16936 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
16939 2017-08-06 Tom de Vries <tom@codesourcery.com>
16941 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
16944 2017-08-04 Yury Gribov <tetra2005@gmail.com>
16946 PR tree-optimization/57371
16947 * match.pd: New pattern.
16949 2017-08-04 Marek Polacek <polacek@redhat.com>
16951 PR middle-end/81695
16952 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
16953 perform the computation in offset_int.
16955 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
16957 PR tree-optimization/81136
16958 * tree-vectorizer.h: Include tree-hash-traits.h.
16959 (vec_base_alignments): New typedef.
16960 (vec_info): Add a base_alignments field.
16961 (vect_record_base_alignments): Declare.
16962 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
16964 (DR_IS_CONDITIONAL_IN_STMT): New macro.
16965 (create_data_ref): Add an is_conditional_in_stmt argument.
16966 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
16967 the is_conditional_in_stmt field.
16968 (data_ref_loc): Add an is_conditional_in_stmt field.
16969 (get_references_in_stmt): Set the is_conditional_in_stmt field.
16970 (find_data_references_in_stmt): Update call to create_data_ref.
16971 (graphite_find_data_references_in_stmt): Likewise.
16972 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
16973 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
16974 (vect_record_base_alignment): New function.
16975 (vect_record_base_alignments): Likewise.
16976 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
16977 for nested statements even if we fail to compute a misalignment.
16978 Use pooled base alignments for unconditional references.
16979 (vect_find_same_alignment_drs): Compare base addresses instead
16981 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
16982 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
16984 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
16986 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
16987 Add an explicit name for the enum. Use auto_vec for slp_instances
16988 and grouped_stores.
16989 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
16991 (_bb_vec_info): Add a constructor and destructor.
16992 (vinfo_for_stmt): Return NULL for uids of -1 as well.
16993 (destroy_loop_vec_info): Delete.
16994 (vect_destroy_datarefs): Likewise.
16995 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
16996 (vec_info::vec_info): New function.
16997 (vec_info::~vec_info): Likewise.
16998 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
16999 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
17000 destroy_loop_vec_info.
17001 * tree-vect-loop.c (new_loop_vec_info): Replace with...
17002 (_loop_vec_info::_loop_vec_info): ...this.
17003 (destroy_loop_vec_info): Replace with...
17004 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
17005 the stmt_vec_infos. Leave handling of vec_info information to its
17006 destructor. Remove explicit vector releases.
17007 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
17008 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
17009 * tree-vect-slp.c (new_bb_vec_info): Replace with...
17010 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
17011 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
17012 (destroy_bb_vec_info): Replace with...
17013 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
17014 information to its destructor.
17015 (vect_slp_analyze_bb_1): Use new and delete instead of
17016 new_bb_vec_info and destroy_bb_vec_info.
17017 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
17020 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
17022 * tree-data-ref.h (subscript): Add access_fn field.
17023 (data_dependence_relation): Add could_be_independent_p.
17024 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
17025 (same_access_functions): Move to tree-data-ref.c.
17026 * tree-data-ref.c (ref_contains_union_access_p): New function.
17027 (access_fn_component_p): Likewise.
17028 (access_fn_components_comparable_p): Likewise.
17029 (dr_analyze_indices): Add a reference to access_fn_component_p.
17030 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
17032 (constant_access_functions): Likewise.
17033 (add_other_self_distances): Likewise.
17034 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
17035 (initialize_data_dependence_relation): Use XCNEW and remove
17036 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
17037 of access functions that have the same type. Allow the
17038 subsequence to end with different bases in some circumstances.
17039 Record the chosen access functions in SUB_ACCESS_FN.
17040 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
17041 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
17042 (subscript_dependence_tester_1): Likewise dra and drb.
17043 (build_classic_dist_vector): Update calls accordingly.
17044 (subscript_dependence_tester): Likewise.
17045 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
17046 DDR_COULD_BE_INDEPENDENT_P.
17047 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
17048 comp_alias_ddrs instead of may_alias_ddrs.
17049 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
17051 (vect_analyze_data_ref_dependence): Use it if
17052 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
17053 distance vectors if that fails.
17054 (dependence_distance_ge_vf): New function.
17055 (vect_prune_runtime_alias_test_list): Use it. Don't clear
17056 LOOP_VINFO_MAY_ALIAS_DDRS.
17058 2017-08-04 Richard Biener <rguenther@suse.de>
17060 PR middle-end/81705
17061 * fold-const.c (fold_binary_loc): Properly restrict
17062 minus_var0 && minus_var1 case when associating undefined overflow
17065 2017-08-04 Tom de Vries <tom@codesourcery.com>
17067 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
17069 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17071 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17072 Don't start diagnostic messages with a capital letter.
17073 * config/rs6000/rs6000.c (rs6000_option_override_internal):
17075 (rs6000_invalid_builtin): Likewise.
17076 (rs6000_trampoline_init): Likewise.
17078 2017-08-03 Jakub Jelinek <jakub@redhat.com>
17081 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
17082 after setting changeable df flags.
17084 2017-08-03 Richard Biener <rguenther@suse.de>
17086 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
17087 up if the use is in USE - X.
17089 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
17091 * toplev.c (dumpfile.h): New include.
17092 (internal_error_reentered): New static function. Use it...
17093 (internal_error_function): ...here to handle reentered internal_error.
17095 2017-08-03 Richard Biener <rguenther@suse.de>
17097 PR middle-end/81148
17098 * fold-const.c (split_tree): Add minus_var and minus_con
17099 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
17100 here but always use minus_*.
17101 (associate_trees): Assert we never associate with MINUS_EXPR
17102 and NULL first operand. Do not recurse for PLUS_EXPR operands
17103 when associating as MINUS_EXPR either.
17104 (fold_binary_loc): Track minus_var and minus_con.
17106 2017-08-03 Tom de Vries <tom@codesourcery.com>
17109 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
17110 ACCEL_COMPILER, apply finish_options on
17111 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
17113 2017-08-03 Tom de Vries <tom@codesourcery.com>
17116 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
17117 function_entry_patch_area_size > 0.
17119 2017-08-03 Jakub Jelinek <jakub@redhat.com>
17122 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
17123 instead of 10??LU, perform unit multiplication in wide_int,
17124 don't change alloc_object_size_limit if the limit is larger
17127 PR tree-optimization/81655
17128 PR tree-optimization/81588
17129 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
17130 the case when ranges[i].low and high are 1 for unsigned type with
17133 PR middle-end/81052
17134 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
17135 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
17137 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17139 * tree-vrp.h: Add include guard.
17141 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
17144 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
17145 (ud2): New insn pattern.
17146 * config/i386/i386.c (ix86_expand_epilogue):
17147 For naked functions, generate ud2 instead of trap insn.
17149 2017-08-02 Marek Polacek <polacek@redhat.com>
17152 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
17154 2017-08-02 Tom de Vries <tom@codesourcery.com>
17155 Cesar Philippidis <cesar@codesourcery.com>
17157 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
17158 Add missing edge probabilities.
17160 2017-08-02 Tamar Christina <tamar.christina@arm.com>
17162 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
17163 Correct endianness.
17165 2017-08-02 Jakub Jelinek <jakub@redhat.com>
17167 PR middle-end/79499
17168 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
17169 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
17170 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
17171 prologue_seq sequences - if any.
17173 2017-08-02 Richard Biener <rguenther@suse.de>
17175 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
17176 via vectors if supported, integer extracts via punning if supported
17177 or otherwise vector extracts.
17179 2017-08-02 Richard Biener <rguenther@suse.de>
17181 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
17183 (bitmap_insert_into_set): ... this.
17185 2017-08-02 Richard Biener <rguenther@suse.de>
17187 PR tree-optimization/81633
17189 2015-08-17 Alan Hayward <alan.hayward@arm.com>
17191 PR tree-optimization/71752
17192 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
17194 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
17196 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
17197 (machine_function::call_ms2sysv_pad_out): Remove field.
17198 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
17199 (ix86_compute_frame_layout): Likewise.
17201 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
17204 * config/i386/i386.c (ix86_set_func_type): Disallow naked
17205 attribute with interrupt attribute.
17207 2017-08-01 Andrew Pinski <apinski@cavium.com>
17209 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
17210 BIT_INSERT_EXPR's operand 1
17211 to see if the types precision matches.
17213 2017-08-01 Martin Liska <mliska@suse.cz>
17215 PR middle-end/70140
17216 * builtins.c (expand_builtin_memcpy_args): Remove.
17217 (expand_builtin_memcpy): Call newly added function
17218 expand_builtin_memory_copy_args.
17219 (expand_builtin_memcpy_with_bounds): Likewise.
17220 (expand_builtin_mempcpy): Remove last argument.
17221 (expand_builtin_mempcpy_with_bounds): Likewise.
17222 (expand_builtin_memory_copy_args): New function created from
17223 expand_builtin_mempcpy_args with small modifications.
17224 (expand_builtin_mempcpy_args): Remove.
17225 (expand_builtin_stpcpy): Remove unused argument.
17226 (expand_builtin): Likewise.
17227 (expand_builtin_with_bounds): Likewise.
17229 2017-08-01 Martin Liska <mliska@suse.cz>
17232 Make mempcpy more optimal (PR middle-end/70140).
17234 2017-08-01 Jakub Jelinek <jakub@redhat.com>
17237 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
17238 __builtin_vec_cmpne verify both arguments are compatible vectors
17239 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
17240 verify arg1_type is a pointer or array type. For __builtin_vec_st,
17241 move computation of aligned to after checking the argument types.
17245 * config/rs6000/vsx.md (vextract_fp_from_shorth,
17246 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
17249 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
17250 Doug Rupp <rupp@adacore.com>
17251 Olivier Hainque <hainque@adacore.com>
17253 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
17254 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
17256 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
17257 for TARGET_OS_CPP_BUILTIN.
17258 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
17259 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
17261 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
17262 passing required abi options to the assembler for EABI configurations.
17263 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
17264 of .text.hot and .text.unlikely sections for kernel modules when
17265 using ARM style exceptions.
17266 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
17267 options. Add EXTRA_CC1_SPEC.
17268 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
17270 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
17272 (ARM_TARGET2_DWARF_FORMAT): Define.
17273 * config/arm/t-vxworks: Adjust multilib control to removal of the
17274 Diab command line options.
17276 2017-08-01 Martin Liska <mliska@suse.cz>
17278 PR gcov-profile/81561
17279 * gcov.c (unblock): Make unblocking safe as we need to preserve
17280 index correspondence of blocks and block_lists.
17282 2017-08-01 Richard Biener <rguenther@suse.de>
17284 PR tree-optimization/81181
17285 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
17286 (compute_antic): ... end of iteration here.
17288 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
17290 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
17291 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
17292 (ftree-slp-vectorize): Likewise.
17293 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
17294 can no longer be set independent of flag_tree_loop_vectorize.
17295 * omp-general.c (emp_max_vf): Likewise.
17296 * opts.c (enable_fdo_optimizations): Remove references to
17297 flag_tree_vectorize, these are now implicit.
17298 (common_handle_option): Remove handling for OPT_ftree_vectorize,
17299 and leave it for the options machinery.
17301 2017-08-01 Martin Liska <mliska@suse.cz>
17303 PR middle-end/70140
17304 * builtins.c (expand_builtin_memcpy_args): Remove.
17305 (expand_builtin_memcpy): Call newly added function
17306 expand_builtin_memory_copy_args.
17307 (expand_builtin_memcpy_with_bounds): Likewise.
17308 (expand_builtin_mempcpy): Remove last argument.
17309 (expand_builtin_mempcpy_with_bounds): Likewise.
17310 (expand_builtin_memory_copy_args): New function created from
17311 expand_builtin_mempcpy_args with small modifications.
17312 (expand_builtin_mempcpy_args): Remove.
17313 (expand_builtin_stpcpy): Remove unused argument.
17314 (expand_builtin): Likewise.
17315 (expand_builtin_with_bounds): Likewise.
17317 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
17320 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
17321 print "ds:" only for immediates in generic address space.
17323 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
17326 * config/i386/i386.c (ix86_funciton_naked): New prototype.
17327 (ix86_function_ok_for_sibcall): Return false for naked functions.
17329 2017-08-01 Richard Biener <rguenther@suse.de>
17331 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
17332 (compute_antic): Seed worklist with exit block predecessors.
17333 * cfganal.c (dfs_find_deadend): For a cycle return the source
17334 of the edge closing it.
17336 2017-08-01 Tamar Christina <tamar.christina@arm.com>
17338 * config/aarch64/aarch64.c
17339 (aarch64_can_const_movi_rtx_p): Move 0 check.
17341 2017-08-01 Bin Cheng <bin.cheng@arm.com>
17343 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
17344 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
17346 * match.pd: Ditto in address comparison pattern.
17348 2017-08-01 Bin Cheng <bin.cheng@arm.com>
17350 PR tree-optimization/81627
17351 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
17352 closed ssa form for store-store chain.
17354 2017-08-01 Bin Cheng <bin.cheng@arm.com>
17356 PR tree-optimization/81620
17357 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
17358 for store-store chain.
17360 2017-08-01 Jakub Jelinek <jakub@redhat.com>
17362 PR tree-optimization/81588
17363 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
17364 ranges[i].in_p, invert comparison code ccode. For >/>=,
17365 swap rhs1 and rhs2 and comparison code unconditionally,
17366 for </<= don't do that. Don't swap rhs1/rhs2 again if
17367 ranges[i].in_p, instead invert comparison code ccode if
17368 opcode or oe->rank is BIT_IOR_EXPR.
17371 * optabs.def (vec_extract_optab, vec_init_optab): Change from
17372 a direct optab to conversion optab.
17373 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
17374 with GET_MODE_INNER as last argument instead of optab_handler.
17375 * expmed.c (extract_bit_field_1): Likewise. Use vector from
17376 vector extraction if possible and optab is available.
17377 * expr.c (store_constructor): Use convert_optab_handler instead
17378 of optab_handler. Use vector initialization from smaller
17379 vectors if possible and optab is available.
17380 * tree-vect-stmts.c (vectorizable_load): Likewise.
17381 * doc/md.texi (vec_extract, vec_init): Document that the optabs
17382 now have two modes.
17383 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
17384 of vec_init from half-sized vectors with the same element mode.
17385 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
17386 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
17387 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
17388 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
17389 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
17390 after mode in gen_vec_extract* calls.
17391 (vec_extract<mode>): Renamed to ...
17392 (vec_extract<mode><ssescalarmodelower>): ... this.
17393 (vec_extract<mode><ssehalfvecmodelower>): New expander.
17394 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
17395 element mode after mode in gen_vec_init* calls.
17396 (VEC_INIT_HALF_MODE): New mode iterator.
17397 (vec_init<mode>): Renamed to ...
17398 (vec_init<mode><ssescalarmodelower>): ... this.
17399 (vec_init<mode><ssehalfvecmodelower>): New expander.
17400 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
17401 (vec_extractv2sfsf): ... this.
17402 (vec_initv2sf): Renamed to ...
17403 (vec_initv2sfsf): ... this.
17404 (vec_extractv2si): Renamed to ...
17405 (vec_extractv2sisi): ... this.
17406 (vec_initv2si): Renamed to ...
17407 (vec_initv2sisi): ... this.
17408 (vec_extractv4hi): Renamed to ...
17409 (vec_extractv4hihi): ... this.
17410 (vec_initv4hi): Renamed to ...
17411 (vec_initv4hihi): ... this.
17412 (vec_extractv8qi): Renamed to ...
17413 (vec_extractv8qiqi): ... this.
17414 (vec_initv8qi): Renamed to ...
17415 (vec_initv8qiqi): ... this.
17416 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
17417 (vec_init<mode>): Renamed to ...
17418 (vec_init<mode><VEC_base_l>): ... this.
17419 (vec_extract<mode>): Renamed to ...
17420 (vec_extract<mode><VEC_base_l>): ... this.
17421 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
17422 (vec_initv2sfsf): ... this.
17423 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
17424 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
17425 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
17426 element mode after mode in gen_vec_init* calls.
17427 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
17428 (vec_init<mode><Vel>): ... this.
17429 (vec_extract<mode>): Renamed to ...
17430 (vec_extract<mode><Vel>): ... this.
17431 * config/aarch64/iterators.md (Vel): New mode attribute.
17432 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
17433 Add element mode after mode in gen_vec_extract* calls.
17434 * config/s390/vector.md (non_vec_l): New mode attribute.
17435 (vec_extract<mode>): Renamed to ...
17436 (vec_extract<mode><non_vec_l>): ... this.
17437 (vec_init<mode>): Renamed to ...
17438 (vec_init<mode><non_vec_l>): ... this.
17439 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
17440 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
17442 * config/arm/iterators.md (V_elem_l): New mode attribute.
17443 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
17444 (vec_extract<mode><V_elem_l>): ... this.
17445 (vec_extractv2di): Renamed to ...
17446 (vec_extractv2didi): ... this.
17447 (vec_init<mode>): Renamed to ...
17448 (vec_init<mode><V_elem_l>): ... this.
17449 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
17450 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
17451 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
17452 Add element mode after gen_vec_extract* calls.
17453 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
17454 (vec_init<mode><unitmode>): ... this.
17455 (vec_extract<mode>): Renamed to ...
17456 (vec_extract<mode><unitmode>): ... this.
17457 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
17458 (vec_init<mode><unitmode>): ... this.
17459 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
17460 (vec_initv2sfsf): ... this.
17461 (vec_extractv2sf): Renamed to ...
17462 (vec_extractv2sfsf): ... this.
17463 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
17464 Add element mode after gen_vec_extract* calls.
17465 * config/mips/mips.md (unitmode): New mode iterator.
17466 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
17467 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
17468 * config/spu/spu.md (inner_l): New mode attribute.
17469 (vec_init<mode>): Renamed to ...
17470 (vec_init<mode><inner_l>): ... this.
17471 (vec_extract<mode>): Renamed to ...
17472 (vec_extract<mode><inner_l>): ... this.
17473 * config/sparc/sparc.md (veltmode): New mode iterator.
17474 (vec_init<VMALL:mode>): Renamed to ...
17475 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
17476 * config/ia64/vect.md (vec_initv2si): Renamed to ...
17477 (vec_initv2sisi): ... this.
17478 (vec_initv2sf): Renamed to ...
17479 (vec_initv2sfsf): ... this.
17480 (vec_extractv2sf): Renamed to ...
17481 (vec_extractv2sfsf): ... this.
17482 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
17483 (vec_init<mode>): Renamed to ...
17484 (vec_init<mode><VEC_base_l>): ... this.
17485 (vec_extract<mode>): Renamed to ...
17486 (vec_extract<mode><VEC_base_l>): ... this.
17487 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
17488 (vec_initv2sfsf): ... this.
17489 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
17490 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
17491 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
17492 gen_vec_init* calls.
17494 2017-08-01 Richard Biener <rguenther@suse.de>
17496 PR tree-optimization/81297
17497 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
17498 TREE_OVERFLOW from INTEGER_CSTs.
17500 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
17502 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
17504 2017-07-31 Carl Love <cel@us.ibm.com>
17506 * config/rs6000/rs6000-c: Add support for built-in functions
17507 vector signed char vec_xl_be (signed long long, signed char *);
17508 vector unsigned char vec_xl_be (signed long long, unsigned char *);
17509 vector signed int vec_xl_be (signed long long, signed int *);
17510 vector unsigned int vec_xl_be (signed long long, unsigned int *);
17511 vector signed long long vec_xl_be (signed long long, signed long long *);
17512 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
17513 vector signed short vec_xl_be (signed long long, signed short *);
17514 vector unsigned short vec_xl_be (signed long long, unsigned short *);
17515 vector double vec_xl_be (signed long long, double *);
17516 vector float vec_xl_be (signed long long, float *);
17517 * config/rs6000/altivec.h (vec_xl_be): Add #define.
17518 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
17519 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
17521 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
17522 (altivec_expand_builtin): Add switch statement to call
17523 altivec_expand_xl_be for each builtin.
17524 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
17525 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
17526 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
17527 __builtin_vsx_le_be_v16qi.
17528 * doc/extend.texi: Update the built-in documentation file for the
17529 new built-in functions.
17531 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
17534 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
17536 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17538 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17540 * config.gcc: Add z14.
17541 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
17542 CPU model numbers for z13s and z14.
17543 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
17545 * config/s390/s390-opts.h (enum processor_type): Rename
17546 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
17547 * config/s390/s390.c (processor_table): Add field for CPU name to
17548 be passed to Binutils.
17549 (s390_asm_output_machine_for_arch): Use the new field in
17550 processor_table for Binutils.
17551 (s390_expand_builtin): Replace arch12 with z14.
17552 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
17553 (s390_get_sched_attrmask): Likewise.
17554 (s390_get_unit_mask): Likewise.
17555 * config/s390/s390.opt: Add z14 to processor_type enum.
17557 2017-07-31 Martin Jambor <mjambor@suse.cz>
17560 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
17561 regardless of optimization level.
17563 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
17564 Martin Liska <mliska@suse.cz>
17566 * predict.def: Remove old comment and adjust probability.
17567 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
17568 PREDICT statements.
17570 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
17573 * config/i386/i386.c (ix86_function_naked): New function.
17574 (ix86_can_use_return_insn_p): Return false for naked functions.
17575 (ix86_expand_prologue): Skip prologue for naked functions.
17576 (ix86_expand_epilogue): Skip epilogue for naked functions
17577 and emit trap instruction.
17578 (ix86_warn_func_return): New function.
17579 (ix86_attribute_table): Add "naked" attribute specification.
17580 (TARGET_WARN_FUNC_RETURN): Define.
17581 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
17583 2017-07-31 Martin Liska <mliska@suse.cz>
17585 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
17586 (dump_gimple_bb_header): Always dump BB info.
17587 (pp_cfg_jump): Do not append info about BB when dumping a jump.
17589 2017-07-31 Martin Liska <mliska@suse.cz>
17592 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
17593 also with current_function_decl non-null equality.
17595 2017-07-31 Jakub Jelinek <jakub@redhat.com>
17598 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
17599 change type to the element type, instead add eltype variable and
17600 use it where we are interested in the element type.
17602 PR tree-optimization/81603
17603 * ipa-polymorphic-call.c
17604 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
17605 offset arithmetic in offset_int, bail out if the resulting bit offset
17606 doesn't fit into shwi.
17608 2017-07-31 Martin Liska <mliska@suse.cz>
17610 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
17611 (gimplify_save_expr): Fix comment.
17613 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
17616 * config/i386/i386.c (ix86_function_arg): Update arguments for
17618 (ix86_compute_frame_layout): Set the initial stack offset to
17619 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
17620 INCOMING_FRAME_SP_OFFSET.
17621 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
17622 stack before exception handler returns.
17623 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
17624 the 'ERROR_CODE' for exception handler.
17626 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
17628 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
17629 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
17630 (ASM_OUTPUT_REG_POP): Ditto.
17631 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
17632 instead of asm_fprintf to output pure string.
17634 2017-07-29 Jakub Jelinek <jakub@redhat.com>
17636 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
17637 to imported_module_or_decl hook.
17638 (debug_nothing_tree_tree_tree_bool): Remove.
17639 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
17640 * debug.c (do_nothing_debug_hooks): Use
17641 debug_nothing_tree_tree_tree_bool_bool instead of
17642 debug_nothing_tree_tree_tree_bool.
17643 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
17644 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
17645 * sdbout.c (sdb_debug_hooks): Likewise.
17646 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
17647 (gen_namespace_die): Add DW_AT_export_symbols attribute if
17649 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
17650 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
17651 attribute, don't add anything.
17653 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17655 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
17656 (fold_build2_stat_loc): Likewise.
17657 (fold_build3_stat_loc): Likewise.
17658 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
17659 (fold_build1_loc): Remove macro.
17660 (fold_build2_loc): Likewise.
17661 (fold_build3_loc): Likewise.
17663 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17665 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
17666 (gimple_build_debug_bind_source_stat): Likewise.
17667 * gimple.h (gimple_build_debug_bind): Remove macro.
17668 (gimple_build_debug_bind_source): Likewise.
17670 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17672 * bitmap.c (bitmap_alloc): Adjust.
17673 (bitmap_gc_alloc): Likewise.
17674 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
17676 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17678 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
17679 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
17680 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
17681 (bitmap_gc_alloc_stat): Likewise.
17682 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
17684 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17686 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
17687 * rtl.h (shallow_copy_rtx): Remove macro.
17689 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17691 * emit-rtl.c (gen_raw_REG): Adjust.
17692 * gengenrtl.c (gendef): Likewise.
17693 * rtl.c (rtx_alloc_stat): Remove _stat from name.
17694 * rtl.h (rtx_alloc): Remove macro.
17696 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17698 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
17699 (build_tree_list_stat): Likewise.
17700 * tree.h (build_tree_list): Remove macro.
17701 (build_tree_list_vec): Likewise.
17703 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17705 * tree.c (make_vector_stat): Remove _stat from name.
17706 (build_vector_stat): Likewise.
17707 * tree.h (make_vector_stat): Remove macro.
17708 (build_vector_stat): Likewise.
17710 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17712 * tree.h (build_var_debug_value): Remove prototype.
17714 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17716 * tree.c (tree_cons_stat): Remove _stat from name.
17717 * tree.h (tree_cons): Remove macro.
17719 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17721 * tree.c (build_vl_exp_stat): Remove _stat from name.
17722 * tree.h (build_vl_exp): Remove macro.
17724 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17726 * tree.c (build_decl_stat): Remove _stat from name.
17727 * tree.h (build_decl): Remove macro.
17729 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17731 * gimple.c (gimple_build_with_ops_stat): Adjust.
17732 (gimple_alloc_stat): Remove _stat from name.
17733 * gimple.h (gimple_alloc): Remove macro.
17735 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17737 * tree.c (make_tree_vec_stat): Remove _stat from name.
17738 (grow_tree_vec_stat): Likewise.
17739 * tree.h (make_tree_vec_stat): Adjust prototype.
17740 (grow_tree_vec_stat): Likewise.
17741 (make_tree_vec): Remove macro.
17742 (grow_tree_vec): Likewise.
17744 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17746 * fold-const.c (fold_build1_stat_loc): Adjust.
17747 (fold_build2_stat_loc): Likewise.
17748 (fold_build3_stat_loc): Likewise.
17749 * tree.c (build0_stat): Remove _stat from name.
17750 (build1_stat): Likewise.
17751 (build2_stat): Likewise.
17752 (build3_stat): Likewise.
17753 (build4_stat): Likewise.
17754 (build5_stat): Likewise.
17755 * tree.h (build1_loc): Remove macro, and rename _stat function
17757 (build2_loc): Likewise.
17758 (build3_loc): Likewise.
17759 (build4_loc): Likewise.
17760 (build5_loc): Likewise.
17762 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17764 * tree.c (make_int_cst_stat): Remove _stat from name.
17765 * tree.h (make_int_cst_stat): Adjust prototype.
17766 (make_int_cst): Remove macro.
17768 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17770 * tree.c (make_tre_binfo_stat): Remove _stat from name.
17771 * tree.h (make_tree_binfo_stat): Adjust prototype.
17772 (make_tree_binfo): Remove.
17774 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17776 * tree.c (copy_node_stat): Rename to copy_node.
17777 (build_distinct_type_copy): Adjust.
17778 * tree.h (copy_node_stat): Adjust prototype.
17779 (copy_node): Remove macro.
17781 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17783 * tree.c (make_node_stat): rename to make_node.
17784 (build_tree_list_stat): Adjust.
17785 (build0_stat): Likewise.
17786 (build2_stat): Likewise.
17787 (build3_stat): Likewise.
17788 (build4_stat): Likewise.
17789 (build5_stat): Likewise.
17790 (build_decl_stat): Likewise.
17791 * tree.h (make_node_stat): Adjust prototype.
17792 (make_node): remove macro.
17794 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
17796 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
17797 (PPC_FEATURE2_SCV): Likewise.
17798 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
17800 2017-07-28 Tamar Christina <tamar.christina@arm.com>
17802 * config/aarch64/aarch64.c
17803 (aarch64_internal_mov_immediate): Add new special pattern.
17804 * config/aarch64/aarch64.md (*movdi_aarch64):
17805 Add reg/32bit const mov case.
17807 2017-07-28 Tamar Christina <tamar.christina@arm.com>
17808 Richard Sandiford <richard.sandiford@linaro.org>
17810 * config/aarch64/aarch64.md (mov<mode>): Generalize.
17811 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
17812 Add integer and movi cases.
17813 (movi-split-hf-df-sf split, fp16): New.
17814 (enabled): Added TARGET_FP_F16INST.
17815 * config/aarch64/iterators.md (GPF_HF): New.
17816 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
17818 2017-07-28 Tamar Christina <tamar.christina@arm.com>
17820 * config/aarch64/aarch64.c
17821 (aarch64_simd_container_mode): Add prototype.
17822 (aarch64_expand_mov_immediate): Add HI support.
17823 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
17824 (aarch64_can_const_movi_rtx_p): New.
17825 (aarch64_preferred_reload_class):
17826 Remove restrictions of using FP registers for certain SIMD operations.
17827 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
17828 (aarch64_valid_floating_const): Add integer move validation.
17829 (aarch64_simd_imm_scalar_p): Remove.
17830 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
17831 (aarch64_legitimate_constant_p): Expand list of supported cases.
17832 * config/aarch64/aarch64-protos.h
17833 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
17834 (aarch64_reinterpret_float_as_int): New.
17835 (aarch64_simd_imm_scalar_p): Remove.
17836 * config/aarch64/constraints.md (Uvi): New.
17837 (Dd): Split into Ds and new Dd.
17838 * config/aarch64/aarch64.md (*movsi_aarch64):
17840 (*movdi_aarch64): Add SIMD mov case.
17842 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17844 * tree-predcom.c: (struct chain): Handle store-store chain in which
17845 stores for elimination only store loop invariant values.
17846 (execute_pred_commoning_chain): Ditto.
17847 (prepare_initializers_chain_store_elim): Ditto.
17848 (prepare_finalizers): Ditto.
17849 (is_inv_store_elimination_chain): New function.
17850 (initialize_root_vars_store_elim_1): New function.
17852 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17854 * tree-predcom.c: Revise general description of the pass.
17855 (enum chain_type): New enum type for store elimination.
17856 (struct chain): New field supporting store elimination.
17857 (struct component): Ditto.
17858 (dump_chain): Dump store-stores chain.
17859 (release_chain): Release resources.
17860 (split_data_refs_to_components): Compute and create component
17861 contains only stores for elimination.
17862 (get_chain_last_ref_at): New function.
17863 (make_invariant_chain): Initialization.
17864 (make_rooted_chain): Specify chain type in parameter and record it.
17865 (add_looparound_copies): Skip for store-stores chain.
17866 (determine_roots_comp): Compute type of chain and pass it to
17868 (initialize_root_vars_store_elim_2): New function.
17869 (finalize_eliminated_stores): New function.
17870 (remove_stmt): Handle store for elimination.
17871 (execute_pred_commoning_chain): Execute predictive commoning on
17872 store-store chains.
17873 (determine_unroll_factor): Skip unroll for store-stores chain.
17874 (prepare_initializers_chain_store_elim): New function.
17875 (prepare_initializers_chain): Hanlde store-store chain.
17876 (prepare_finalizers_chain, prepare_finalizers): New function.
17877 (tree_predictive_commoning_loop): Return integer value indicating
17878 if loop is unrolled or lcssa form is corrupted.
17879 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
17881 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17883 * tree-predcom.c (initialize_root): Delete.
17884 (execute_pred_commoning_chain): Initialize root vars and replace
17885 reference of non-combined chain directly, rather than call above
17888 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17890 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
17891 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
17893 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17895 * tree-predcom.c (struct chain): New field init_seq.
17896 (release_chain): Release init_seq.
17897 (prepare_initializers_chain): Record intialization stmts in above
17899 (insert_init_seqs): New function.
17900 (tree_predictive_commoning_loop): Call insert_init_seqs.
17902 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17904 * tree-predcom.c (determine_roots_comp): Skip trivial components.
17906 2017-07-28 Richard Biener <rguenther@suse.de>
17908 * match.pd: Remove superfluous :c.
17909 * genmatch.c (simplify::id): Add member.
17910 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
17912 (current_id): New global.
17913 (dt_node::parent): Move from ...
17914 (dt_operand::parent): ... here. Add for_id member.
17915 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
17916 (decision_tree::find_node): Relax order requirement when
17917 merging DT_TRUE nodes to ones inbetween the current simplify
17918 and the one we try to merge with. Add diagnostic whenever
17919 we need to enforce pattern order by not merging.
17920 (decision_tree::insert): Set current_id.
17921 (decision_tree::print_node): Dump parent node and for_id.
17922 (parser::last_id): Add member.
17923 (parser::push_simplify): Assign unique id.
17924 (parser::parser): Initialize last_id.
17926 2017-07-28 Martin Liska <mliska@suse.cz>
17929 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
17930 gimple_build_debug_bind.
17932 2017-07-28 Richard Biener <rguenther@suse.de>
17934 PR tree-optimization/81502
17935 * match.pd: Add pattern combining BIT_INSERT_EXPR with
17937 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
17939 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
17940 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
17941 for BIT_FIELD_REF args.
17942 * fold-const.c (make_bit_field_ref): Likewise.
17943 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
17945 2017-07-28 Jakub Jelinek <jakub@redhat.com>
17948 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
17949 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
17950 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
17951 Or it into SANITIZER_UNDEFINED.
17952 * ubsan.c: Include gimple-fold.h and varasm.h.
17953 (ubsan_expand_ptr_ifn): New function.
17954 (instrument_pointer_overflow): New function.
17955 (maybe_instrument_pointer_overflow): New function.
17956 (instrument_object_size): Formatting fix.
17957 (pass_ubsan::execute): Call instrument_pointer_overflow
17958 and maybe_instrument_pointer_overflow.
17959 * internal-fn.c (expand_UBSAN_PTR): New function.
17960 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
17961 * sanitizer.def (__ubsan_handle_pointer_overflow,
17962 __ubsan_handle_pointer_overflow_abort): New builtins.
17963 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
17964 * internal-fn.def (UBSAN_PTR): New internal function.
17965 * opts.c (sanitizer_opts): Add pointer-overflow.
17966 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
17967 * fold-const.c (build_range_check): Compute pointer range check in
17968 integral type if pointer arithmetics would be needed. Formatting
17971 2017-07-28 Martin Liska <mliska@suse.cz>
17974 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
17975 parameters that are of a variable-length.
17977 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
17979 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
17981 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
17982 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
17983 (CRT_CALL_STATIC_FUNCTION): Likewise.
17984 (ASM_DEFAULT_SPEC): New define.
17985 (ASM_SPEC32): Likewise.
17986 (ASM_SPEC64): Likewise.
17987 (ASM_SPEC_COMMON): Likewise.
17988 (ASM_SPEC): Likewise.
17989 (INVALID_64BIT): Likewise.
17990 (LINK_OS_DEFAULT_SPEC): Likewise.
17991 (LINK_OS_SPEC32): Likewise.
17992 (LINK_OS_SPEC64): Likewise.
17993 (POWERPC_LINUX): Likewise.
17994 (PTRDIFF_TYPE): Likewise.
17995 (RESTORE_FP_PREFIX): Likewise.
17996 (RESTORE_FP_SUFFIX): Likewise.
17997 (SAVE_FP_PREFIX): Likewise.
17998 (SAVE_FP_SUFFIX): Likewise.
17999 (SIZE_TYPE): Likewise.
18000 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
18001 (TARGET_64BIT): Likewise.
18002 (TARGET_64BIT): Likewise.
18003 (TARGET_AIX): Likewise.
18004 (WCHAR_TYPE_SIZE): Likewise.
18005 (WCHAR_TYPE): Undefine.
18006 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
18007 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
18008 (CPP_OS_RTEMS_SPEC): Delete.
18009 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
18010 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
18012 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
18014 2017-07-28 Jakub Jelinek <jakub@redhat.com>
18016 PR tree-optimization/81578
18017 * tree-parloops.c (build_new_reduction): Bail out if
18018 reduction_code isn't one of the standard OpenMP reductions.
18019 Move the details printing after that decision.
18021 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18023 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
18024 related to reload_in_progress.
18025 (splat_input_operand): Likewise.
18026 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
18028 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
18030 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
18031 (TARGET_INSTANTIATE_DECLS): Likewise.
18032 (legitimate_indexed_address_p): Delete reload_in_progress code.
18033 (rs6000_debug_legitimate_address_p): Likewise.
18034 (rs6000_eliminate_indexed_memrefs): Likewise.
18035 (rs6000_emit_le_vsx_store): Likewise.
18036 (rs6000_emit_move_si_sf_subreg): Likewise.
18037 (rs6000_emit_move): Likewise.
18038 (register_to_reg_type): Likewise.
18039 (rs6000_pre_atomic_barrier): Likewise.
18040 (rs6000_machopic_legitimize_pic_address): Likewise.
18041 (rs6000_allocate_stack_temp): Likewise.
18042 (rs6000_address_for_fpconvert): Likewise.
18043 (rs6000_address_for_altivec): Likewise.
18044 (rs6000_secondary_memory_needed_rtx): Delete function.
18045 (rs6000_check_sdmode): Likewise.
18046 (rs6000_alloc_sdmode_stack_slot): Likewise.
18047 (rs6000_instantiate_decls): Likewise.
18048 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
18049 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
18050 Delete reload_in_progress.
18051 (*vec_reload_and_plus_<mptrsize>): Likewise.
18052 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
18053 (vsx_div_v2di): Likewise.
18054 (vsx_udiv_v2di): Likewise.
18056 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18058 * config/rs6000/rs6000.opt (mlra): Replace with stub.
18059 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
18060 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
18061 (rs6000_debug_reg_global): Delete print of LRA status.
18062 (rs6000_option_override_internal): Delete dead LRA related code.
18063 (rs6000_lra_p): Delete function.
18064 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
18066 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
18068 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
18069 * config/riscv/rtems.h: New file.
18071 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18072 Sudakshina Das <sudi.das@arm.com>
18074 * config/aarch64/aarch64.md
18075 (define_split for and<mode>3nr_compare): Move
18076 non aarch64_logical_operand to a register.
18077 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
18078 register immediate operand to a register.
18079 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
18081 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18083 PR middle-end/81564
18084 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
18086 2017-07-27 Richard Biener <rguenther@suse.de>
18088 PR tree-optimization/81573
18089 PR tree-optimization/81494
18090 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
18091 multi defuse cycle case.
18093 2017-07-27 Richard Biener <rguenther@suse.de>
18095 PR tree-optimization/81571
18096 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
18099 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
18101 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
18102 earlier and only if MASK_FPU is set. Adjust formatting.
18104 2017-07-27 Martin Liska <mliska@suse.cz>
18106 * opt-functions.awk: Add validation of value of Init.
18107 * optc-gen.awk: Pass new argument.
18109 2017-07-27 Martin Liska <mliska@suse.cz>
18111 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
18112 Fix wrong condition.
18114 2017-07-27 Martin Liska <mliska@suse.cz>
18116 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
18117 BBs and edges seen by autoFDO.
18119 2017-07-27 Richard Biener <rguenther@suse.de>
18121 PR tree-optimization/81502
18122 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
18123 with incompatible but same sized type.
18124 (execute_update_addresses_taken): Likewise.
18126 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
18128 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
18129 flag_tree_loop_vectorize rather than flag_tree_vectorize.
18131 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18134 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
18135 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
18136 Change s_operand to memory_operand.
18138 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
18140 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
18141 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
18142 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
18143 Emit instructions rather than returning an expression. Handle TFmode
18144 and KFmode by casting to TImode.
18145 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
18146 (rs6000_emit_le_vsx_store): Likewise.
18147 * config/rs6000/vsx.md (VSX_TI): New iterator.
18148 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
18149 (*vsx_le_undo_permute_<mode>): Likewise.
18150 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
18151 emit the split sequence.
18152 (*vsx_le_perm_store_<mode>): Likewise.
18154 2017-07-27 Jakub Jelinek <jakub@redhat.com>
18156 PR tree-optimization/81555
18157 PR tree-optimization/81556
18158 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
18159 if true, force CHANGED for the recursive invocation.
18160 (reassociate_bb): Remember original length of ops array, pass
18161 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
18163 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
18164 attributes for noipa attribute. For naked attribute use
18165 lookup_attribute first before lookup_attribute_spec.
18166 * final.c (rest_of_handle_final): Disable IPA RA for functions with
18168 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
18169 for functions with noipa attribute.
18170 (cgraph_externally_visible_p): Return true for functions with noipa
18172 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
18173 for functions with noipa attribute.
18174 * doc/extend.texi: Document noipa function attribute.
18175 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
18176 also for functions with noipa attribute.
18177 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
18179 2017-07-26 Andrew Pinski <apinski@cavium.com>
18181 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
18182 vec_unalign_load_cost and vec_unalign_store_cost.
18184 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
18186 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
18187 -mvsx-small-integer option.
18188 (ISA_3_0_MASKS_IEEE): Likewise.
18189 (OTHER_VSX_VECTOR_MASKS): Likewise.
18190 (POWERPC_MASKS): Likewise.
18191 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
18192 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
18193 code, only testing for DImode being allowed in non-VSX floating
18195 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
18196 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
18198 (rs6000_option_override_internal): Delete -mvsx-small-integer.
18199 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
18200 TARGET_P8_VECTOR test.
18201 (rs6000_secondary_reload_simple_move): Likewise.
18202 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
18203 since TARGET_P9_VECTOR was already tested.
18204 (rs6000_opt_masks): Remove -mvsx-small-integer.
18205 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
18206 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18208 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
18209 test for TARGET_VEXTRACTUB was used, and that uses
18211 (p9 extract splitter): Likewise.
18212 (vsx_extract_<mode>_di_p9): Likewise.
18213 (vsx_extract_<mode>_store_p9): Likewise.
18214 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
18215 for TARGET_P9_VECTOR was used. Delete code that is now dead with
18216 the elimination of TARGET_VSX_SMALL_INTEGER.
18217 (vsx_extract_<mode>_p8): Likewise.
18218 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
18219 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
18220 (vsx_set_<mode>_p9): Likewise.
18221 (vsx_set_v4sf_p9): Likewise.
18222 (vsx_set_v4sf_p9_zero): Likewise.
18223 (vsx_insert_extract_v4sf_p9): Likewise.
18224 (vsx_insert_extract_v4sf_p9_2): Likewise.
18225 * config/rs6000/rs6000.md (sign extend splitter): Change
18226 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
18227 (floatsi<mode>2_lfiwax_mem): Likewise.
18228 (floatunssi<mode>2_lfiwzx_mem): Likewise.
18229 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
18230 since a test for TARGET_P9_VECTOR was used.
18231 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18232 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
18233 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18234 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
18235 TARGET_P8_VECTOR test.
18236 (fix_trunc<mode>si2_stfiwx): Likewise.
18237 (fix_trunc<mode>si2_internal): Likewise.
18238 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
18239 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18241 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
18242 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
18243 TARGET_P8_VECTOR test.
18244 (fixuns_trunc<mode>si2_stfiwx): Likewise.
18245 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
18246 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18248 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
18249 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
18250 since a test for TARGET_P9_VECTOR was used.
18251 (splitter for loading small constants): Likewise.
18253 2017-07-26 Andrew Pinski <apinski@cavium.com>
18255 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
18258 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
18261 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
18262 (fp_valid_at): Likewise.
18264 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
18266 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
18267 (qdf24xx_addrcost_table): Likewise.
18268 (cortexa57_tunings): Update to use generic_branch_cost.
18269 (cortexa72_tunings): Likewise.
18270 (cortexa73_tunings): Likewise.
18271 (qdf24xx_tunings): Likewise.
18273 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
18275 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
18276 (thunderx2t99_branch_cost): Likewise.
18277 (cortexa35_tunings): Update to use generic_branch_cost.
18278 (cortexa53_tunings): Likewise.
18279 (cortexa57_tunings): Likewise.
18280 (cortexa72_tunings): Likewise.
18281 (cortexa73_tunings): Likewise.
18282 (thunderx2t99_tunings): Likewise.
18284 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
18286 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
18287 (sparc_option_override): Honour MASK_FSMULD.
18288 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
18289 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
18290 * config/sparc/sparc.opt (mfsmuld): New option.
18291 * doc/invoke.texi (mfsmuld): Document option.
18293 2017-07-26 Marek Polacek <polacek@redhat.com>
18295 PR middle-end/70992
18296 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
18298 2017-07-26 Richard Biener <rguenther@suse.de>
18300 * gimple-match-head.c (do_valueize): Return OP if valueize
18302 (get_def): New helper to get at the def stmt of a SSA name
18303 if valueize allows.
18304 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
18305 do_valueize to get at the def stmt.
18306 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
18308 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
18310 PR middle-end/46932
18311 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
18313 2017-07-26 Martin Liska <mliska@suse.cz>
18316 * function.c (expand_function_start): Make expansion of
18317 nonlocal_goto_save_area after parm_birth_insn.
18319 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
18321 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
18322 from all CPU target flags enable members.
18324 2017-07-26 Richard Biener <rguenther@suse.de>
18326 * genmatch.c (dt_simplify::gen): Make iterator vars const.
18327 (decision_tree::gen): Make 'type' const.
18328 (write_predicate): Likewise.
18330 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
18332 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
18333 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
18334 (rs6000_option_override_internal): Likewise.
18335 (rs6000_expand_vector_set): Likewise.
18336 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
18337 (TARGET_UPPER_REGS_SF): Likewise.
18338 (TARGET_UPPER_REGS_DI): Likewise.
18339 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
18340 (TARGET_DIRECT_MOVE_64BIT): Likewise.
18341 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
18342 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18343 (Splitters for DI constants in Altivec registers): Likewise.
18344 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
18345 (vsx_set_v4sf_p9): Likewise.
18346 (vsx_set_v4sf_p9_zero): Likewise.
18347 (vsx_insert_extract_v4sf_p9): Likewise.
18348 (vsx_insert_extract_v4sf_p9_2): Likewise.
18350 2017-07-25 Carl Love <cel@us.ibm.com>
18352 * doc/extend.texi: Update the built-in documentation file for the
18353 existing built-in functions
18354 vector signed char vec_cnttz (vector signed char);
18355 vector unsigned char vec_cnttz (vector unsigned char);
18356 vector signed short vec_cnttz (vector signed short);
18357 vector unsigned short vec_cnttz (vector unsigned short);
18358 vector signed int vec_cnttz (vector signed int);
18359 vector unsigned int vec_cnttz (vector unsigned int);
18360 vector signed long long vec_cnttz (vector signed long long);
18361 vector unsigned long long vec_cnttz (vector unsigned long long);
18363 2017-07-25 Andrew Pinski <apinski@cavium.com>
18365 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
18366 accesses where the use is for the first operand of a BIT_INSERT.
18368 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
18371 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
18372 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
18374 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
18376 * config/i386/gstabs.h: Delete.
18377 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
18379 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
18381 * config/i386/i386.c (ix86_decompose_address): Do not check for
18382 register RTX when looking at index_reg or base_reg.
18383 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
18385 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
18387 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
18388 to update EH info here.
18390 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18392 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
18394 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18396 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
18398 2017-07-25 Torsten Duwe <duwe@suse.de>
18400 * common.opt: Introduce -fpatchable-function-entry
18401 command line option, and its variables function_entry_patch_area_size
18402 and function_entry_patch_area_start.
18403 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
18404 including a two-value parser.
18405 * target.def (print_patchable_function_entry): New target hook.
18406 * targhooks.h (default_print_patchable_function_entry): New function.
18407 * targhooks.c (default_print_patchable_function_entry): Likewise.
18408 * toplev.c (process_options): Switch off IPA-RA if
18409 patchable function entries are being generated.
18410 * varasm.c (assemble_start_function): Look at the
18411 patchable-function-entry command line switch and current
18412 function attributes and maybe generate NOP instructions by
18413 calling the print_patchable_function_entry hook.
18414 * doc/extend.texi: Document patchable_function_entry attribute.
18415 * doc/invoke.texi: Document -fpatchable_function_entry
18416 command line option.
18417 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
18419 * doc/tm.texi: Re-generate.
18421 2017-07-25 Jakub Jelinek <jakub@redhat.com>
18424 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
18425 TARGET_AVX512DQ rather than TARGET_AVX512BW.
18427 2017-07-25 Tamar Christina <tamar.christina@arm.com>
18429 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
18431 2017-07-25 Richard Biener <rguenther@suse.de>
18433 PR tree-optimization/81455
18434 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
18435 not walk in cycles when looking for guards.
18437 2017-07-25 Richard Biener <rguenther@suse.de>
18439 PR tree-optimization/81529
18440 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
18441 when optimizing backedge uses.
18443 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
18445 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
18447 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
18448 to dl_section_ref. On AIX, append an expression to subtract
18449 the size of the section length to dl_section_ref.
18451 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
18453 * configure.ac: If any of the config.* scripts fail, exit 1.
18454 * configure: Regenerate.
18456 2017-07-25 Richard Biener <rguenther@suse.de>
18458 PR middle-end/81546
18459 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
18460 of immediate uses, be more verbose on errors.
18462 2017-07-25 Richard Biener <rguenther@suse.de>
18464 PR tree-optimization/81510
18465 * tree-vect-loop.c (vect_is_simple_reduction): When the
18466 reduction stmt is not inside the loop bail out.
18468 2017-07-25 Richard Biener <rguenther@suse.de>
18470 PR tree-optimization/81303
18471 * tree-vect-loop-manip.c (vect_loop_versioning): Build
18472 profitability check against LOOP_VINFO_NITERSM1.
18474 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18476 * domwalk.c (cmp_bb_postorder): Simplify.
18477 (sort_bbs_postorder): New function. Use it...
18478 (dom_walker::walk): ...here to optimize common cases.
18480 2017-07-25 Martin Liska <mliska@suse.cz>
18483 * ipa-visibility.c (function_and_variable_visibility): Make the
18484 redirection just on target that supports aliasing.
18485 Fix GNU coding style.
18487 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
18490 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
18492 * config/arm/rtems.h: Likewise.
18493 * config/bfin/rtems.h: Likewise.
18494 * config/i386/rtemself.h: Likewise.
18495 * config/lm32/rtems.h: Likewise.
18496 * config/m32c/rtems.h: Likewise.
18497 * config/m68k/rtemself.h: Likewise.
18498 * config/microblaze/rtems.h: Likewise.
18499 * config/mips/rtems.h: Likewise.
18500 * config/moxie/rtems.h: Likewise.
18501 * config/nios2/rtems.h: Likewise.
18502 * config/powerpcspe/rtems.h: Likewise.
18503 * config/rs6000/rtems.h: Likewise.
18504 * config/rtems.h: Likewise.
18505 * config/sh/rtems.h: Likewise.
18506 * config/sh/rtemself.h: Likewise.
18507 * config/sparc/rtemself.h: Likewise.
18509 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
18512 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
18513 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
18514 * tree-ssa-structalias.c (alias_get_name): Same.
18516 2017-07-25 Bin Cheng <bin.cheng@arm.com>
18519 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
18520 instructions if no du chain is found.
18522 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
18524 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
18526 2017-07-25 Richard Biener <rguenther@suse.de>
18528 PR middle-end/81505
18529 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
18532 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
18534 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
18535 upper-regs options.
18536 (ISA_2_7_MASKS_SERVER): Likewise.
18537 (ISA_3_0_MASKS_IEEE): Likewise.
18538 (OTHER_P8_VECTOR_MASKS): Likewise.
18539 (OTHER_VSX_VECTOR_MASKS): Likewise.
18540 (POWERPC_MASKS): Likewise.
18541 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
18542 duplicate list of options.
18543 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
18544 explicit -mupper-regs options.
18545 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
18546 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
18547 alias for -mupper-regs-df.
18548 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
18549 (rs6000_init_hard_regno_mode_ok): Likewise.
18550 (rs6000_option_override_internal): Likewise.
18551 (rs6000_opt_masks): Likewise.
18552 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
18553 options in terms of whether -mvsx or -mpower8-vector was used.
18554 (TARGET_UPPER_REGS_DI): Likewise.
18555 (TARGET_UPPER_REGS_SF): Likewise.
18556 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
18557 -mupper-regs-* options.
18559 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
18561 * passes.c (emergency_dump_function): Print some empty lines and a
18562 header before the RTL dump.
18564 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
18566 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
18568 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
18571 * config/aarch64/aarch64.c (aarch64_classify_symbol):
18572 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
18574 2017-07-24 Carl Love <cel@us.ibm.com>
18576 * config/rs6000/rs6000-c.c: Add support for built-in functions
18577 vector float vec_extract_fp32_from_shorth (vector unsigned short);
18578 vector float vec_extract_fp32_from_shortl (vector unsigned short);
18579 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
18580 vec_extract_fp_from_shortl): Add defines for the two builtins.
18581 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
18582 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
18584 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
18585 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
18586 * doc/extend.texi: Update the built-in documentation file for the
18587 new built-in function.
18589 2017-07-24 Jakub Jelinek <jakub@redhat.com>
18592 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
18594 * doc/generic.texi: Likewise.
18595 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
18596 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
18598 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
18600 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
18601 (aarch64_mls_elt_merge<mode>): Likewise.
18603 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
18605 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
18606 having __cxa_atexit.
18608 2017-07-23 Michael Collison <michael.collison@arm.com>
18610 * config/arm/arm.c (arm_option_override): Deprecate
18611 use of -mstructure-size-boundary.
18612 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
18613 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
18615 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18618 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
18619 Reduce cost estimate for direct moves.
18621 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
18624 * config/i386/i386.c (ix86_option_override_internal): Disable
18625 BMI, BMI2 and TBM instructions for -m16.
18627 2017-07-21 Carl Love <cel@us.ibm.com>
18629 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18630 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
18631 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
18632 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
18633 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
18634 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
18635 VMULOSW): New enum "unspec" values.
18636 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
18637 altivec_vmulosw): New patterns.
18638 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
18639 VMULOSW): Add definitions.
18641 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
18643 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
18644 (qdf24xx): Likewise.
18645 * config/aarch64/aarch64-options-extensions.def (rdma); New.
18646 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
18647 (AARCH64_FL_V8_1): Renumber.
18648 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
18649 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
18650 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
18651 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
18652 rdma to feature modifiers list.
18654 2017-07-21 Yury Gribov <tetra2005@gmail.com>
18656 PR middle-end/56727
18657 * ipa-visibility (function_and_variable_visibility): Convert
18658 recursive PLT call to direct call if appropriate.
18660 2017-07-21 Andrew Pinski <apinski@cavium.com>
18662 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
18663 operand 1 to see if the types precision matches.
18664 * fold-const.c (operand_equal_p): Likewise.
18666 2017-07-21 Richard Biener <rguenther@suse.de>
18668 PR tree-optimization/81303
18669 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
18670 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
18671 (vect_peeling_hash_get_lowest_cost): Adjust.
18672 (vect_enhance_data_refs_alignment): Likewise. Use
18673 vect_get_peeling_costs_all_drs to compute the penalty for no
18674 peeling to match up costs.
18676 2017-07-21 Richard Biener <rguenther@suse.de>
18678 PR tree-optimization/81500
18679 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
18680 we didn't identify a reduction path.
18682 2017-07-21 Tom de Vries <tom@codesourcery.com>
18683 Cesar Philippidis <cesar@codesourcery.com>
18685 PR gcov-profile/81442
18686 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
18689 2017-07-21 Tom de Vries <tom@codesourcery.com>
18692 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
18694 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
18695 nvptx_override_options_after_change.
18697 2017-07-21 Ulrich Drepper <drepper@redhat.com>
18699 * dwarf2out.c (output_file_names): Avoid double testing for
18700 dwarf_version >= 5.
18702 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
18704 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
18706 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
18708 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
18710 (try_crossjump_to_edge): Do not punt on partitioned functions.
18712 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
18714 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
18715 Put all BBs reachable only via paths crossing cold region to cold
18717 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
18719 2016-07-21 Richard Biener <rguenther@suse.de>
18721 PR tree-optimization/81303
18722 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
18723 into account prologue and epilogue iterations when raising
18724 min_profitable_iters to sth at least covering one vector iteration.
18726 2017-07-21 Tamar Christina <tamar.christina@arm.com>
18728 * config/arm/arm.c (arm_test_cpu_arch_dat):
18731 2017-07-20 Nathan Sidwell <nathan@acm.org>
18733 Remove TYPE_METHODS.
18734 * tree.h (TYPE_METHODS): Delete.
18735 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
18736 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
18737 (dbxout_type_methods): Scan TYPE_FIELDS.
18738 (dbxout_type): Don't check TYPE_METHODS here.
18739 * function.c (use_register_for_decl): Always ignore register for
18740 class types when not optimizing.
18741 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
18742 * tree.c (free_lang_data_in_type): Stitch out member functions and
18743 templates from TYPE_FIELDS.
18744 (build_distinct_type_copy, verify_type_variant,
18745 verify_type): Member fns are on TYPE_FIELDS.
18746 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
18747 * tree-pretty-print.c (dump_generic_node): Likewise.
18749 2017-07-20 Jakub Jelinek <jakub@redhat.com>
18752 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
18753 V2TImode and V4TImode.
18754 (ix86_expand_vector_extract): Likewise.
18755 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
18756 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
18757 (ssescalarmode): Handle V4TImode and V2TImode.
18758 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
18759 (*vec_extractv2ti, *vec_extractv4ti): New insns.
18760 (VEXTRACTI128_MODE): New mode iterator.
18761 (splitter for *vec_extractv?ti first element): New.
18762 (VEC_INIT_MODE): New mode iterator.
18763 (vec_init<mode>): Consolidate 3 expanders into one using
18764 VEC_INIT_MODE mode iterator.
18766 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
18768 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
18769 non_spilled_static_chain_regno_p.
18771 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
18773 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
18775 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
18777 * bb-reorder.c (connect_traces): Allow copying of blocks within
18780 2017-07-20 Richard Biener <rguenther@suse.de>
18782 * gimple.h (gimple_phi_result): Add gphi * overload.
18783 (gimple_phi_result_ptr): Likewise.
18784 (gimple_phi_arg): Likewise. Adjust index assert to only
18785 allow actual argument accesses rather than all slots available
18787 (gimple_phi_arg_def): Add gphi * overload.
18788 * tree-phinodes.c (make_phi_node): Initialize only actual
18790 (resize_phi_node): Clear memory not covered by old node,
18791 do not initialize excess argument slots.
18792 (reserve_phi_args_for_new_edge): Initialize new argument slot
18795 2017-07-20 Bin Cheng <bin.cheng@arm.com>
18797 PR tree-optimization/81388
18799 2016-07-21 Bin Cheng <bin.cheng@arm.com>
18801 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
18802 by removing computation of may_be_zero.
18804 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
18805 Tom de Vries <tom@codesourcery.com>
18807 PR middle-end/81030
18808 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
18809 when gimple level profile disagrees with what RTL expander did.
18811 2017-07-20 Richard Biener <rguenther@suse.de>
18813 PR tree-optimization/61171
18814 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
18815 (vect_analyze_stmt): Add slp instance parameter.
18816 (vectorizable_reduction): Likewise.
18817 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
18818 (vect_is_simple_reduction): Deal with chains not detected
18819 as SLP reduction chain, specifically not properly associated
18820 chains containing a mix of plus/minus.
18821 (get_reduction_op): Remove.
18822 (get_initial_defs_for_reduction): Simplify, pass in whether
18823 this is a reduction chain, pass in the SLP node for the PHIs.
18824 (vect_create_epilog_for_reduction): Get the SLP instance as
18826 (vectorizable_reduction): Get the SLP instance as arg.
18827 During analysis remember the SLP node with the PHIs in the
18828 instance. Simplify getting at the vectorized reduction PHIs.
18829 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
18830 through SLP instance.
18831 (vect_slp_analyze_operations): Likewise.
18832 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
18833 (vect_transform_stmt): Likewise.
18835 2017-07-20 Tom de Vries <tom@codesourcery.com>
18837 PR tree-optimization/81489
18838 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
18839 read of phi arg location to before loop that modifies phi.
18841 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18843 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
18846 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
18848 PR middle-end/81331
18849 * except.c (execute): Fix ordering issue.
18851 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
18853 PR rtl-optimization/81423
18854 * combine.c (make_compound_operation_int): Don't try to optimize
18855 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
18857 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
18859 PR rtl-optimization/81423
18860 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
18861 with a constant that is -1 in the truncated to mode.
18863 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
18865 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
18866 (determine_unlikely_bbs): ... here.
18867 * predict.h (propagate_unlikely_bbs_forward): Declare.
18868 * cfgexpand.c (pass_expand::execute): Use it.
18869 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
18871 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
18872 propagate_unlikely_bbs_forward.
18874 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
18876 PR middle-end/81331
18877 * except.c (maybe_add_nop_after_section_switch): New function.
18880 2017-07-19 Tom de Vries <tom@codesourcery.com>
18882 * gimple.h (gimple_phi_set_arg): Make assert more strict.
18884 2017-07-19 Tom de Vries <tom@codesourcery.com>
18886 * gimple.h (gimple_phi_arg): Make assert more strict.
18888 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
18890 * config.gcc (powerpc*-*-*): Add mmintrin.h.
18891 * config/rs6000/mmintrin.h: New file.
18892 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
18894 2017-07-19 Jakub Jelinek <jakub@redhat.com>
18896 PR tree-optimization/81346
18897 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
18899 2017-07-19 Tom de Vries <tom@codesourcery.com>
18901 * config/nvptx/nvptx.md (VECIM): Add V2DI.
18903 2017-07-19 Tom de Vries <tom@codesourcery.com>
18905 * config/nvptx/nvptx-modes.def: Add V2DImode.
18906 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
18907 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
18908 (nvptx_output_mov_insn): Handle lack of mov.b128.
18909 (nvptx_print_operand): Handle 'H' and 'L' codes.
18910 (nvptx_vector_mode_supported): Allow V2DImode.
18911 (nvptx_preferred_simd_mode): New function.
18912 (nvptx_data_alignment): New function.
18913 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
18914 nvptx_preferred_simd_mode.
18915 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
18917 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
18919 2017-07-19 Tom de Vries <tom@codesourcery.com>
18921 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
18922 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
18923 (nvptx_vector_mode_supported): New function. Allow V2SImode.
18924 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
18925 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
18926 (mov<VECIM>_insn): New define_insn.
18927 (define_expand "mov<VECIM>): New define_expand.
18929 2017-07-19 Tom de Vries <tom@codesourcery.com>
18931 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
18933 2017-07-19 Jakub Jelinek <jakub@redhat.com>
18935 PR tree-optimization/81346
18936 * fold-const.h (fold_div_compare, range_check_type): Declare.
18937 * fold-const.c (range_check_type): New function.
18938 (build_range_check): Use range_check_type.
18939 (fold_div_compare): No longer static, rewritten into
18940 a match.pd helper function.
18941 (fold_comparison): Don't call fold_div_compare here.
18942 * match.pd (X / C1 op C2): New optimization using fold_div_compare
18943 as helper function.
18945 2017-07-19 Nathan Sidwell <nathan@acm.org>
18947 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
18948 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
18949 * tree.c (find_decls_types_r, verify_type): Use
18950 TYPE_{MIN,MAX}_VALUE_RAW.
18951 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
18952 (hash_tree): Likewise.
18953 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
18955 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
18958 2017-07-18 Tom de Vries <tom@codesourcery.com>
18960 PR middle-end/81464
18961 * omp-expand.c (expand_omp_for_static_chunk): Handle
18962 equal-argument loop exit phi.
18964 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
18967 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
18968 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
18969 operand 2 predicate.
18970 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
18971 operand 2 predicate.
18972 (ror,rol -> rorx splitters): Use const_int_operand as
18973 operand 2 predicate.
18975 2017-06-18 Richard Biener <rguenther@suse.de>
18977 PR tree-optimization/81410
18978 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
18979 the gap in the ! slp_perm SLP case after each group.
18981 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
18983 PR middle-end/81463
18984 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
18987 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
18989 PR middle-end/81462
18990 * predict.c (set_even_probabilities): Cleanup; do not affect
18991 probabilities that are already known.
18992 (combine_predictions_for_bb): Call even when count is set.
18994 2017-07-18 Nathan Sidwell <nathan@acm.org>
18996 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
18999 2017-07-18 Bin Cheng <bin.cheng@arm.com>
19002 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
19003 optimization for loop niter analysis.
19005 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
19008 * config/avr/avr.c (avr_optimize_casesi): Don't use
19009 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
19011 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
19013 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
19014 body_cost_vec from _vect_peel_extended_info.
19015 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
19016 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
19019 2017-07-18 Bin Cheng <bin.cheng@arm.com>
19021 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
19023 2017-07-18 Richard Biener <rguenther@suse.de>
19025 PR tree-optimization/80620
19026 PR tree-optimization/81403
19027 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
19028 info when re-using a VN table entry.
19030 2017-07-18 Richard Biener <rguenther@suse.de>
19032 PR tree-optimization/81418
19033 * tree-vect-loop.c (vectorizable_reduction): Properly compute
19034 vectype_in. Verify that with lane-reducing reduction operations
19035 we have a single def-use cycle.
19037 2017-07-17 Carl Love <cel@us.ibm.com>
19039 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
19041 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19042 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19043 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19044 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19045 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19046 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19047 VMULOSW): New enum "unspec" values.
19048 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19049 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19050 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19051 altivec_vmulosw): New patterns.
19052 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19053 VMULOSW): Add definitions.
19054 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
19056 * config/alpha/alpha.c: Include predict.h.
19058 2017-07-17 Yury Gribov <tetra2005@gmail.com>
19060 * tree-vrp.c (compare_assert_loc): Fix comparison function
19061 to return predictable results.
19063 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19065 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
19067 (subdi3): Likewise.
19068 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
19069 * doc/invoke.texi (mexpand-adddi): Update text.
19071 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19073 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
19074 that also clobbers the CC register. The old expand code is moved
19076 (*arc_clzsi2): ... here.
19077 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
19078 the CC register. The old expand code is moved to ...
19079 (arc_ctzsi2): ... here.
19081 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19083 * config/arc/arc.opt (mindexed-loads): Use initial value
19084 TARGET_INDEXED_LOADS_DEFAULT.
19085 (mauto-modify-reg): Use initial value
19086 TARGET_AUTO_MODIFY_REG_DEFAULT.
19087 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
19088 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
19089 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
19090 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
19092 2017-07-17 Martin Liska <mliska@suse.cz>
19095 * opts.c (finish_options): Do not allow -fgnu-tm
19096 w/ -fsanitize={kernel-,}address. Say sorry.
19098 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19101 * tree-loop-distribution.c (classify_partition): Only assert on
19102 numer of iterations.
19103 (merge_dep_scc_partitions): Delete prameter. Update function call.
19104 (distribute_loop): Remove code handling loop with unknown niters.
19105 (pass_loop_distribution::execute): Skip loop with unknown niters.
19107 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19110 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
19111 function sort_partitions_by_post_order.
19113 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19115 PR tree-optimization/81374
19116 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
19117 the max index of basic blocks, rather than number of basic blocks.
19119 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19121 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
19123 (arc_legitimate_pic_operand_p): Likewise.
19124 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
19126 (arc_needs_pcl_p): Likewise.
19127 (arc_legitimate_pc_offset_p): Likewise.
19128 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
19129 function is also used in constrains.md.
19130 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
19131 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
19132 PLUS. Only return true/false in known cases, otherwise assert.
19133 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
19134 is already called in arc_legitimate_constant_p.
19135 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
19137 (LEGITIMATE_PIC_OPERAND_P): Use
19138 arc_raw_symbolic_reference_mentioned_p function.
19139 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
19144 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19145 Andrew Burgess <andrew.burgess@embecosm.com>
19147 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
19148 (arc_return_address_register): New function.
19149 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
19150 (arc_handle_fndecl_attribute): Add naked attribute.
19151 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
19152 (TARGET_WARN_FUNC_RETURN): Likewise.
19153 (arc_allocate_stack_slots_for_args): New function.
19154 (arc_warn_func_return): Likewise.
19155 (machine_function): Change type fn_type.
19156 (arc_compute_function_type): Consider new naked function type,
19157 change function return type.
19158 (arc_must_save_register): Adapt to handle new
19159 arc_compute_function_type's return type.
19160 (arc_expand_prologue): Likewise.
19161 (arc_expand_epilogue): Likewise.
19162 (arc_return_address_regs): Delete.
19163 (arc_return_address_register): New function.
19164 (arc_epilogue_uses): Use above function.
19165 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
19166 (arc_function_type): Change encoding, add naked type.
19167 (ARC_INTERRUPT_P): Change to handle the new encoding.
19168 (ARC_FAST_INTERRUPT_P): Likewise.
19169 (ARC_NORMAL_P): Define.
19170 (ARC_NAKED_P): Likewise.
19171 (arc_compute_function_type): Delete prototype.
19172 * config/arc/arc.md (in_ret_delay_slot): Use
19173 arc_return_address_register function.
19174 (simple_return): Likewise.
19175 (p_return_i): Likewise.
19177 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19179 PR tree-optimization/81428
19180 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
19181 can't be built for those types.
19183 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
19185 Remove stuff dead since r239246.
19187 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
19188 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
19189 (avr_inform_devices): Remove dead stuff.
19191 2017-07-17 Tamar Christina <tamar.christina@arm.com>
19193 * config/arm/arm_neon.h: Fix softp typo.
19195 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19197 PR tree-optimization/81365
19198 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
19199 aggregate moves onto bb predecessor edges, make sure there are no
19200 loads that could alias the lhs in between the start of bb and the
19203 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
19206 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
19207 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
19208 [LSHIFTRT, outer_code = TRUNCATE]: Same.
19210 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19212 PR tree-optimization/81396
19213 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
19214 (init_symbolic_number): Initialize it to 1.
19215 (perform_symbolic_merge): Add n_ops from both operands into the new
19217 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
19218 without base_addr as useless if they need more than one operation.
19219 (bswap_replace): Handle !bswap case for NULL base_addr.
19221 2017-07-17 Tom de Vries <tom@codesourcery.com>
19224 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
19227 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
19229 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
19230 conditional builtin define __FIX_LEON3FT_B2BST.
19232 2017-07-17 Daniel Cederman <cederman@gaisler.com>
19234 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
19235 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
19238 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
19240 PR rtl-optimization/81424
19241 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
19242 to remove potential trapping from operands if -fnon-call-exceptions.
19244 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19246 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
19247 profile_proability for scalling.
19248 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
19250 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19252 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
19254 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19256 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
19257 fixpoint arithmetics.
19259 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19261 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
19262 fixpoint arithmetics.
19264 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19266 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
19267 fixpoint arithmetics.
19269 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19271 * profile-count.h (profile_probability::from_reg_br_prob_note,
19272 profile_probability::to_reg_br_prob_note): New functions.
19273 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
19274 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
19275 * predict.c (probability_reliable_p): Update.
19276 (edge_probability_reliable_p): Update.
19277 (br_prob_note_reliable_p): Update.
19278 (invert_br_probabilities): Update.
19279 (add_reg_br_prob_note): New function.
19280 (combine_predictions_for_insn): Update.
19281 * asan.c (asan_clear_shadow): Update.
19282 * cfgbuild.c (compute_outgoing_frequencies): Update.
19283 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
19284 (update_br_prob_note): Update.
19285 (rtl_verify_edges): Update.
19286 (purge_dead_edges): Update.
19287 (fixup_reorder_chain): Update.
19288 * emit-rtl.c (try_split): Update.
19289 * ifcvt.c (cond_exec_process_insns): Update.
19290 (cond_exec_process_if_block): Update.
19291 (dead_or_predicable): Update.
19292 * internal-fn.c (expand_addsub_overflow): Update.
19293 (expand_neg_overflow): Update.
19294 (expand_mul_overflow): Update.
19295 * loop-doloop.c (doloop_modify): Update.
19296 * loop-unroll.c (compare_and_jump_seq): Update.
19297 * optabs.c (emit_cmp_and_jump_insn_1): Update.
19298 * predict.h: Update.
19299 * reorg.c (mostly_true_jump): Update.
19301 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
19302 * config/alpha/alpha.c (emit_unlikely_jump): Update.
19303 * config/arc/arc.c: (emit_unlikely_jump): Update.
19304 * config/arm/arm.c: (emit_unlikely_jump): Update.
19305 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
19306 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
19307 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
19308 (ix86_print_operand): Update.
19309 (ix86_split_fp_branch): Update.
19310 (predict_jump): Update.
19311 * config/ia64/ia64.c (ia64_print_operand): Update.
19312 * config/mmix/mmix.c (mmix_print_operand): Update.
19313 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
19314 (rs6000_expand_split_stack_prologue): Update.
19315 * config/rs6000/rs6000.c: Update.
19316 * config/s390/s390.c (s390_expand_vec_strlen): Update.
19317 (s390_expand_vec_movstr): Update.
19318 (s390_expand_cs_tdsi): Update.
19319 (s390_expand_split_stack_prologue): Update.
19320 * config/sh/sh.c (sh_print_operand): Update.
19321 (expand_cbranchsi4): Update.
19322 (expand_cbranchdi4): Update.
19323 * config/sparc/sparc.c (output_v9branch): Update.
19324 * config/spu/spu.c (get_branch_target): Update.
19325 (ea_load_store_inline): Update.
19326 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
19327 * config/tilepro/tilepro.c: Update.
19329 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
19331 * gimplify.c (mostly_copy_tree_r): Revert latest change.
19332 (gimplify_save_expr): Likewise.
19334 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19336 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
19338 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19340 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
19342 * timevar.def (TV_IPA_FNSUMMARY): Define.
19344 2017-07-16 Daniel Cederman <cederman@gaisler.com>
19346 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
19347 to back store errata sensitive sequence from being generated.
19348 (sqrtdf2_fix): Likewise.
19350 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19352 * tree-ssa-threadupdate.c (compute_path_counts,
19353 update_joiner_offpath_counts): Use profile_probability.
19355 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
19358 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19360 * config/arm/arm-c.c (arm_cpu_builtins): Define
19361 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
19363 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
19365 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19366 array entries to represent __ieee128 versions of the
19367 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
19368 scalar_extract_sig, and scalar_insert_exp built-in functions.
19369 (altivec_resolve_overloaded_builtin): Add special case handling
19370 for the __builtin_scalar_insert_exp function, as represented by
19371 the P9V_BUILTIN_VEC_VSIEDP constant.
19372 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
19373 exponent support for __ieee128 argument.
19374 (VSESQP): Add scalar extract signature support for __ieee128
19376 (VSTDCNQP): Add scalar test negative support for __ieee128
19378 (VSIEQP): Add scalar insert exponent support for __int128 argument
19379 with __ieee128 result.
19380 (VSIEQPF): Add scalar insert exponent support for __ieee128
19381 argument with __ieee128 result.
19382 (VSTDCQP): Add scalar test data class support for __ieee128
19384 (VSTDCNQP): Add overload support for scalar test negative with
19385 __ieee128 argument.
19386 (VSTDCQP): Add overload support for scalar test data class
19387 __ieee128 argument.
19388 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
19389 UNSPEC_VSX_SXSIGDP.
19390 (UNSPEC_VSX_SIEXPQP): New constant.
19391 (xsxexpqp): New insn for VSX scalar extract exponent quad
19393 (xsxsigqp): New insn for VSX scalar extract significand quad
19395 (xsiexpqpf): New insn for VSX scalar insert exponent quad
19396 precision with floating point argument.
19397 (xststdcqp): New expand for VSX scalar test data class quad
19399 (xststdcnegqp): New expand for VSX scalar test negative quad
19401 (xststdcqp): New insn to match expansions for VSX scalar test data
19402 class quad precision and VSX scalar test negative quad precision.
19403 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
19404 special case operand checking to enforce that second operand of
19405 VSX scalar test data class with quad precision argument is a 7-bit
19407 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19408 prototypes and descriptions of __ieee128 versions of
19409 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
19410 scalar_test_data_class, and scalar_test_neg built-in functions.
19412 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19414 PR tree-optimization/81162
19415 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
19416 replace a negate with an add.
19418 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
19420 * doc/invoke.texi (arm/-mcpu): Document +crypto.
19422 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19424 * config/arm/arm-c.c (arm_cpu_builtins): Define
19425 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
19427 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19429 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
19430 (armv8-r): Set ARM Cortex-R52 as default CPU.
19431 * config/arm/arm-tables.opt: Regenerate.
19432 * config/arm/arm-tune.md: Regenerate.
19433 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
19435 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
19436 extension for -mcpu=cortex-r52.
19438 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19440 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
19441 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
19442 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
19443 (fp-armv8): Define it as FP_ARMv8 only.
19444 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
19445 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
19447 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
19448 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
19449 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
19450 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
19451 than TARGET_FPU_ARMV8.
19452 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
19453 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
19454 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
19456 * config/arm/neon.md (neon_vrint): Likewise.
19457 (neon_vcvt): Likewise.
19458 (neon_<fmaxmin_op><mode>): Likewise.
19459 (<fmaxmin><mode>3): Likewise.
19460 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
19461 * config/arm/predicates.md (arm_cond_move_operator): Check against
19462 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
19464 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
19466 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
19467 to top of function.
19469 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19471 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
19472 loop in comment with memset.
19474 2017-07-14 Martin Liska <mliska@suse.cz>
19476 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
19477 * dwarf2out.c (is_java): Remove the function.
19478 (output_pubname): Remove usage of the function.
19479 (lower_bound_default): Remove usage of DW_LANG_Java.
19480 (gen_compile_unit_die): Likewise.
19481 * gcc.c: Remove compiler defaults for .java and .zip files.
19482 * gimple-expr.c (remove_suffix): Change as there's no longer
19483 extension than 4-letter one.
19484 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
19485 (gimplify_save_expr): Likewise.
19486 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
19487 as it's possible even for other languages than Java.
19488 * langhooks.h (struct lang_hooks): Remove Java from a comment.
19489 * lto-opts.c (lto_write_options): Remove reference to Java.
19490 * opts.c (strip_off_ending): Update file extension handling.
19491 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
19492 * tree-eh.c (lower_resx): Likewise.
19493 * tree.c (free_lang_data_in_type): Remove dead code.
19494 (find_decls_types_r): Likewise.
19495 (build_common_builtin_nodes): Remove Java from a comment.
19496 (verify_type): Remove dead code.
19497 * varasm.c (assemble_external): Remove Java from a comment.
19499 2017-07-14 Martin Liska <mliska@suse.cz>
19501 * opts.c (finish_options): Add quotes.
19502 (common_handle_option): Likewise.
19504 2017-07-14 Martin Liska <mliska@suse.cz>
19506 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
19507 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
19508 Remove N_SO_PASCAL.
19509 * dwarf2out.c (lower_bound_default): Do not handle
19511 (gen_compile_unit_die): Likewise.
19512 * gcc.c: Remove default extension binding for GNU Pascal.
19513 * stmt.c: Remove Pascal language from a comment.
19514 * xcoffout.c: Likewise.
19516 2017-07-13 David Malcolm <dmalcolm@redhat.com>
19519 * diagnostic-show-locus.c (fixit_cmp): New function.
19520 (layout::layout): Sort m_fixit_hints.
19521 (column_range::column_range): Assert that the values are valid.
19522 (struct char_span): New struct.
19523 (correction::overwrite): New method.
19524 (struct source_line): New struct.
19525 (line_corrections::add_hint): Add assertions. Reimplement memcpy
19526 calls in terms of classes source_line and char_span, and
19527 correction::overwrite.
19528 (selftest::test_overlapped_fixit_printing_2): New function.
19529 (selftest::diagnostic_show_locus_c_tests): Call it.
19531 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
19533 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
19534 early if there is no lhs.
19536 2017-07-13 Martin Liska <mliska@suse.cz>
19538 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
19539 (gen_reference_type_die): Likewise.
19540 * stor-layout.c: Remove Pascal-related comment.
19542 2017-07-13 Martin Liska <mliska@suse.cz>
19544 * opts.c (finish_options): Add quotes to error messages.
19545 (parse_sanitizer_options): Likewise.
19547 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
19549 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
19551 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
19553 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
19555 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
19557 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
19559 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
19561 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
19564 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
19565 provides the hardware capability bits, define the macro
19566 __BUILTIN_CPU_SUPPORTS__.
19567 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
19568 if GLIBC does not provide the hardware capability bits. Add a
19569 gcc_unreachable call if the built-in cpu function is neither
19570 __builtin_cpu_is nor __builtin_cpu_supports.
19571 (rs6000_get_function_versions_dispatcher): Change the warning
19572 that an old GLIBC is used which does not export the capability
19573 bits to be an error.
19574 * doc/extend.texi (target_clones attribute): Document the
19575 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
19576 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
19577 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
19578 the macros defined by GCC if the newer GLIBC is available.
19580 2017-07-12 Jeff Law <law@redhat.com>
19582 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
19583 remaining includes slightly.
19584 * config/riscv/riscv-builtins.c: Include profile-count.h.
19586 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
19589 * config/avr/avr.c (avr_set_current_function): In diagnostic
19590 messages: Quote keywords and (parts of) identifiers.
19591 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
19594 2017-07-12 Carl Love <cel@us.ibm.com>
19596 * config/rs6000/rs6000-c.c: Add support for built-in functions
19597 vector bool char vec_revb (vector bool char);
19598 vector bool short vec_revb (vector short char);
19599 vector bool int vec_revb (vector bool int);
19600 vector bool long long vec_revb (vector bool long long);
19601 * doc/extend.texi: Update the built-in documentation file for the
19602 new built-in functions.
19604 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19606 * config/s390/s390.md: Remove movcc splitter.
19608 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19610 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
19611 load/store on condition.
19613 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
19616 * config/avr/avr.c (avr_encode_section_info)
19617 [progmem && !TREE_READONLY]: Error if progmem object needs
19620 2017-07-11 Michael Collison <michael.collison@arm.com>
19622 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
19625 2017-07-11 Carl Love <cel@us.ibm.com>
19627 * config/rs6000/rs6000-c.c: Add support for builtins
19628 vector unsigned int vec_parity_lsbb (vector signed int);
19629 vector unsigned int vec_parity_lsbb (vector unsigned int);
19630 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
19631 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
19632 vector unsigned long long vec_parity_lsbb (vector signed long long);
19633 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
19634 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
19635 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
19636 * doc/extend.texi: Update the built-in documentation file for the
19637 new built-in functions.
19639 2017-07-11 David Malcolm <dmalcolm@redhat.com>
19641 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
19642 (layout::m_primary_loc): New field.
19643 (layout::layout): Initialize new field. Move location filtering
19644 logic from here to...
19645 (layout::maybe_add_location_range): ...this new method. Add
19646 support for filtering to just the lines already specified by other
19648 (layout::will_show_line_p): New method.
19649 (gcc_rich_location::add_location_if_nearby): New method.
19650 (selftest::test_add_location_if_nearby): New test function.
19651 (selftest::diagnostic_show_locus_c_tests): Call it.
19652 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
19655 2017-07-11 Tom de Vries <tom@codesourcery.com>
19657 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
19658 (bb_first_real_insn): New function.
19659 (nvptx_single): Add extra initialization of broadcasted condition
19662 2017-07-11 Nathan Sidwell <nathan@acm.org>
19664 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
19666 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
19668 * doc/extend.texi (AVR Function Attributes): Remove weblink to
19669 Binutils doc as TEXI will mess them up.
19670 * doc/invoke.texi (AVR Options): Same here.
19672 2017-07-11 Daniel Cederman <cederman@gaisler.com>
19674 * config/sparc/sparc.opt (mfix-ut700): New option.
19675 (mfix-gr712rc): Likewise.
19676 (sparc_fix_b2bst): New variable.
19677 * doc/invoke.texi (SPARC options): Document them.
19678 (ARM options): Fix warnings.
19679 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
19680 instructions to prevent sequences that can trigger the store-store
19681 errata for certain LEON3FT processors.
19682 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
19683 (sparc_option_override): Set sparc_fix_b2bst appropriately.
19684 * config/sparc/sparc.md (fix_b2bst): New attribute.
19685 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
19687 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
19690 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
19692 (*rsqrtsf2_sse): Ditto.
19694 (div<mode>3): Macroize insn from divdf3 and divsf3
19695 using MODEF mode iterator.
19697 2017-07-10 Martin Sebor <msebor@redhat.com>
19699 PR tree-optimization/80397
19700 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
19701 instead of testing for equality to INTEGER_TYPE.
19703 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
19705 * config.gcc: Remove uclibc from arc target spec.
19707 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
19709 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
19711 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19714 * lto-wrapper.c (remove_option): New function.
19715 (merge_and_complain): Merge PIC/PIE options more realistically.
19717 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
19719 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
19723 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
19724 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
19725 * config.in: Regenerate.
19726 * configure: Regenerate.
19727 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
19728 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
19729 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
19730 (TARGET_GASISR_PROLOGUES): ...target mask.
19731 * common/config/avr/avr-common.c
19732 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
19733 Set -mgas-isr-prologues.
19734 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
19735 INSERT_PASS_BEFORE for it.
19736 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
19737 * config/avr/avr.c (avr_option_override)
19738 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
19739 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
19740 (avr_attribute_table) <no_gccisr>: Add new function attribute.
19741 (avr_set_current_function) <is_no_gccisr>: Init machine field.
19742 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
19744 (make_avr_pass_pre_proep): New function.
19745 (emit_push_sfr) <treg>: Add argument to function and use it
19746 instead of TMP_REG.
19747 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
19748 and set machine->gasisr.yes.
19749 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
19750 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
19751 __gcc_isr.n_pushed to .L__stack_usage.
19752 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
19753 (avr_asm_final_postscan_insn): ...this new static function.
19754 * config/avr/avr.h (machine_function)
19755 <is_no_gccisr, use_L__stack_usage>: New fields.
19756 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
19757 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
19758 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
19759 (gasisr, *gasisr): New expander and insn.
19760 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
19761 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
19762 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
19764 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
19766 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
19769 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
19771 Move jump-tables out of .text again.
19774 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
19775 (ASM_OUTPUT_ADDR_VEC): New function.
19776 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
19777 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
19778 INSN_ADDRESSes as asm comment.
19779 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
19780 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
19781 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
19782 * config/avr/avr.md (*tablejump): Adjust comment.
19783 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
19784 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
19786 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
19787 (avr_output_addr_vec): New proto.
19788 (avr_log_t) <insn_addresses>: New field.
19790 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
19793 * config/i386/i386.c (ix86_function_arg_advance): Set
19794 outgoing_args_on_stack to true if there are outgoing arguments
19796 (ix86_function_arg): Likewise.
19797 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
19798 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
19799 * config/i386/i386.h (machine_function): Add
19800 outgoing_args_on_stack.
19802 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
19804 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
19805 supporting pthreds.
19806 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
19808 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
19810 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
19811 (REAL_H): Remove $(MACHMODE_H).
19812 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
19814 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
19815 $(MACHMODE_H) and double-int.h.
19816 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
19818 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
19821 2017-07-07 Andrew Pinski <apinski@cavium.com>
19823 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
19824 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
19826 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
19828 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
19829 Add warning if GCC was not configured to link against a GLIBC that
19830 exports the hardware capability bits.
19831 (make_resolver_func): Make resolver function private and not a
19832 COMDAT function. Create the name with clone_function_name instead
19833 of make_unique_name.
19836 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
19837 correct operand in doing the split.
19839 2017-07-07 Carl Love <cel@us.ibm.com>
19841 * config/rs6000/rs6000-c: Add support for built-in function
19842 vector unsigned short vec_pack_to_short_fp32 (vector float,
19844 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
19845 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
19846 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
19847 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
19848 (convert_4f32_8i16): Add define_expand.
19849 * doc/extend.texi: Update the built-in documentation file for the
19850 new built-in function.
19852 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19854 * config/sparc/m8.md: New file.
19855 * config/sparc/sparc.md: Include m8.md.
19857 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19859 * config/sparc/sparc.opt: New option -mvis4b.
19860 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
19861 (sparc_option_override): Handle VIS4B.
19862 (enum sparc_builtins): Define
19863 SPARC_BUILTIN_DICTUNPACK{8,16,32},
19864 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
19865 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
19866 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
19867 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
19868 (check_constant_argument): New function.
19869 (sparc_vis_init_builtins): Define builtins
19870 __builtin_vis_dictunpack{8,16,32},
19871 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
19872 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
19873 __builtin_vis_fpcmpde{8,16,32}shl and
19874 __builtin_vis_fpcmpur{8,16,32}shl.
19875 (sparc_expand_builtin): Check that the constant operands to
19876 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
19877 constant and in range.
19878 * config/sparc/sparc-c.c (sparc_target_macros): Handle
19880 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
19881 (SPARC_IMM5_P): Likewise.
19882 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
19883 (enabled): Handle vis4b.
19884 (UNSPEC_DICTUNPACK): New unspec.
19885 (UNSPEC_FPCMPSHL): Likewise.
19886 (UNSPEC_FPUCMPSHL): Likewise.
19887 (UNSPEC_FPCMPDESHL): Likewise.
19888 (UNSPEC_FPCMPURSHL): Likewise.
19889 (cpu_feature): New CPU feature `vis4b'.
19890 (dictunpack{8,16,32}): New insns.
19891 (FPCSMODE): New mode iterator.
19892 (fpcscond): New code iterator.
19893 (fpcsucond): Likewise.
19894 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
19895 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
19896 (fpcmpde{8,16,32}{si,di}shl): Likewise.
19897 (fpcmpur{8,16,32}{si,di}shl): Likewise.
19898 * config/sparc/constraints.md: Define constraints `q' for unsigned
19899 2-bit integer constants and `t' for unsigned 5-bit integer
19901 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
19903 (imm5_operand_dictunpack16): Likewise.
19904 (imm5_operand_dictunpack32): Likewise.
19905 (imm2_operand): Likewise.
19906 * doc/invoke.texi (SPARC Options): Document -mvis4b.
19907 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
19908 ditunpack* and fpcmp*shl builtins.
19910 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19912 * config.gcc: Handle m8 in --with-{cpu,tune} options.
19913 * config.in: Add HAVE_AS_SPARC6 define.
19914 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
19916 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
19918 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
19919 (CPP_CPU_SPEC): Handle m8.
19920 (ASM_CPU_SPEC): Likewise.
19921 * config/sparc/sparc-opts.h (enum processor_type): Add
19923 * config/sparc/sparc.c (m8_costs): New struct.
19924 (sparc_option_override): Handle TARGET_CPU_m8.
19925 (sparc32_initialize_trampoline): Likewise.
19926 (sparc64_initialize_trampoline): Likewise.
19927 (sparc_issue_rate): Likewise.
19928 (sparc_register_move_cost): Likewise.
19929 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
19930 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
19931 (ASM_CPU64_DEFAULT_SPEC): Likewise.
19932 (CPP_CPU_SPEC): Handle M8.
19933 (ASM_CPU_SPEC): Likewise.
19934 (AS_M8_FLAG): Define.
19935 * config/sparc/sparc.md: Add m8 to the cpu attribute.
19936 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
19937 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
19939 * configure: Regenerate.
19940 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
19943 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19945 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
19947 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
19948 ("*movdi_insn_sp32"): Do not set v3pipe.
19949 ("*movsi_insn"): Likewise.
19950 ("*movdi_insn_sp64"): Likewise.
19951 ("*movsf_insn"): Likewise.
19952 ("*movdf_insn_sp32"): Likewise.
19953 ("*movdf_insn_sp64"): Likewise.
19954 ("*zero_extendsidi2_insn_sp64"): Likewise.
19955 ("*sign_extendsidi2_insn"): Likewise.
19956 ("*mov<VM32:mode>_insn"): Likewise.
19957 ("*mov<VM64:mode>_insn_sp64"): Likewise.
19958 ("*mov<VM64:mode>_insn_sp32"): Likewise.
19959 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
19960 ("<vlop:code><VL:mode>3"): Likewise.
19961 ("*not_<vlop:code><VL:mode>3"): Likewise.
19962 ("*nand<VL:mode>_vis"): Likewise.
19963 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
19964 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
19965 ("one_cmpl<VL:mode>2"): Likewise.
19966 ("faligndata<VM64:mode>_vis"): Likewise.
19967 ("alignaddrsi_vis"): Likewise.
19968 ("alignaddrdi_vis"): Likweise.
19969 ("alignaddrlsi_vis"): Likewise.
19970 ("alignaddrldi_vis"): Likewise.
19971 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
19972 ("bmaskdi_vis"): Likewise.
19973 ("bmasksi_vis"): Likewise.
19974 ("bshuffle<VM64:mode>_vis"): Likewise.
19975 ("cmask8<P:mode>_vis"): Likewise.
19976 ("cmask16<P:mode>_vis"): Likewise.
19977 ("cmask32<P:mode>_vis"): Likewise.
19978 ("pdistn<P:mode>_vis"): Likewise.
19979 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
19981 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19983 * config/sparc/sparc.md ("subtype"): New insn attribute.
19984 ("*wrgsr_sp64"): Set insn subtype.
19985 ("*rdgsr_sp64"): Likewise.
19986 ("alignaddrsi_vis"): Likewise.
19987 ("alignaddrdi_vis"): Likewise.
19988 ("alignaddrlsi_vis"): Likewise.
19989 ("alignaddrldi_vis"): Likewise.
19990 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
19991 ("fexpand_vis"): Likewise.
19992 ("fpmerge_vis"): Likewise.
19993 ("faligndata<VM64:mode>_vis"): Likewise.
19994 ("bshuffle<VM64:mode>_vis"): Likewise.
19995 ("cmask8<P:mode>_vis"): Likewise.
19996 ("cmask16<P:mode>_vis"): Likewise.
19997 ("cmask32<P:mode>_vis"): Likewise.
19998 ("fchksm16_vis"): Likewise.
19999 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
20000 ("fmean16_vis"): Likewise.
20001 ("fp<plusminus_insn>64_vis"): Likewise.
20002 ("<plusminus_insn>v8qi3"): Likewise.
20003 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
20004 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
20005 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
20006 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
20007 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
20008 ("*movqi_insn"): Likewise.
20009 ("*movhi_insn"): Likewise.
20010 ("*movsi_insn"): Likewise.
20011 ("movsi_pic_gotdata_op"): Likewise.
20012 ("*movdi_insn_sp32"): Likewise.
20013 ("*movdi_insn_sp64"): Likewise.
20014 ("movdi_pic_gotdata_op"): Likewise.
20015 ("*movsf_insn"): Likewise.
20016 ("*movdf_insn_sp32"): Likewise.
20017 ("*movdf_insn_sp64"): Likewise.
20018 ("*zero_extendhisi2_insn"): Likewise.
20019 ("*zero_extendqihi2_insn"): Likewise.
20020 ("*zero_extendqisi2_insn"): Likewise.
20021 ("*zero_extendqidi2_insn"): Likewise.
20022 ("*zero_extendhidi2_insn"): Likewise.
20023 ("*zero_extendsidi2_insn_sp64"): Likewise.
20024 ("ldfsr"): Likewise.
20025 ("prefetch_64"): Likewise.
20026 ("prefetch_32"): Likewise.
20027 ("tie_ld32"): Likewise.
20028 ("tie_ld64"): Likewise.
20029 ("*tldo_ldub_sp32"): Likewise.
20030 ("*tldo_ldub1_sp32"): Likewise.
20031 ("*tldo_ldub2_sp32"): Likewise.
20032 ("*tldo_ldub_sp64"): Likewise.
20033 ("*tldo_ldub1_sp64"): Likewise.
20034 ("*tldo_ldub2_sp64"): Likewise.
20035 ("*tldo_ldub3_sp64"): Likewise.
20036 ("*tldo_lduh_sp32"): Likewise.
20037 ("*tldo_lduh1_sp32"): Likewise.
20038 ("*tldo_lduh_sp64"): Likewise.
20039 ("*tldo_lduh1_sp64"): Likewise.
20040 ("*tldo_lduh2_sp64"): Likewise.
20041 ("*tldo_lduw_sp32"): Likewise.
20042 ("*tldo_lduw_sp64"): Likewise.
20043 ("*tldo_lduw1_sp64"): Likewise.
20044 ("*tldo_ldx_sp64"): Likewise.
20045 ("*mov<VM32:mode>_insn"): Likewise.
20046 ("*mov<VM64:mode>_insn_sp64"): Likewise.
20047 ("*mov<VM64:mode>_insn_sp32"): Likewise.
20049 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20051 * config/sparc/sparc.md ("type"): New insn type viscmp.
20052 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
20054 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
20055 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
20056 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
20057 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
20059 ("n7_vis_logical_11cycle"): Likewise.
20060 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
20061 * config/sparc/niagara2.md ("niag3_vis": Likewise.
20062 * config/sparc/niagara.md ("niag_vis"): Likewise.
20063 * config/sparc/ultra3.md ("us3_fga"): Likewise.
20064 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
20066 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20068 * config/sparc/sparc.md: New instruction type `bmask'.
20069 (bmaskdi_vis): Use the `bmask' type.
20070 (bmasksi_vis): Likewise.
20071 * config/sparc/ultra3.md (us3_array): Likewise.
20072 * config/sparc/niagara7.md (n7_array): Likewise.
20073 * config/sparc/niagara4.md (n4_array): Likewise.
20074 * config/sparc/niagara2.md (niag2_vis): Likewise.
20075 (niag3_vis): Likewise.
20076 * config/sparc/niagara.md (niag_vis): Likewise.
20078 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20080 * ipa-comdats.c: Remove optimize check from gate.
20081 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
20082 for functions not optimized.
20083 (ipa_fn_summary_read): Skip optimize check.
20084 (ipa_fn_summary_write): Likewise.
20085 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
20087 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
20089 (can_inline_edge_p): Check flag_pcc_struct_return for match.
20090 (check_callers): Give up on caller which is not optimized.
20091 (inline_small_functions): Likewise.
20092 (ipa_inline): Do not give up when not optimizing.
20093 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
20094 away unoptimizes cdtors.
20095 (whole_program_function_and_variable_visibility): Do
20096 ipa_discover_readonly_nonaddressable_vars in LTO mode.
20097 * ipa.c (process_references): Do not check optimize.
20098 (symbol_table::remove_unreachable_nodes): Update optimize check.
20099 (set_writeonly_bit): Update optimize check.
20100 (pass_ipa_cdtor_merge::gate): Do not check optimize.
20101 (pass_ipa_single_use::gate): Remove.
20103 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20105 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
20106 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
20107 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
20108 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
20109 permute_load, permute_store, adjust_extract, adjust_splat,
20110 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
20111 replace_swap_with_copy, dump_swap_insn_table,
20112 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
20113 recombine_lvx_pattern, recombine_stvx_pattern,
20114 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
20115 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
20116 to file rs6000-p8swap.c.
20117 * config/rs6000/rs6000-p8swap.c: New file.
20118 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
20119 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
20120 and rs6000*-*-* targets.
20122 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20124 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
20126 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20128 * lto-wrapper.c (merge_and_complain): Do not merge
20129 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
20130 fsigned_zeros, ftrapping_math, fwrapv.
20131 (append_compiler_options): Do not track these options.
20132 (append_linker_options): Likewie
20134 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20136 * cgraphunit.c (cgraph_node::finalize_function): When
20137 !flag_toplevel_reorde set no_reorder flag.
20138 (varpool_node::finalize_decl): Likewise.
20139 (symbol_table::compile): Drop no toplevel reorder path.
20141 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20143 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
20144 edges; zero probability is not better than uninitialized.
20146 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
20148 * asan.h (asan_sanitize_allocas_p): Declare.
20149 * asan.c (asan_sanitize_allocas_p): New function.
20150 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
20151 (handle_builtin_alloca): Likewise.
20152 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
20153 if !asan_sanitize_allocas_p.
20154 * params.def (asan-instrument-allocas): Add new option.
20155 * params.h (ASAN_PROTECT_ALLOCAS): Define.
20156 * opts.c (common_handle_option): Disable allocas sanitization for
20159 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
20161 * asan.c: Include gimple-fold.h.
20162 (get_last_alloca_addr): New function.
20163 (handle_builtin_stackrestore): Likewise.
20164 (handle_builtin_alloca): Likewise.
20165 (asan_emit_allocas_unpoison): Likewise.
20166 (get_mem_refs_of_builtin_call): Add new parameter, remove const
20167 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
20168 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
20169 (instrument_builtin_call): Pass gimple iterator to
20170 get_mem_refs_of_builtin_call.
20171 (last_alloca_addr): New global.
20172 * asan.h (asan_emit_allocas_unpoison): Declare.
20173 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
20174 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
20175 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
20176 if function calls alloca.
20177 * gimple-fold.c (replace_call_with_value): Remove static keyword.
20178 * gimple-fold.h (replace_call_with_value): Declare.
20179 * internal-fn.c: Include asan.h.
20180 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
20181 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
20183 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20185 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
20186 (C_SELFTEST_FLAGS): New.
20187 (CPP_SELFTEST_FLAGS): New.
20188 (SELFTEST_DEPS): New, from deps of s-selftest.
20189 (C_SELFTEST_DEPS): New, from deps of s-selftest.
20190 (CPP_SELFTEST_DEPS): New.
20191 (selftest): Add dependency on s-selftest-c++.
20192 (s-selftest): Rename to...
20193 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
20194 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
20195 than SELFTEST_FLAGS.
20196 (selftest-gdb): Rename to...
20197 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
20199 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
20200 (selftest-valgrind): Rename to...
20201 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
20203 (selftest-valgrind): Reintroduce as an alias for
20204 selftest-c-valgrind.
20205 (s-selftest-c++): New.
20206 (selftest-c++-gdb): New.
20207 (selftest-c++-valgrind): New.
20209 2017-07-06 Olivier Hainque <hainque@adacore.com>
20211 * gcc.c (process_command): When deciding if undefined variables
20212 should be ignored when processing specs, accept "gcc -v" as well.
20214 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20216 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
20217 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
20219 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
20221 * config/arm/arm-cpus.in (armv8-r): Add new entry.
20222 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
20223 * config/arm/arm-tables.opt: Regenerate.
20224 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
20226 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
20228 2017-07-06 Carl Love <cel@us.ibm.com>
20230 * ChangeLog: Clean up from mid air collision
20232 2017-07-06 Carl Love <cel@us.ibm.com>
20234 * config/rs6000/rs6000-c.c: Add support for built-in functions
20235 vector signed int vec_subc (vector signed int, vector signed int);
20236 vector signed __int128 vec_subc (vector signed __int128,
20237 vector signed __int128);
20238 vector unsigned __int128 vec_subc (vector unsigned __int128,
20239 vector unsigned __int128);
20240 vector signed int vec_sube (vector signed int, vector signed int,
20241 vector signed int);
20242 vector unsigned int vec_sube (vector unsigned int,
20243 vector unsigned int,
20244 vector unsigned int);
20245 vector signed __int128 vec_sube (vector signed __int128,
20246 vector signed __int128,
20247 vector signed__int128);
20248 vector unsigned __int128 vec_sube (vector unsigned __int128,
20249 vector unsigned __int128,
20250 vector unsigned __int128);
20251 vector signed int vec_subec (vector signed int, vector signed int,
20252 vector signed int);
20253 vector unsigned int vec_subec (vector unsigned int,
20254 vector unsigned int,
20255 vector unsigned int);
20256 vector signed __int128 vec_subec (vector signed __int128,
20257 vector signed __int128,
20258 vector signed__int128);
20259 vector unsigned __int128 vec_subec (vector unsigned __int128,
20260 vector unsigned __int128,
20261 vector unsigned __int128);
20262 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
20263 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
20264 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
20265 BU_ALTIVEC_OVERLOAD_X definitions.
20266 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
20267 * doc/extend.texi: Update the built-in documentation file for the new
20268 built-in functions.
20270 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20273 * diagnostic-show-locus.c (layout::layout): Use start and finish
20274 spelling location for the start and finish of each range.
20275 * genmatch.c (linemap_client_expand_location_to_spelling_point):
20276 Add unused aspect param.
20277 * input.c (expand_location_1): Add "aspect" param, and use it
20278 to access the correct part of the location.
20279 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
20281 (expand_location_to_spelling_point): Likewise.
20282 (linemap_client_expand_location_to_spelling_point): Add "aspect"
20283 param, and pass it to expand_location_1.
20285 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
20287 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
20288 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
20289 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
20290 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
20291 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
20292 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
20293 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
20294 _mm_maskz_getmant_ss): New intrinsics.
20295 (__builtin_ia32_getexpss128_mask): Changed to ...
20296 __builtin_ia32_getexpss128_round ... this.
20297 (__builtin_ia32_getexpsd128_mask): Changed to ...
20298 __builtin_ia32_getexpsd128_round ... this.
20299 * config/i386/i386-builtin-types.def
20300 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
20301 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
20302 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
20303 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
20304 __builtin_ia32_getmantss_mask_round): New builtins.
20305 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
20306 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
20307 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
20308 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
20309 * config/i386/sse.md
20310 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
20311 avx512f_sgetexp<mode><mask_scalar_name>
20312 <round_saeonly_scalar_name> ... this.
20313 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
20314 %0, %1, %2<round_saeonly_op3>}): Changed to ...
20315 vgetexp<ssescalarmodesuffix>
20316 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20317 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
20318 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
20319 avx512f_vgetmant<mode><mask_scalar_name>
20320 <round_saeonly_scalar_name> ... this.
20321 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
20322 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
20323 vgetmant<ssescalarmodesuffix>
20324 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
20325 %0<mask_scalar_operand4>, %1, %2
20326 <round_saeonly_scalar_mask_op4>, %3} ... this.
20327 * config/i386/subst.md (mask_scalar_operand4,
20328 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
20329 round_saeonly_scalar_nimm_predicate): New subst attributes.
20331 2017-07-06 Julia Koval <julia.koval@intel.com>
20333 * config/i386/i386.c (ix86_erase_embedded_rounding):
20334 Remove code for old rounding pattern.
20336 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
20338 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
20340 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
20342 * doc/sourcebuild.texi (Test Directives, Variants of
20343 dg-require-support): Add documentation for dg-require-stack-check.
20345 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
20347 * config/i386/subst.md (mask_scalar, round_scalar,
20348 round_saeonly_scalar): New meta-templates.
20349 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
20350 round_scalar_mask_operand3, round_scalar_mask_op3,
20351 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
20352 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
20353 round_saeonly_scalar_constraint,
20354 round_saeonly_scalar_prefix): New subst attribute.
20355 * config/i386/sse.md
20356 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
20357 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
20358 <round_scalar_name> ... this.
20359 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
20360 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
20361 <round_scalar_name> ... this.
20362 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
20363 <sse>_vm<code><mode>3<mask_scalar_name>
20364 <round_saeonly_scalar_name> ... this.
20365 (v<plusminus_mnemonic><ssescalarmodesuffix>
20366 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
20367 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
20368 v<plusminus_mnemonic><ssescalarmodesuffix>
20369 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20370 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
20371 (v<multdiv_mnemonic><ssescalarmodesuffix>
20372 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
20373 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
20374 v<multdiv_mnemonic><ssescalarmodesuffix>
20375 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20376 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
20377 (v<maxmin_float><ssescalarmodesuffix>
20378 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
20379 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
20380 v<maxmin_float><ssescalarmodesuffix>
20381 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20382 %0<mask_scalar_operand3>, %1, %<iptr>2
20383 <round_saeonly_scalar_mask_op3>} ... this.
20385 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
20387 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
20388 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
20390 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
20391 Alan Hayward <alan.hayward@arm.com>
20392 David Sherwood <david.sherwood@arm.com>
20394 * combine.c (simplify_if_then_else): Remove "enum" before
20396 * compare-elim.c (can_eliminate_compare): Likewise.
20397 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
20399 (aarch64_lookup_simd_builtin_type): Likewise.
20400 (aarch64_simd_builtin_type): Likewise.
20401 (aarch64_init_simd_builtin_types): Likewise.
20402 (aarch64_simd_expand_args): Likewise.
20403 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
20405 (aarch64_reverse_mask): Likewise.
20406 (aarch64_simd_emit_reg_reg_move): Likewise.
20407 (aarch64_gen_adjusted_ldpstp): Likewise.
20408 (aarch64_ccmp_mode_to_code): Likewise.
20409 (aarch64_operands_ok_for_ldpstp): Likewise.
20410 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
20411 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
20413 (aarch64_min_divisions_for_recip_mul): Likewise.
20414 (aarch64_reassociation_width): Likewise.
20415 (aarch64_get_condition_code_1): Likewise.
20416 (aarch64_simd_emit_reg_reg_move): Likewise.
20417 (aarch64_simd_attr_length_rglist): Likewise.
20418 (aarch64_reverse_mask): Likewise.
20419 (aarch64_operands_ok_for_ldpstp): Likewise.
20420 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
20421 (aarch64_gen_adjusted_ldpstp): Likewise.
20422 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
20424 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
20425 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
20426 (arm_lookup_simd_builtin_type): Likewise.
20427 (arm_simd_builtin_type): Likewise.
20428 (arm_init_simd_builtin_types): Likewise.
20429 (arm_expand_builtin_args): Likewise.
20430 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
20431 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
20432 (ft32_setup_incoming_varargs): Likewise.
20433 (ft32_function_arg): Likewise.
20434 (ft32_function_arg_advance): Likewise.
20435 (ft32_pass_by_reference): Likewise.
20436 (ft32_arg_partial_bytes): Likewise.
20437 (ft32_valid_pointer_mode): Likewise.
20438 (ft32_addr_space_pointer_mode): Likewise.
20439 (ft32_addr_space_legitimate_address_p): Likewise.
20440 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
20442 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
20443 (ix86_emit_outlined_ms2sysv_restore): Likewise.
20444 (iamcu_alignment): Likewise.
20445 (canonicalize_vector_int_perm): Likewise.
20446 (ix86_noce_conversion_profitable_p): Likewise.
20447 (ix86_mpx_bound_mode): Likewise.
20448 (ix86_operands_ok_for_move_multiple): Likewise.
20449 * config/microblaze/microblaze-protos.h
20450 (microblaze_expand_conditional_branch_reg): Likewise.
20451 * config/microblaze/microblaze.c
20452 (microblaze_expand_conditional_branch_reg): Likewise.
20453 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
20455 (rs6000_reassociation_width): Likewise.
20456 (rs6000_invalid_binary_op): Likewise.
20457 (fusion_p9_p): Likewise.
20458 (emit_fusion_p9_load): Likewise.
20459 (emit_fusion_p9_store): Likewise.
20460 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
20462 (riscv_hard_regno_mode_ok_p): Likewise.
20463 (riscv_address_insns): Likewise.
20464 (riscv_split_symbol): Likewise.
20465 (riscv_legitimize_move): Likewise.
20466 (riscv_function_value): Likewise.
20467 (riscv_hard_regno_nregs): Likewise.
20468 (riscv_expand_builtin): Likewise.
20469 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
20470 (riscv_build_integer): Likewise.
20471 (riscv_split_integer): Likewise.
20472 (riscv_legitimate_constant_p): Likewise.
20473 (riscv_cannot_force_const_mem): Likewise.
20474 (riscv_regno_mode_ok_for_base_p): Likewise.
20475 (riscv_valid_base_register_p): Likewise.
20476 (riscv_valid_offset_p): Likewise.
20477 (riscv_valid_lo_sum_p): Likewise.
20478 (riscv_classify_address): Likewise.
20479 (riscv_legitimate_address_p): Likewise.
20480 (riscv_address_insns): Likewise.
20481 (riscv_load_store_insns): Likewise.
20482 (riscv_force_binary): Likewise.
20483 (riscv_split_symbol): Likewise.
20484 (riscv_force_address): Likewise.
20485 (riscv_legitimize_address): Likewise.
20486 (riscv_move_integer): Likewise.
20487 (riscv_legitimize_const_move): Likewise.
20488 (riscv_legitimize_move): Likewise.
20489 (riscv_address_cost): Likewise.
20490 (riscv_subword): Likewise.
20491 (riscv_output_move): Likewise.
20492 (riscv_canonicalize_int_order_test): Likewise.
20493 (riscv_emit_int_order_test): Likewise.
20494 (riscv_function_arg_boundary): Likewise.
20495 (riscv_pass_mode_in_fpr_p): Likewise.
20496 (riscv_pass_fpr_single): Likewise.
20497 (riscv_pass_fpr_pair): Likewise.
20498 (riscv_get_arg_info): Likewise.
20499 (riscv_function_arg): Likewise.
20500 (riscv_function_arg_advance): Likewise.
20501 (riscv_arg_partial_bytes): Likewise.
20502 (riscv_function_value): Likewise.
20503 (riscv_pass_by_reference): Likewise.
20504 (riscv_setup_incoming_varargs): Likewise.
20505 (riscv_print_operand): Likewise.
20506 (riscv_elf_select_rtx_section): Likewise.
20507 (riscv_save_restore_reg): Likewise.
20508 (riscv_for_each_saved_reg): Likewise.
20509 (riscv_register_move_cost): Likewise.
20510 (riscv_hard_regno_mode_ok_p): Likewise.
20511 (riscv_hard_regno_nregs): Likewise.
20512 (riscv_class_max_nregs): Likewise.
20513 (riscv_memory_move_cost): Likewise.
20514 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
20515 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
20516 (rl78_addr_space_address_mode): Likewise.
20517 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20519 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20520 (rs6000_reassociation_width): Likewise.
20521 (rs6000_invalid_binary_op): Likewise.
20522 (fusion_p9_p): Likewise.
20523 (emit_fusion_p9_load): Likewise.
20524 (emit_fusion_p9_store): Likewise.
20525 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
20526 (ok_for_simple_move_operands): Likewise.
20527 (ok_for_simple_move_strict_operands): Likewise.
20528 (ok_for_simple_arith_logic_operands): Likewise.
20529 (visium_legitimize_reload_address): Likewise.
20530 (visium_select_cc_mode): Likewise.
20531 (output_cbranch): Likewise.
20532 (visium_split_double_move): Likewise.
20533 (visium_expand_copysign): Likewise.
20534 (visium_expand_int_cstore): Likewise.
20535 (visium_expand_fp_cstore): Likewise.
20536 * config/visium/visium.c (visium_pass_by_reference): Likewise.
20537 (visium_function_arg): Likewise.
20538 (visium_function_arg_advance): Likewise.
20539 (visium_libcall_value): Likewise.
20540 (visium_setup_incoming_varargs): Likewise.
20541 (visium_legitimate_constant_p): Likewise.
20542 (visium_legitimate_address_p): Likewise.
20543 (visium_legitimize_address): Likewise.
20544 (visium_secondary_reload): Likewise.
20545 (visium_register_move_cost): Likewise.
20546 (visium_memory_move_cost): Likewise.
20547 (prepare_move_operands): Likewise.
20548 (ok_for_simple_move_operands): Likewise.
20549 (ok_for_simple_move_strict_operands): Likewise.
20550 (ok_for_simple_arith_logic_operands): Likewise.
20551 (visium_function_value_1): Likewise.
20552 (rtx_ok_for_offset_p): Likewise.
20553 (visium_legitimize_reload_address): Likewise.
20554 (visium_split_double_move): Likewise.
20555 (visium_expand_copysign): Likewise.
20556 (visium_expand_int_cstore): Likewise.
20557 (visium_expand_fp_cstore): Likewise.
20558 (visium_split_cstore): Likewise.
20559 (visium_select_cc_mode): Likewise.
20560 (visium_split_cbranch): Likewise.
20561 (output_cbranch): Likewise.
20562 (visium_print_operand_address): Likewise.
20563 * expmed.c (flip_storage_order): Likewise.
20564 * expmed.h (emit_cstore): Likewise.
20565 (flip_storage_order): Likewise.
20566 * genrecog.c (validate_pattern): Likewise.
20567 * hsa-gen.c (gen_hsa_addr): Likewise.
20568 * internal-fn.c (expand_arith_overflow): Likewise.
20569 * ira-color.c (allocno_copy_cost_saving): Likewise.
20570 * lra-assigns.c (find_hard_regno_for_1): Likewise.
20571 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
20572 (process_invariant_for_inheritance): Likewise.
20573 * lra-eliminations.c (move_plus_up): Likewise.
20574 * omp-low.c (lower_oacc_reductions): Likewise.
20575 * simplify-rtx.c (simplify_subreg): Likewise.
20576 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
20577 (TARGET_CHKP_BOUND_MODE): Likewise..
20578 * targhooks.c (default_chkp_bound_mode): Likewise.
20579 (default_setup_incoming_vararg_bounds): Likewise.
20580 * targhooks.h (default_chkp_bound_mode): Likewise.
20581 (default_setup_incoming_vararg_bounds): Likewise.
20582 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
20583 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
20584 (have_whole_vector_shift): Likewise.
20585 * tree-vect-stmts.c (vectorizable_load): Likewise.
20586 * doc/tm.texi: Regenerate.
20588 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
20590 Graceful degrade if Binutils PR21472 is not available.
20593 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
20594 .rodata in flash test fails.
20595 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
20596 * confgure: Regenerate.
20597 * config.in: Regenerate.
20598 * config/avr/avr.c (avr_asm_named_section)
20599 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
20600 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
20601 (avr_asm_init_sections): Same.
20603 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20605 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
20606 (fma<VH:mode>4_intrinsic): Likewise.
20607 (*fmsub<VCVTF:mode>4): Likewise.
20608 (*fmsub<VH:mode>4_intrinsic): Likewise.
20610 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
20613 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
20614 Don't depend on "optimize > 0".
20615 (out_movhi_r_mr, out_movqi_mr_r): Same.
20616 (out_movhi_mr_r, out_movqi_r_mr): Same.
20617 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
20618 io_address_operand on "optimize > 0".
20620 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20622 * tree-loop-distribution.c: Add general explanantion on the pass.
20623 (generate_loops_for_partition): Mark distributed loop.
20624 (pg_add_dependence_edges): New parameter. Handle alias data
20625 dependence specially and record it in the parameter if asked.
20626 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
20627 (init_partition_graph_vertices, add_partition_graph_edge): New.
20628 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
20629 (free_partition_graph_vdata, build_partition_graph): New.
20630 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
20631 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
20632 (data_ref_segment_size, latch_dominated_by_data_ref): New.
20633 (compute_alias_check_pairs, version_loop_by_alias_check): New.
20634 (version_for_distribution_p, finalize_partitions): New.
20635 (distribute_loop): Handle alias data dependence specially. Factor
20636 out loop fusion code as functions and call these functions.
20638 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20640 * tree-loop-distribution.c (classify_partition): New parameter and
20641 better handle reduction statement.
20642 (rdg_build_partitions): Revise comment.
20643 (distribute_loop): Compute statements in all partitions and pass it
20644 to classify_partition.
20646 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20648 * tree-loop-distribution.c (enum partition_type): New.
20649 (struct partition): New field type.
20650 (partition_merge_into): Add parameter. Update partition type.
20651 (data_dep_in_cycle_p, update_type_for_merge): New functions.
20652 (build_rdg_partition_for_vertex): Compute partition type.
20653 (rdg_build_partitions): Dump partition type.
20654 (distribute_loop): Update calls to partition_merge_into.
20656 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20658 * tree-loop-distribution.c (struct ddr_hasher): New.
20659 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
20661 (classify_partition): Call get_data_dependence.
20662 (pg_add_dependence_edges): Ditto.
20663 (distribute_loop): Release data dependence hash table.
20665 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20667 * tree-loop-distribution.c (ref_base_address): Delete.
20668 (similar_memory_accesses): Rename ...
20669 (share_memory_accesses): ... to this. Check if partitions access
20670 the same memory reference.
20671 (distribute_loop): Call share_memory_accesses.
20673 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20675 * tree-loop-distribution.c (struct partition): New field recording
20676 its data reference.
20677 (partition_alloc, partition_free): Init and release data refs.
20678 (partition_merge_into): Merge data refs.
20679 (build_rdg_partition_for_vertex): Collect data refs for partition.
20680 (pg_add_dependence_edges): Change parameters from vector to bitmap.
20682 (distribute_loop): Remve data refs from vertice data of partition
20685 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20687 * tree-loop-distribution.c (params.h): Include header file.
20688 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
20689 (datarefs_vec): New global var.
20690 (create_rdg_vertices): Use datarefs_vec directly.
20691 (free_rdg): Don't free data references.
20692 (build_rdg): Update use. Don't free data references.
20693 (distribute_loop): Compute global variable for data references.
20694 Bail out if there are too many data references.
20696 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20698 * tree-loop-distribution.c (loop_nest): New global var.
20699 (build_rdg): Use loop directly, rather than loop nest.
20700 (pg_add_dependence_edges): Remove loop nest parameter. Use global
20702 (distribute_loop): Compute global variable loop nest. Update use.
20704 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20706 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
20707 (partition_merge_into): New parameter. Dump reason for fusion.
20708 (distribute_loop): Update use of partition_merge_into.
20710 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20712 * tree-loop-distribution.c (bb_top_order_index): New.
20713 (bb_top_order_index_size, bb_top_order_cmp): New.
20714 (stmts_from_loop): Use topological order.
20715 (pass_loop_distribution::execute): Compute and release topological
20716 order for basic blocks.
20718 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20720 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
20723 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20725 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
20726 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
20727 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
20728 * internal-fn.def (LOOP_DIST_ALIAS): New.
20729 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
20730 (fold_loop_internal_call): ... this.
20731 (vect_loop_dist_alias_call): New function.
20732 (set_uid_loop_bbs): Call fold_loop_internal_call.
20733 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
20736 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
20739 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
20740 Require dead FLAGS_REG at the beginning of a peephole.
20742 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
20745 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
20746 arguments in the call to __builtin_ia32_sbb_u32.
20747 (_subborrow_u64): Swap _X and _Y arguments in the call to
20748 __builtin_ia32_sbb_u64.
20750 2017-07-04 Jakub Jelinek <jakub@redhat.com>
20753 * tree-vrp.c (compare_assert_loc): Turn into a function template
20754 with stable template parameter. Only test if a->e is NULL,
20755 !a->e == !b->e has been verified already. Use e == NULL or
20756 e != NULL instead of e or ! e tests. If stable is true, don't use
20757 iterative_hash_expr, on the other side allow a or b or both NULL
20758 and sort the NULLs last.
20759 (process_assert_insertions): Sort using compare_assert_loc<false>
20760 instead of compare_assert_loc, later sort using
20761 compare_assert_loc<true> before calling process_assert_insertions_for
20762 in a loop. Use break instead of continue once seen NULL pointer.
20764 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20766 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
20767 Cortex-R7 and Cortex-R8 processors.
20769 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
20771 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
20772 uninitialized while src is not.
20774 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
20776 * common/config/arm/arm-common.c: Adjust include path for
20778 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
20779 (arm-cpu.h): Create in build directory. Adjust dependency rules.
20780 (arm-cpu-data.h): Likewise.
20781 (arm-cpu-cdata.h): Likewise.
20782 * config/arm/arm-cpu.h: Delete.
20783 * config/arm/arm-cpu-cdata.h: Delete.
20784 * config/arm/arm-cpu-data.h: Delete.
20786 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
20788 * config/arm/arm-cpus.in (cortex-a55): New.
20789 (cortex-a75): Likewise.
20790 (cortex-a75.cortex-a55): Likewise.
20791 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
20793 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
20794 * config/arm/arm-cpu-cdata.h: Regenerate.
20795 * config/arm/arm-cpu-data.h: Regenerate.
20796 * config/arm/arm-cpu.h: Regenerate.
20797 * config/arm/arm-tables.opt: Regenerate.
20798 * config/arm/arm-tune.md: Regenerate.
20800 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
20802 * haifa-sched.c (sched_create_recovery_edges): Update profile.
20804 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
20806 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
20809 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
20811 PR tree-optimization/81292
20812 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
20813 full_string_p, also call adjust_related_strinfos if the adjustment
20814 is simple, otherwise invalidate related strinfos.
20816 2017-07-04 Martin Liska <mliska@suse.cz>
20819 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
20820 newly created variable as DECL_IGNORED_P.
20822 2017-07-04 Martin Liska <mliska@suse.cz>
20825 * ipa-inline.c (inline_small_functions):
20826 Use xstrdup_for_dump.
20828 2017-07-04 Tom de Vries <tom@codesourcery.com>
20830 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
20832 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
20835 * config/darwin.c (darwin_function_switched_text_sections):
20838 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
20840 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
20842 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
20844 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
20846 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20848 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
20849 min_profitable_iters, and th as inclusive lower bounds.
20850 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
20851 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
20852 for min_profitable_iters and min_profitable_estimate.
20853 (vect_transform_loop): Treat th as an inclusive lower bound.
20854 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
20856 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
20859 * config/darwin.c (darwin_function_switched_text_sections):
20860 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
20861 in two pieces, and suppress the use of buf.
20863 2017-07-03 Nathan Sidwell <nathan@acm.org>
20865 * hash-table.h (hash_table_mod1): Fix indentation.
20867 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20869 PR middle-end/81290
20870 * predict.c (force_edge_cold): Be more careful about propagation
20872 * profile-count.h (profile_probability::guessed,
20873 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
20875 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
20877 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
20879 * doc/invoke.texi (rcpc architecture extension): Document it.
20881 2017-07-03 Richard Biener <rguenther@suse.de>
20883 PR tree-optimization/60510
20884 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
20885 the scalar reduction PHI and use it.
20886 (vectorizable_reduction): Properly guard the single_defuse_cycle
20887 path for non-SLP reduction chains where we cannot use it.
20888 Rework reduc_def/index and vector type deduction. Rework
20889 vector operand gathering during reduction op code-gen.
20890 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
20891 chains dissolve the chain and leave it to non-SLP reduction
20894 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20896 * tree-data-ref.h (dr_alignment): Declare.
20897 * tree-data-ref.c (dr_alignment): New function.
20898 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
20899 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
20901 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
20903 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20905 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
20906 and base_misalignment fields.
20907 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
20908 * tree-data-ref.c: Include builtins.h.
20909 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
20910 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
20911 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
20912 * tree-vect-data-refs.c: Include tree-cfg.h.
20913 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
20914 fields instead of calculating an alignment here.
20915 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
20916 innermost_loop_behavior fields.
20918 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20920 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
20922 (DR_STEP_ALIGNMENT): New macro.
20923 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
20924 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
20925 (create_data_ref): Print it.
20926 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
20927 to tell whether the step preserves vector (mis)alignment.
20928 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
20929 Move the check for an integer step and generalise to all INTEGER_CST.
20930 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
20931 Print the outer step alignment.
20933 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20935 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
20936 with offset_alignment.
20937 (DR_ALIGNED_TO): Delete.
20938 (DR_OFFSET_ALIGNMENT): New macro.
20939 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
20940 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
20941 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
20942 (create_data_ref): Likewise.
20943 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
20944 (vect_analyze_data_refs): Likewise.
20945 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
20946 creating dummy innermost behavior.
20948 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20950 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
20951 with a "innermost_loop_behavior *" and refeence tree.
20952 * tree-data-ref.c (dr_analyze_innermost): Likewise.
20953 (create_data_ref): Update call accordingly.
20954 * tree-predcom.c (find_looparound_phi): Likewise.
20956 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20958 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
20959 fields with dr_wrt_vec_loop.
20960 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
20961 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
20962 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
20963 (vect_dr_behavior): New function.
20964 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
20965 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
20966 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
20967 track whether the step preserves the misalignment.
20968 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
20969 Use vect_dr_behavior.
20970 (vect_setup_realignment): Update call accordingly.
20971 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
20972 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
20973 call to vect_create_addr_base_for_vector_ref.
20974 (vect_create_cond_for_align_checks): Likewise.
20975 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
20976 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
20977 (vect_recog_mask_conversion_pattern): Likewise.
20978 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
20979 (new_stmt_vec_info): Remove redundant zeroing.
20981 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
20983 * common/config/arm/arm-common.c (arm_be8_option): New function.
20984 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
20985 (ISA_ARMv6): Add isa_bit_be8.
20986 * config/arm/arm.h (arm_be8_option): Add prototype.
20987 (BE8_SPEC_FUNCTION): New define.
20988 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
20989 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
20990 (mlittle-endian): Similarly.
20991 (mbe8, mbe32): New options.
20992 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
20993 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
20995 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20997 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
20999 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21001 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
21002 (cleanup_tree_cfg_bb): Use it.
21003 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
21005 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
21007 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21010 * loop-doloop.c (add_test): Update profile.
21012 2017-07-03 Martin Liska <mliska@suse.cz>
21015 * sanopt.c (rewrite_usage_of_param): New function.
21016 (sanitize_rewrite_addressable_params): Likewise.
21017 (pass_sanopt::execute): Call rewrite_usage_of_param.
21019 2017-07-03 Richard Biener <rguenther@suse.de>
21021 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
21022 back to using VIEW_CONVERT_EXPR.
21024 2017-07-03 Martin Liska <mliska@suse.cz>
21027 * doc/extend.texi: Document when a resolver function is
21028 generated for target_clones.
21030 2017-07-03 Martin Liska <mliska@suse.cz>
21032 * asan.c (asan_emit_stack_protection): Unpoison just red zones
21033 and shadow memory of auto variables which are subject of
21034 use-after-scope sanitization.
21035 (asan_expand_mark_ifn): Add do set only when is_poison.
21037 2016-07-03 Richard Biener <rguenther@suse.de>
21039 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
21041 (vect_force_simple_reduction): Record reduction def -> phi mapping.
21042 (vectorizable_reduction): Perform reduction PHI creation when
21043 visiting a reduction PHI and adjust and simplify code generation
21044 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
21045 (vect_transform_loop): Visit reduction PHIs.
21046 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
21047 defs into the SLP tree.
21048 (vect_build_slp_tree): Reduction defs terminate the recursion.
21049 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
21051 (vect_get_vec_defs_for_stmt_copy): Export.
21052 (vect_get_vec_defs): Likewise.
21053 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
21055 (vect_get_vec_defs_for_stmt_copy): Declare.
21056 (vect_get_vec_defs): Likewise.
21058 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21060 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
21061 parameter with a "loop" parameter and use it instead of the
21062 loop containing DR_STMT. Don't check simple_iv when doing
21063 BB analysis. Describe the two analysis modes in the comment.
21065 2017-07-03 Tom de Vries <tom@codesourcery.com>
21067 PR tree-optimization/69468
21068 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
21069 (find_same_succ_bb): Handle ignore_edge_flags.
21071 2017-07-03 Tom de Vries <tom@codesourcery.com>
21073 PR tree-optimization/81192
21074 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
21076 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
21078 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
21080 2017-07-03 Tom de Vries <tom@codesourcery.com>
21082 PR tree-optimization/81192
21083 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
21084 BB_SAME_SUCC (bb) == NULL.
21086 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21088 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
21091 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21093 * dumpfile.c: Include profile-count.h
21094 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
21096 (insert_cond_bb): Update profile.
21097 * tree-cfg.h (insert_cond_bb): Update prototype.
21098 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
21099 * tree-dump.c: Do not include tree-cfg.
21101 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21103 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
21105 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21107 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
21110 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21112 * tree-complex.c (expand_complex_div_wide): update profile.
21114 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21115 Alan Hayward <alan.hayward@arm.com>
21116 David Sherwood <david.sherwood@arm.com>
21118 * Makefile.in (MACHMODE_H): Remove insn-modes.h
21119 (CORETYPES_H): New define.
21120 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
21121 (insn-modes-inline.h, s-modes-inline-h): New rules.
21122 (generated_files): Add insn-modes-inline.h.
21123 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
21124 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
21125 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
21126 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
21127 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
21128 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
21129 (build/gencodes.o, build/genconditions.o): Likewise.
21130 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
21131 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
21132 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
21133 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
21134 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
21135 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
21136 * coretypes.h: Include everything up to real.h for generators.
21137 Include insn-modes.h first. Include wide-int-print.h after
21138 wide-int.h. Include insn-modes-inline.h and then machmode.h.
21139 * machmode.h: Don't include insn-modes.h here.
21140 * function-tests.c: Remove includes of signop.h, machmode.h,
21141 double-int.h and wide-int.h.
21143 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
21145 * optc-save-gen.awk: Likewise.
21146 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
21147 * godump.c: Remove include of wide-int-print.h.
21148 * pretty-print.h: Likewise.
21149 * wide-int-print.cc: Likewise.
21150 * wide-int.cc: Likewise.
21151 * hash-map-tests.c: Remove include of signop.h.
21152 * hash-set-tests.c: Likewise.
21153 * rtl-tests.c: Likewise.
21154 * mkconfig.sh: Remove include of machmode.h.
21155 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
21157 (emit_insn_modes_inline_h): ...this new function. Emit the code
21158 into an insn-modes-inline.h header file, adding appropriate
21159 include guards and end comments.
21160 (emit_insn_modes_c_header): Remove include of machmode.h.
21161 (emit_min_insn_modes_c_header): Include coretypes.h rather than
21163 (main): Handle -i flag and call emit_insn_modes_inline_h when
21166 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21168 * tree-ssa-strlen.c (strinfo): Rename the length field to
21169 nonzero_chars. Add a full_string_p field.
21170 (compare_nonzero_chars, zero_length_string_p): New functions.
21171 (get_addr_stridx): Add an offset_out parameter.
21172 Use compare_nonzero_chars.
21173 (get_stridx): Update accordingly. Use compare_nonzero_chars.
21174 (new_strinfo): Update after above changes to strinfo.
21175 (set_endptr_and_length): Set full_string_p.
21176 (get_string_length): Update after above changes to strinfo.
21177 (unshare_strinfo): Update call to new_strinfo.
21178 (maybe_invalidate): Likewise.
21179 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
21180 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
21181 as a uhwi instead of an shwi. Update after above changes to
21182 strinfo and new_strinfo.
21183 (zero_length_string): Assert that chainsi contains full strings.
21184 Use zero_length_string_p. Update call to new_strinfo.
21185 (adjust_related_strinfos): Update after above changes to strinfo.
21186 Copy full_string_p from origsi.
21187 (adjust_last_stmt): Use zero_length_string_p.
21188 (handle_builtin_strlen): Update after above changes to strinfo and
21189 new_strinfo. Install the lhs as the string length if the previous
21190 entry didn't describe a full string.
21191 (handle_builtin_strchr): Update after above changes to strinfo
21193 (handle_builtin_strcpy): Likewise.
21194 (handle_builtin_strcat): Likewise.
21195 (handle_builtin_malloc): Likewise.
21196 (handle_pointer_plus): Likewise.
21197 (handle_builtin_memcpy): Likewise. Track nonzero characters
21198 that aren't necessarily followed by a nul terminator.
21199 (handle_char_store): Likewise.
21201 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21203 PR tree-optimization/80769
21204 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
21205 for malloc and calloc. Document the new invariant that all related
21206 strinfos have delayed lengths or none do.
21207 (verify_related_strinfos): Move earlier in file.
21208 (set_endptr_and_length): New function, split out from...
21209 (get_string_length): ...here. Also set the lengths of related
21211 (zero_length_string): Assert that chainsi has known (rather than
21213 (adjust_related_strinfos): Likewise.
21215 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21217 PR tree-optimization/81136
21218 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
21219 assert that two references with the same misalignment have the same
21220 compile-time misalignment if those compile-time misalignments
21223 2017-07-01 Andi Kleen <ak@linux.intel.com>
21225 * print-tree.c (print_node): Print all attributes.
21227 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21229 * cfg.c (scale_bbs_frequencies): New function.
21230 * cfg.h (scale_bbs_frequencies): Declare it.
21231 * cfgloopanal.c (single_likely_exit): Cleanup.
21232 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
21234 (scale_loop_profile): Likewise.
21235 (loop_version): Likewise.
21236 (create_empty_loop_on_edge): Update.
21237 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
21238 scale_loop_frequencies, scale_loop_profile, loopify,
21239 loop_version): Update prototypes.
21240 * modulo-sched.c (sms_schedule): Update.
21241 * predict.c (unlikely_executed_edge_p): Also check probability.
21242 (probably_never_executed_edge_p): Fix typo.
21243 * tree-if-conv.c (version_loop_for_if_conversion): Update.
21244 * tree-parloops.c (gen_parallel_loop): Update.
21245 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
21246 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21247 * tree-ssa-loop-split.c (split_loop): Update.
21248 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
21249 * tree-vect-loop-manip.c (vect_do_peeling): Update.
21250 (vect_loop_versioning): Update.
21251 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21253 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21255 * trans-mem.c (split_bb_make_tm_edge): Update profile.
21257 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21259 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
21260 to keep profile consistent.
21262 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21264 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
21265 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
21266 * profile-count.h (max_safe_multiplier): Make unsigned.
21267 (profile_count::guessed_zero): New.
21269 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21271 * bb-reorder.c (fix_up_crossing_landing_pad,
21272 fix_crossing_conditional_branches): Use make_single_succ_edge
21273 to keep profile consistent.
21275 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21277 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
21280 2017-07-01 Jakub Jelinek <jakub@redhat.com>
21283 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
21284 the right scopes, make sure cond_jump isn't preserved between multiple
21285 iterations. Search for fallthru edge whenever there are 3+ edges and
21286 use find_fallthru_edge for it.
21288 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21290 Patch by Alexander Monakov <amonakov@ispras.ru>
21291 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
21292 probabilities consistently.
21294 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21296 * pa.c (pa_expand_compare_and_swap_loop): Update call of
21297 emit_cmp_and_jump_insns.
21299 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21302 * tree-inline.c (expand_call_inline): Combine profile statuses.
21304 2017-06-30 Andrew Pinski <apinski@cavium.com>
21306 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
21307 fold_stmt returned true.
21309 2017-06-30 Nathan Sidwell <nathan@acm.org>
21311 * ggc.h (empty_string): Delete.
21312 * cfgexpand.c (expand_asm_stmt): Use plain "".
21313 * optabs.c (expand_asm_memory_barrier): Likewise.
21314 * stringpool.c (empty_string): Delete.
21315 (digit_vector, digit_string): Delete.
21316 (ggc_alloc_string): Use plain "", don't optimize single digit
21317 strings. Use ggc_alloc_atomic.
21319 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
21321 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
21322 comparison set and one other set, use the cost of the non-comparison
21325 2017-06-30 Nathan Sidwell <nathan@acm.org>
21327 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
21330 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
21332 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
21333 loops. Remove now unneeded calls to gimple_switch_set_label() that
21334 just set removed labels to NULL_TREE.
21336 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
21338 * tree-ssanames.c (set_range_info_raw): Abstract from ...
21339 (set_range_info): ...here. Only call set_range_info_raw if domain
21341 (set_nonzero_bits): Call set_range_info_raw.
21342 * tree-ssanames.h (set_range_info_raw): New.
21344 2017-06-30 Jakub Jelinek <jakub@redhat.com>
21347 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
21348 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
21349 of nonimmediate_operand and <store_mask_constraint> instead of m
21350 for the input operand. For V8FI iterator, always split if input
21351 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
21352 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
21353 <store_mask_predicate> instead of register_operand and
21354 <store_mask_constraint> instead of v for the input operand. Make
21355 sure both operands aren't MEMs for if not <mask_applied>.
21357 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
21359 * lto-wrapper.c (copy_file) Close both file descriptors before
21362 2017-06-30 Martin Liska <mliska@suse.cz>
21365 * multiple_target.c (create_dispatcher_calls): Make ifunc
21366 also for function that don't have calls or are not referenced.
21368 2017-06-30 Richard Biener <rguenther@suse.de>
21370 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
21371 analyze the first scalar stmt. Move vector type computation
21372 for the BB case here from ...
21373 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
21374 live operation processing in the SLP case properly.
21376 2017-06-30 Richard Biener <rguenther@suse.de>
21378 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
21380 2017-06-30 Martin Liska <mliska@suse.cz>
21383 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
21384 before BUILT_IN_UNWIND_RESUME when ASAN is used.
21386 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
21388 * doc/invoke.texi (AArch64): Add missing options and remove redundant
21391 2017-06-30 Richard Biener <rguenther@suse.de>
21393 PR tree-optimization/81249
21394 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
21395 condition reduction result to original scalar type.
21397 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21399 * profile-count.h (enum profile_quality): Fix typos and whitespace
21402 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21404 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
21405 type for branch probabilities.
21407 2017-06-29 Julian Brown <julian@codesourcery.com>
21408 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21410 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
21411 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
21412 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
21413 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
21415 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21417 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
21418 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
21419 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
21420 CC usage from generic code to here.
21421 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
21422 CC usage into the target macros.
21424 2017-06-29 Maya Rashish <coypu@sdf.org>
21426 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
21429 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21431 * arm/arm-builtins.c: Include profile-count.h
21432 * except.c (sjlj_emit_function_enter): Use
21433 profile_probability::unlikely.
21435 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21437 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
21438 and tocrel_offset be pointer args rather than implicitly using
21440 (legitimate_constant_pool_address_p, rs6000_emit_move,
21441 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
21442 tocrel_offset and use in toc_relative_expr_p call.
21443 (print_operand, print_operand_address): Use static tocrel_base_oac
21444 and tocrel_offset_oac.
21445 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
21448 2017-06-29 Maya Rashish <coypu@sdf.org>
21450 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
21452 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
21454 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
21455 objects, take into account only the alignment of 'op0' and 'mode1' if
21458 2017-06-29 Steve Ellcey <sellcey@cavium.com>
21460 * ccmp.c (ccmp_tree_comparison_p): New function.
21461 (ccmp_candidate_p): Update to use above function.
21462 (get_compare_parts): New function.
21463 (expand_ccmp_next): Update to use new functions.
21464 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
21466 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
21467 take mode as argument.
21468 * ccmp.h (expand_ccmp_expr): Add mode as argument.
21469 * expr.c (expand_expr_real_1): Pass mode as argument.
21471 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
21473 * combine.c (combine_instructions): Print insns to dump_file, together
21476 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21478 * asan.c (asan_emit_stack_protection): Update.
21479 (create_cond_insert_point): Update.
21480 * auto-profile.c (afdo_propagate_circuit): Update.
21481 * basic-block.h (struct edge_def): Turn probability to
21482 profile_probability.
21483 (EDGE_FREQUENCY): Update.
21484 * bb-reorder.c (find_traces_1_round): Update.
21485 (better_edge_p): Update.
21486 (sanitize_hot_paths): Update.
21487 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
21488 (make_single_succ_edge): Update.
21489 (check_bb_profile): Update.
21490 (dump_edge_info): Update.
21491 (update_bb_profile_for_threading): Update.
21492 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
21493 probabilitycount to 0.
21494 * cfgbuild.c (compute_outgoing_frequencies): Update.
21495 * cfgcleanup.c (try_forward_edges): Update.
21496 (outgoing_edges_match): Update.
21497 (try_crossjump_to_edge): Update.
21498 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
21499 (expand_gimple_tailcall): Update.
21500 (construct_init_block): Use make_single_succ_edge.
21501 (construct_exit_block): Use make_single_succ_edge.
21502 * cfghooks.c (verify_flow_info): Update.
21503 (redirect_edge_succ_nodup): Update.
21504 (split_edge): Update.
21505 (account_profile_record): Update.
21506 * cfgloopanal.c (single_likely_exit): Update.
21507 * cfgloopmanip.c (scale_loop_profile): Update.
21508 (set_zero_probability): Remove.
21509 (duplicate_loop_to_header_edge): Update.
21510 * cfgloopmanip.h (loop_version): Update prototype.
21511 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
21512 (force_nonfallthru_and_redirect): Update.
21513 (update_br_prob_note): Update.
21514 (rtl_verify_edges): Update.
21515 (purge_dead_edges): Update.
21516 (rtl_lv_add_condition_to_bb): Update.
21517 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
21518 * cgraphunit.c (init_lowered_empty_function): Update.
21519 (cgraph_node::expand_thunk): Update.
21520 * cilk-common.c: Include profile-count.h
21521 * dojump.c (inv): Remove.
21522 (jumpifnot): Update.
21523 (jumpifnot_1): Update.
21524 (do_jump_1): Update.
21526 (do_jump_by_parts_greater_rtx): Update.
21527 (do_compare_rtx_and_jump): Update.
21528 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
21529 do_jump_1. do_compare_rtx_and_jump): Update prototype.
21530 * dwarf2cfi.c: Include profile-count.h
21531 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
21532 (sjlj_emit_dispatch_table): Likewise.
21533 * explow.c: Include profile-count.h
21534 * expmed.c (emit_store_flag_force): Update.
21535 (do_cmp_and_jump): Update.
21536 * expr.c (compare_by_pieces_d::generate): Update.
21537 (compare_by_pieces_d::finish_mode): Update.
21538 (emit_block_move_via_loop): Update.
21539 (store_expr_with_bounds): Update.
21540 (store_constructor): Update.
21541 (expand_expr_real_2): Update.
21542 (expand_expr_real_1): Update.
21543 * expr.h (try_casesi, try_tablejump): Update prototypes.
21544 * gimple-pretty-print.c (dump_probability): Update.
21545 (dump_profile): New.
21546 (dump_gimple_label): Update.
21547 (dump_gimple_bb_header): Update.
21548 * graph.c (draw_cfg_node_succ_edges): Update.
21549 * hsa-gen.c (convert_switch_statements): Update.
21550 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
21551 (find_if_case_1): Update.
21552 (find_if_case_2): Update.
21553 * internal-fn.c (expand_arith_overflow_result_store): Update.
21554 (expand_addsub_overflow): Update.
21555 (expand_neg_overflow): Update.
21556 (expand_mul_overflow): Update.
21557 (expand_vector_ubsan_overflow): Update.
21558 * ipa-cp.c (good_cloning_opportunity_p): Update.
21559 * ipa-split.c (split_function): Use make_single_succ_edge.
21560 * ipa-utils.c (ipa_merge_profiles): Update.
21561 * loop-doloop.c (add_test): Update.
21562 (doloop_modify): Update.
21563 * loop-unroll.c (compare_and_jump_seq): Update.
21564 (unroll_loop_runtime_iterations): Update.
21565 * lra-constraints.c (lra_inheritance): Update.
21566 * lto-streamer-in.c (input_cfg): Update.
21567 * lto-streamer-out.c (output_cfg): Update.
21568 * mcf.c (adjust_cfg_counts): Update.
21569 * modulo-sched.c (sms_schedule): Update.
21570 * omp-expand.c (expand_omp_for_init_counts): Update.
21571 (extract_omp_for_update_vars): Update.
21572 (expand_omp_ordered_sink): Update.
21573 (expand_omp_for_ordered_loops): Update.
21574 (expand_omp_for_generic): Update.
21575 (expand_omp_for_static_nochunk): Update.
21576 (expand_omp_for_static_chunk): Update.
21577 (expand_cilk_for): Update.
21578 (expand_omp_simd): Update.
21579 (expand_omp_taskloop_for_outer): Update.
21580 (expand_omp_taskloop_for_inner): Update.
21581 * omp-simd-clone.c (simd_clone_adjust): Update.
21582 * optabs.c (expand_doubleword_shift): Update.
21583 (expand_abs): Update.
21584 (emit_cmp_and_jump_insn_1): Update.
21585 (expand_compare_and_swap_loop): Update.
21586 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
21587 * predict.c (predictable_edge_p): Update.
21588 (edge_probability_reliable_p): Update.
21589 (set_even_probabilities): Update.
21590 (combine_predictions_for_insn): Update.
21591 (combine_predictions_for_bb): Update.
21592 (propagate_freq): Update.
21593 (estimate_bb_frequencies): Update.
21594 (force_edge_cold): Update.
21595 * profile-count.c (profile_count::dump): Add missing space into dump.
21596 (profile_count::debug): Add newline.
21597 (profile_count::differs_from_p): Explicitly convert to unsigned.
21598 (profile_count::stream_in): Update.
21599 (profile_probability::dump): New member function.
21600 (profile_probability::debug): New member function.
21601 (profile_probability::differs_from_p): New member function.
21602 (profile_probability::differs_lot_from_p): New member function.
21603 (profile_probability::stream_in): New member function.
21604 (profile_probability::stream_out): New member function.
21605 * profile-count.h (profile_count_quality): Rename to ...
21606 (profile_quality): ... this one.
21607 (profile_probability): New.
21608 (profile_count): Update.
21609 * profile.c (compute_branch_probabilities): Update.
21610 * recog.c (peep2_attempt): Update.
21611 * sched-ebb.c (schedule_ebbs): Update.
21612 * sched-rgn.c (find_single_block_region): Update.
21613 (compute_dom_prob_ps): Update.
21614 (schedule_region): Update.
21615 * sel-sched-ir.c (compute_succs_info): Update.
21616 * stmt.c (struct case_node): Update.
21617 (do_jump_if_equal): Update.
21618 (get_outgoing_edge_probs): Update.
21619 (conditional_probability): Update.
21620 (emit_case_dispatch_table): Update.
21621 (expand_case): Update.
21622 (expand_sjlj_dispatch_table): Update.
21623 (emit_case_nodes): Update.
21624 * targhooks.c: Update.
21625 * tracer.c (better_p): Update.
21626 (find_best_successor): Update.
21627 * trans-mem.c (expand_transaction): Update.
21628 * tree-call-cdce.c: Update.
21629 * tree-cfg.c (gimple_split_edge): Upate.
21630 (move_sese_region_to_fn): Upate.
21631 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
21632 * tree-eh.c (lower_resx): Upate.
21633 (cleanup_empty_eh_move_lp): Upate.
21634 * tree-if-conv.c (version_loop_for_if_conversion): Update.
21635 * tree-inline.c (copy_edges_for_bb): Update.
21636 (copy_cfg_body): Update.
21637 * tree-parloops.c (gen_parallel_loop): Update.
21638 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
21639 (gimple_gen_time_profiler): Update.
21640 * tree-ssa-dce.c (remove_dead_stmt): Update.
21641 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21642 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
21643 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
21644 (unloop_loops): Update.
21645 (try_peel_loop): Update.
21646 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21647 * tree-ssa-loop-split.c (connect_loops): Update.
21648 (split_loop): Update.
21649 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
21650 (hoist_guard): Update.
21651 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
21652 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
21653 (value_replacement): Update.
21654 * tree-ssa-reassoc.c (branch_fixup): Update.
21655 * tree-ssa-tail-merge.c (replace_block_by): Update.
21656 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
21657 (create_edge_and_update_destination_phis): Update.
21658 (compute_path_counts): Update.
21659 (recompute_probabilities): Update.
21660 (update_joiner_offpath_counts): Update.
21661 (freqs_to_counts_path): Update.
21662 (duplicate_thread_path): Update.
21663 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
21664 (struct switch_conv_info): Update.
21665 (gen_inbound_check): Update.
21666 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21667 (vect_do_peeling): Update.
21668 (vect_loop_versioning): Update.
21669 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21670 (optimize_mask_stores): Update.
21671 * ubsan.c (ubsan_expand_null_ifn): Update.
21672 * value-prof.c (gimple_divmod_fixed_value): Update.
21673 (gimple_divmod_fixed_value_transform): Update.
21674 (gimple_mod_pow2): Update.
21675 (gimple_mod_pow2_value_transform): Update.
21676 (gimple_mod_subtract): Update.
21677 (gimple_mod_subtract_transform): Update.
21678 (gimple_ic): Update.
21679 (gimple_stringop_fixed_value): Update.
21680 (gimple_stringops_transform): Update.
21681 * value-prof.h: Update.
21683 2017-06-29 Carl Love <cel@us.ibm.com>
21685 * config/rs6000/rs6000-c.c: Add support for built-in functions
21686 vector signed int vec_signed (vector float);
21687 vector signed long long vec_signed (vector double);
21688 vector signed int vec_signed2 (vector double, vector double);
21689 vector signed int vec_signede (vector double);
21690 vector signed int vec_signedo (vector double);
21691 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
21692 instruction generator.
21693 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
21694 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
21695 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
21697 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
21698 vunsignede_v2df): Add define_expands.
21699 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
21700 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
21701 VEC_UNSIGNEDO): Add definitions.
21702 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
21703 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
21704 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
21705 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
21706 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
21707 * config/rs6000/altivec.h (vec_signed, vec_signed2,
21708 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
21709 vec_unsignede, vec_unsignedo): Add builtin defines.
21710 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
21712 * doc/extend.texi: Update the built-in documentation file for the
21713 new built-in functions.
21715 2017-06-29 Richard Biener <rguenther@suse.de>
21717 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
21718 reduction chains to LOOP_VINFO_REDUCTIONS.
21719 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
21720 SLP reductions after processing reduction chains.
21722 2017-06-29 Nathan Sidwell <nathan@acm.org>
21724 * builtins.c (fold_builtin_FUNCTION): Use
21725 lang_hooks.decl_printable_name.
21727 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
21729 PR middle-end/81194
21730 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
21731 with only one label.
21732 * stmt.c (expand_case): Assert NCASES is greater than one.
21734 2017-06-29 Richard Biener <rguenther@suse.de>
21736 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
21738 (group_case_labels): Likewise.
21739 (find_taken_edge): Push sanity checking on val to workers...
21740 (find_taken_edge_cond_expr): ... here
21741 (find_taken_edge_switch_expr): ... and here, handle cases
21742 with just a default label.
21743 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
21744 (group_case_labels): Likewise.
21745 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
21746 group_case_labels does anything cleanup the CFG again.
21748 2017-06-29 Bin Cheng <bin.cheng@arm.com>
21750 PR tree-optimization/81196
21751 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
21752 exit condition comparing two IVs.
21754 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
21756 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
21757 profile to the dummy entry at the end of the list of architectures.
21758 * config/arm/arm-cpu-cdata.h: Regenerated.
21760 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21761 Michael Collison <michael.collison@arm.com>
21764 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
21766 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
21767 (*aarch64_reg_<mode>3_minus_mask): New pattern.
21768 (*aarch64_<optab>_reg_di3_mask2): New pattern.
21769 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
21770 of shift when the shift amount is masked with constant equal to
21771 the size of the mode.
21772 * config/aarch64/predicates.md (subreg_lowpart_operator): New
21775 2017-06-29 Martin Liska <mliska@suse.cz>
21777 * config/i386/i386.opt: Change range from [1,5] to [0,5].
21779 2017-06-29 Yury Gribov <tetra2005@gmail.com>
21782 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
21785 (edge_growth_cache_entry::edge_growth_cache_entry): New
21787 (reset_edge_growth_cache): Update to use constructor.
21789 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21791 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
21792 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
21793 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
21795 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
21797 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
21798 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
21800 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
21802 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
21803 (*-linux-uclibc*): Add t-uclibc tmake_file.
21804 * config/t-musl: New.
21805 * config/t-uclibc: New.
21807 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
21809 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
21811 (gen_comm_data): Emit architectural setting of arch_prof.
21812 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
21814 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
21815 (armv8-m.base, armv8-m.main): Likewise.
21816 * arm-protos.h (arm_build_target): Add profile field.
21817 (arch_option): Likewise.
21818 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
21820 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
21821 arm_active_target.profile.
21823 2017-06-28 Richard Biener <rguenther@suse.de>
21825 PR middle-end/81227
21826 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
21827 TYPE_OVERFLOW_WRAPS.
21828 * match.pd (negate_expr_p): Likewise.
21829 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
21830 fold_build2, not fold_binary.
21832 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
21834 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
21835 Convert memory address to Pmode.
21836 (aarch64_print_operand): Assert MEM operands are always Pmode.
21838 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
21841 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
21842 Remove redundant if.
21843 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
21844 * config/arm/aarch-common-protos.h
21845 (aarch_forward_to_shift_is_not_shifted_re): Remove.
21846 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
21848 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
21851 * multiple_target.c (create_dispatcher_calls): Set the default
21852 clone to be static, not public.
21854 2017-06-28 Richard Biener <rguenther@suse.de>
21856 * tree-vect-loop.c (vectorizable_reduction): Move special
21857 cond reduction IV var creation ...
21858 (vect_create_epilog_for_reduction): ... here. Remove induction_index
21859 parameter. Use STMT_VINFO_VECTYPE.
21860 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
21863 2017-06-28 Martin Liska <mliska@suse.cz>
21866 * ipa-visibility.c (non_local_p): Handle visibility.
21868 2017-06-28 Martin Liska <mliska@suse.cz>
21871 * common.opt: Add IntegerRange to various options.
21872 * opt-functions.awk (integer_range_info): New function.
21873 * optc-gen.awk: Add integer_range_info to cl_options struct.
21874 * opts-common.c (decode_cmdline_option): Handle
21875 CL_ERR_INT_RANGE_ARG.
21876 (cmdline_handle_error): Likewise.
21877 * opts.c (print_filtered_help): Show valid interval in
21878 when --help is provided.
21879 * opts.h (struct cl_option): Add range_min and range_max fields.
21880 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
21882 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
21884 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
21885 (x * C EQ/NE y * C): New transformation.
21887 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
21889 * genmultilib (combination_space): Accept '+' in option names.
21891 2017-06-28 Martin Liska <mliska@suse.cz>
21894 * asan.c (instrument_derefs): Bail out inner references
21895 that are hard register variables.
21897 2017-06-28 Jakub Jelinek <jakub@redhat.com>
21900 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
21901 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
21903 2017-06-28 Richard Biener <rguenther@suse.de>
21905 * tree-vectorizer.h (vect_get_vec_defs): Remove.
21906 (vect_get_slp_defs): Adjust.
21907 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
21909 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
21911 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
21912 get_initial_defs_for_reduction instead of vect_get_vec_defs.
21913 (vectorizable_reduction): Adjust.
21914 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
21916 (vect_get_slp_defs): Likewise.
21917 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
21918 (vectorizable_bswap): Adjust.
21919 (vectorizable_call): Likewise.
21920 (vectorizable_conversion): Likewise.
21921 (vectorizable_assignment): Likewise.
21922 (vectorizable_shift): Likewise.
21923 (vectorizable_operation): Likewise.
21924 (vectorizable_store): Likewise.
21925 (vectorizable_condition): Likewise.
21926 (vectorizable_comparison): Likewise.
21928 2017-06-28 Michael Collison <michael.collison@arm.com>
21931 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
21933 (arm_gen_movmemqi): Removed unused variable 'i'.
21934 Convert 'for' loop into 'while' loop.
21935 (arm_expand_prologue): Remove last unnecessary set of insn.
21936 (thumb_pop): Remove unused variable 'pushed_words'.
21937 (thumb_exit): Remove last unnecessary set of regs_to_pop.
21939 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21941 * config/s390/predicates.md: Use s390_rel_address_ok_p.
21942 * config/s390/s390-protos.h: Add prototype of
21943 s390_rel_address_ok_p.
21944 * config/s390/s390.c (s390_got_symbol): New function.
21945 (s390_rel_address_ok_p): New function.
21946 (legitimize_pic_address): Use s390_rel_address_ok_p.
21947 (s390_load_got): Use s390_got_symbol.
21948 (s390_option_override): Issue error if
21949 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
21950 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
21952 * config/s390/s390.opt: New option mpic-data-is-text-relative.
21954 2017-06-27 Andrew Pinski <apinski@cavium.com>
21956 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
21957 (X * copysign (1.0, X)): New pattern.
21958 (X * copysign (1.0, -X)): New pattern.
21959 (copysign (-1.0, CST)): New pattern.
21961 2017-06-27 Joseph Myers <joseph@codesourcery.com>
21963 * genmultilib (combination_space): Remove variable.
21964 Validate reuse rules against regular expression for any sequence
21965 of multilib options in any order.
21967 2017-06-27 Michael Collison <michael.collison@arm.com>
21969 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
21970 call aarch64_split_simd_combine.
21971 * (aarch64_combine_internal<mode>): Delete pattern.
21972 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
21973 Allow register and subreg operands.
21975 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21977 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
21978 specific need, just fallback on defaults.
21979 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
21981 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21982 Olivier Hainque <hainque@adacore.com>
21984 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
21986 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
21987 targets. Pick a default if no particular attempt applied.
21988 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
21991 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
21993 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
21994 (x86_64-wrs-vxworks7): Likewise.
21996 2017-06-27 Marek Polacek <polacek@redhat.com>
21999 * ubsan.c (instrument_null): Check get_base_address's result for null.
22001 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
22003 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
22005 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
22007 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
22008 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
22009 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
22010 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
22011 New function types.
22012 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
22013 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
22014 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
22015 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
22016 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
22017 BUILT_IN_FEUPDATEENV): New builtins.
22018 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
22019 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
22020 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
22021 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
22023 (builtin_structptr_types): Adjust size.
22024 * tree.c (builtin_structptr_types): Add four entries.
22026 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22027 Olivier Hainque <hainque@adacore.com>
22029 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
22030 (TLS_SYM): New local macro, forcing reference to __tls__ on
22031 link command lines for VxWorks 7 RTPs, triggering initialization
22033 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
22034 OS features TLS support, true for RTPs on VxWorks 7.
22035 * config/vxworks.c (vxworks_override_options): Setup emutls
22038 2017-06-27 Jakub Jelinek <jakub@redhat.com>
22040 * predict.c (test_prediction_value_range): Use -1U instead of -1
22041 to avoid narrowing conversion warning.
22042 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
22043 to avoid narrowing conversion warning.
22044 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
22046 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
22048 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22050 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
22051 64bit configurations.
22052 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
22053 (SIZE_TYPE): Likewise.
22054 * config/vxworks.c (vxworks_emutls_var_fields): Use
22055 long_unsigned_type_node instead of unsigned_type_node as the offset
22056 field type, which is "pointer" mode in emutls.c.
22058 2017-06-27 Jakub Jelinek <jakub@redhat.com>
22061 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
22063 PR middle-end/81207
22064 * gimple-fold.c (replace_call_with_call_and_fold): Handle
22065 gimple_vuse copying separately from gimple_vdef copying.
22067 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22069 * value-prof.c (free_hist): Remove call to memset and the enclosing if
22072 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
22073 Olivier Hainque <hainque@adacore.com>
22075 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
22076 for all vxworks7 targets.
22077 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
22078 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
22079 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
22080 variations for VX6/VX7 and 32/64bits later on in ...
22081 (VXWORKS_LIB_SPEC): Leverage new macros.
22082 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
22083 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
22085 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
22087 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
22088 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
22090 2017-06-26 Carl Love <cel@us.ibm.com>
22092 * config/rs6000/rs6000-c.c: Add support for built-in functions
22093 vector bool char vec_reve (vector bool char);
22094 vector signed char vec_reve (vector signed char);
22095 vector unsigned char vec_reve (vector unsigned char);
22096 vector bool int vec_reve (vector bool int);
22097 vector signed int vec_reve (vector signed int);
22098 vector unsigned int vec_reve (vector unsigned int);
22099 vector bool long long vec_reve (vector bool long long);
22100 vector signed long long vec_reve (vector signed long long);
22101 vector unsigned long long vec_reve (vector unsigned long long);
22102 vector bool short vec_reve (vector bool short);
22103 vector signed short vec_reve (vector signed short);
22104 vector double vec_reve (vector double);
22105 vector float vec_reve (vector float);
22106 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
22107 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
22108 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
22109 (altivec_vreve): New pattern.
22110 * config/rs6000/altivec.h (vec_reve): New define.
22111 * doc/extend.texi (vec_rev): Update the built-in documentation file
22112 for the new built-in functions.
22114 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22116 PR tree-optimization/71815
22117 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
22119 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
22121 (slsr_process_phi): Likewise.
22122 (replace_uncond_cands_and_profitable_phis): Don't replace a
22123 multiply candidate with a stride of 1 (copy or cast).
22124 (phi_incr_cost): Call uses_consumed_by_stmt rather than
22126 (lowest_cost_path): Likewise.
22127 (total_savings): Likewise.
22129 2017-06-26 Richard Biener <rguenther@suse.de>
22132 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
22133 Use def_builtin_pure for all gather builtins.
22135 2017-06-26 Richard Biener <rguenther@suse.de>
22137 PR tree-optimization/81203
22138 * tree-tailcall.c (find_tail_calls): Do not move stmts into
22139 non-dominating BBs.
22141 2017-06-26 Marek Polacek <polacek@redhat.com>
22144 * doc/invoke.texi: Document -Wmultistatement-macros.
22146 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
22148 * doc/sourcebuild.texi (ARM-specific attributes): Document new
22149 arm_neon_ok_no_float_abi effective target.
22151 2017-06-26 Richard Biener <rguenther@suse.de>
22153 PR tree-optimization/80928
22154 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
22155 (copy_bbs): Set BB_DUPLICATED flag early.
22156 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
22158 (execute_on_shrinking_pred): Likewise.
22159 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
22160 BB_DUPLICATED blocks.
22161 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
22162 iterate over all PHIs considering removal of *gsi.
22164 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
22166 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
22169 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22171 * config/rs6000/rs6000-string.c: (expand_block_clear,
22172 do_load_for_compare, select_block_compare_mode,
22173 compute_current_alignment, expand_block_compare,
22174 expand_strncmp_align_check, expand_strn_compare,
22175 expand_block_move, rs6000_output_load_multiple)
22176 Move functions related to string/block move/compare
22177 to a separate file.
22178 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
22179 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
22180 for this function which is now used in two files.
22181 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
22182 * config.gcc: Add rs6000-string.o to extra_objs for
22183 targets powerpc*-*-* and rs6000*-*-*.
22185 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
22188 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
22189 32-bit, since indexed is not valid for DImode.
22190 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
22191 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
22192 (define_peephole2 for Altivec d-form load): Add 32-bit support.
22193 (define_peephole2 for Altivec d-form store): Likewise.
22196 * multiple_target.c (create_dispatcher_calls): Only create the
22197 dispatcher call if the function is the default clone of a
22198 versioned function.
22200 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
22202 PR middle-end/80902
22203 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
22204 a call, force the call to not be a tail call.
22206 2017-06-23 Jeff Law <law@redhat.com>
22208 * doc/contrib.texi: Add entry for Steven Pemberton's work on
22211 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
22213 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
22214 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
22215 handling for early expansion of vector shifts (sl,sr,sra,rl).
22216 (builtin_function_type): Add vector shift right instructions
22217 to the unsigned argument list.
22219 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
22221 rtl-optimizatoin/79286
22222 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
22223 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
22224 trap. PIC register plus a const unspec without offset can never trap.
22226 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
22228 * tree.h (builtin_structptr_type): New type.
22229 (builtin_structptr_types): Declare new array.
22230 * tree.c (builtin_structptr_types): New array.
22231 (free_lang_data, build_common_tree_nodes): Use it.
22233 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
22236 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
22239 2017-06-22 Matt Turner <mattst88@gmail.com>
22241 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
22242 Lake models to skylake case. Assume skylake for unknown
22243 models with clflushopt.
22245 2017-06-22 Jeff Law <law@redhat.com>
22247 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
22248 frame sizes that do not satisfy aarch64_uimm12_shift.
22250 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
22252 * profile-count.h (apply_probability,
22253 apply_scale, probability_in): Fix checks for zero.
22255 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22257 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
22258 * doc/cppdiropts.texi (-I @var{dir}): Document it.
22260 2016-06-22 Richard Biener <rguenther@suse.de>
22262 * tree-vect-loop.c (vect_model_reduction_cost): Handle
22263 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
22264 REDUC_MAX_EXPR support.
22265 (vectorizable_reduction): Likewise.
22266 (vect_create_epilog_for_reduction): Likewise.
22268 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
22270 * match.pd (A / (1 << B) -> A >> B): New.
22271 * generic-match-head.c: Include optabs-tree.h.
22272 * gimple-match-head.c: Likewise.
22273 * optabs-tree.h (target_supports_op_p): New.
22274 * optabs-tree.c (target_supports_op_p): New.
22276 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22278 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
22279 $gcc_cv_ld --help output.
22280 (gcc_cv_ld_demangle): Likewise.
22281 (gcc_cv_ld_eh_frame_hdr): Likewise.
22282 (gcc_cv_ld_pie): Likewise.
22283 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
22284 (gcc_cv_ld_buildid): Likewise.
22285 (gcc_cv_ld_sysroot): Likewise.
22286 (ld_bndplt_support): Likewise.
22287 (ld_pushpopstate_support): Likewise.
22288 * configure: Regenerate.
22289 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
22291 2017-06-21 Jakub Jelinek <jakub@redhat.com>
22294 * config/i386/sse.md (round<mode>2): Renumber match_dup and
22295 operands indexes to avoid gap between operands and match_dups.
22297 2017-06-21 Andrew Pinski <apinski@cavium.com>
22299 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
22300 Increment Arith_shift and Arith_shift_reg by 1.
22301 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
22303 * config/aarch64/aarch64.c (thunderx_tunings): Enable
22304 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
22305 (aarch64_strip_extend): Add new argument and test for it.
22306 (aarch64_cheap_mult_shift_p): New function.
22307 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
22308 add a cost if it is true.
22309 Update calls to aarch64_strip_extend.
22310 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
22312 2017-06-21 Andrew Pinski <apinski@cavium.com>
22314 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
22316 (thunderxt88): Likewise.
22317 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
22318 (thunderx_prefetch_tune): New variable.
22319 (thunderx2t99_prefetch_tune): Update for the correct values.
22320 (thunderxt88_tunings): New variable.
22321 (thunderx_tunings): Use thunderx_prefetch_tune instead of
22322 generic_prefetch_tune.
22323 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
22325 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22327 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
22328 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
22329 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
22330 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
22331 (aarch64_atomic_cas<mode>, GPI): Likewise.
22333 2017-06-21 Martin Liska <mliska@suse.cz>
22335 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
22336 statements on cold and hot labels.
22337 * predict.c (tree_estimate_probability_bb): Remove the
22338 prediction from this place.
22340 2017-06-21 Martin Liska <mliska@suse.cz>
22342 PR tree-optimization/79489
22343 * gimplify.c (maybe_add_early_return_predict_stmt): New
22345 (gimplify_return_expr): Call the function.
22346 * predict.c (tree_estimate_probability_bb): Remove handling
22348 * predict.def: Update comment about early return predictor.
22349 * gimple-predict.h (is_gimple_predict): New function.
22350 * predict.def: Change default value of early return to 66.
22351 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
22353 * passes.def: Put pass_strip_predict_hints to the beginning of
22356 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
22358 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
22359 FUNCTION_DECL declarations.
22360 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
22362 (dwaf2out_decl): Likewise.
22363 * godump.c (go_early_global_decl): Skip call to the real debug hook
22364 for FUNCTION_DECL declarations.
22365 * passes.c (rest_of_decl_compilation): Skip call to the
22366 early_global_decl debug hook for FUNCTION_DECL declarations, unless
22367 -fdump-go-spec is passed.
22369 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
22371 * config/i386/i386.c (struct builtin_isa): New field pure_p.
22372 Reorder for compactness.
22373 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
22374 (def_builtin_pure, def_builtin_pure2): New functions.
22375 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
22377 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
22379 * match.pd (nop_convert): New predicate.
22380 ((A +- CST1) +- CST2): Allow some NOP conversions.
22382 2017-06-21 Jakub Jelinek <jakub@redhat.com>
22385 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
22386 with ctors/dtors if GOVD_SHARED is set.
22388 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22390 * config/aarch64/aarch64.md (movti_aarch64):
22391 Emit mov rather than orr.
22392 (movtf_aarch64): Likewise.
22393 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
22394 Emit mov rather than orr.
22396 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22398 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
22399 Swap alternatives, make integer dup more expensive.
22401 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22403 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
22404 Return true for non-tls symbols.
22406 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
22408 * config/aarch64/aarch64-cores.def (cortex-a55): New.
22409 (cortex-a75): Likewise.
22410 (cortex-a75.cortex-a55): Likewise.
22411 * config/aarch64/aarch64-tune.md: Regenerate.
22412 * doc/invoke.texi (-mtune): Document new values for -mtune.
22414 2017-06-21 Tom de Vries <tom@codesourcery.com>
22416 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
22417 stack_size feature.
22418 (Effective-Target Keywords, Other attributes): Suggest using
22419 dg-add-options stack_size feature to get stack limit in stack_size
22420 effective target documentation.
22422 2017-06-21 Julian Brown <julian@codesourcery.com>
22423 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22425 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
22426 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
22427 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
22429 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
22430 attribute type list for neon_multiply.
22431 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
22432 attribute type list for neon_multiply.
22433 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
22434 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
22435 attribute type list for neon_multiply.
22436 * config/arm/types.md (crypto_pmull): Add.
22437 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
22438 attribute type list.
22440 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
22442 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
22445 2017-06-20 Jakub Jelinek <jakub@redhat.com>
22447 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
22448 to make sure not to dereference a NULL cost_classes_ptr pointer.
22450 2017-06-20 Carl Love <cel@us.ibm.com>
22452 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22453 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22454 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22455 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22456 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22457 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22458 VMULOSW): New enum "unspec" values.
22459 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
22460 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
22461 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22462 altivec_vmulosw): New patterns.
22463 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22464 VMULOSW): Add definitions.
22466 2017-06-20 Julia Koval <julia.koval@intel.com>
22468 * config/i386/i386.c: Fix rounding expand for new pattern.
22469 * config/i386/subst.md: Fix pattern (parallel -> unspec).
22471 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
22473 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
22474 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
22476 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
22478 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
22481 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
22483 * config/aarch64/aarch64-cores.def: Rearrange to sort by
22484 architecture, then by implementer ID.
22485 * config/aarch64/aarch64-tune.md: Regenerate.
22487 2017-06-20 Richard Biener <rguenther@suse.de>
22489 PR middle-end/81097
22490 * fold-const.c (split_tree): Fold to type before negating.
22492 2017-06-20 David Malcolm <dmalcolm@redhat.com>
22494 * diagnostic-show-locus.c
22495 (selftest::test_fixit_deletion_affecting_newline): New function.
22496 (selftest::diagnostic_show_locus_c_tests): Call it.
22498 2017-06-20 Andreas Schwab <schwab@suse.de>
22501 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
22504 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22506 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
22507 __ARM_FEATURE_COPROC according to support.
22509 2017-06-20 Jakub Jelinek <jakub@redhat.com>
22511 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
22512 Rewritten to avoid overflow for > 32-bit pointers.
22515 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
22516 by removing enum keyword.
22517 (ubsan_type_descriptor): Likewise. Formatting fix.
22520 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
22521 splitter): Require TARGET_SSE2 in the condition.
22523 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
22526 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
22527 for doing vector set of SFmode on ISA 3.0.
22528 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
22529 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
22531 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
22532 SFmode value into a V4SF variable that was extracted from another
22533 V4SF variable without converting the element to double precision
22534 and back to single precision vector format.
22535 (vsx_insert_extract_v4sf_p9_2): Likewise.
22537 2017-06-19 Jakub Jelinek <jakub@redhat.com>
22539 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
22540 in UWHI to avoid undefined overflow.
22543 * ubsan.h (enum ubsan_encode_value_phase): New.
22544 (ubsan_encode_value): Change second argument to
22545 enum ubsan_encode_value_phase with default value of
22546 UBSAN_ENCODE_VALUE_GENERIC.
22547 * ubsan.c (ubsan_encode_value): Change second argument to
22548 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
22549 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
22550 create_tmp_var_raw instead of create_tmp_var and use a
22552 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
22553 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
22554 ubsan_encode_value callers.
22557 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
22558 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
22559 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
22561 2017-06-19 Richard Biener <rguenther@suse.de>
22563 PR middle-end/81118
22564 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
22565 estimates if we changed anything.
22567 2017-06-19 Richard Biener <rguenther@suse.de>
22569 PR tree-optimization/80887
22570 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
22571 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
22572 simplified lookups, then reset mprts_hook.
22573 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
22575 (try_to_simplify): Likewise.
22577 2017-06-19 Martin Liska <mliska@suse.cz>
22580 * gimplify.c (gimplify_switch_expr):
22581 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
22583 2017-06-19 Martin Liska <mliska@suse.cz>
22585 * doc/install.texi: Document that PGO runs in 4 stages.
22587 2017-06-19 Martin Liska <mliska@suse.cz>
22590 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
22591 to dispatcher function name.
22592 * multiple_target.c (replace_function_decl): New function.
22593 (create_dispatcher_calls): Redirect both edges and references.
22595 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
22597 * profile-count.c (profile_count::dump): Dump quality.
22598 (profile_count::differs_from_p): Update for unsigned val.
22599 * profile-count.h (profile_count_quality): New enum.
22600 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
22602 2017-06-19 Richard Biener <rguenther@suse.de>
22604 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
22605 struct function as arg.
22606 (estimate_numbers_of_iterations): Export overload with loop arg.
22607 (free_numbers_of_iterations_estimates_loop): Use an overload of
22608 free_numbers_of_iterations_estimates instead.
22609 * tree-cfg.c (remove_bb): Adjust.
22610 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
22611 * tree-parloops.c (gen_parallel_loop): Likewise.
22612 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
22614 (tree_unroll_loops_completely): Likewise.
22615 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
22616 Use an overload instead and export.
22617 (estimated_loop_iterations): Adjust.
22618 (max_loop_iterations): Likewise.
22619 (likely_max_loop_iterations): Likewise.
22620 (estimate_numbers_of_iterations): Take struct function as arg
22622 (loop_exits_before_overflow): Adjust.
22623 (free_numbers_of_iterations_estimates_loop): Use an overload.
22624 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
22625 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
22627 2017-06-19 Richard Biener <rguenther@suse.de>
22630 * ipa-prop.c (find_constructor_constant_at_offset): Handle
22631 RANGE_EXPR conservatively.
22633 2017-06-16 Carl Love <cel@us.ibm.com>
22635 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22636 definitions for vec_float, vec_float2, vec_floato,
22637 vec_floate built-ins.
22638 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
22639 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
22641 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
22642 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
22643 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
22644 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
22645 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
22646 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
22647 vec_floato): Add builtin defines.
22648 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
22649 Update the built-in documentation file for the new built-in
22652 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22654 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
22655 (mthumb): Mark as the negative of -marm.
22657 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22659 * doc/invoke.texi (ARM Options, -mcpu): Document supported
22661 (ARM Options, -mtune): Document that this accepts the same
22662 extension options as -mcpu.
22663 (ARM Options, -mfpu): Document addition of -mfpu=auto.
22665 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22667 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
22668 permitted extensions.
22670 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22672 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
22673 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
22674 (armv8-m.main): Add option +nodsp.
22675 * config/arm/arm-cpu-cdata.h: Regenerated.
22677 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22679 * config/arm/t-fuchsia: New file.
22680 * config.gcc (arm*-*-fuchsia*): Use it.
22682 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22684 * config/arm/t-symbian: Rewrite for new option infrastructure.
22686 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22688 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
22689 (MULTILIB_REQUIRED): Likewise.
22691 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22693 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
22694 (MULTILIB_RESUE): Likewise.
22695 (MULTILIB_MATCHES): Likewise.
22696 (MULTLIB_REQUIRED): Likewise.
22698 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22700 * config/arm/t-rtems: Rewrite for new option framework.
22702 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22704 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
22705 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
22706 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
22707 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
22708 * config/arm/t-multilib: ... here.
22709 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
22710 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
22711 armv7-a and armv8*-a when A-profile libraries have not been built.
22712 * config/arm/t-rmprofile: Rewrite.
22714 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22716 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
22717 with a backslash. Remove the backslash after substituting unescaped
22719 * doc/fragments.texi (MULTILIB_REUSE): Document it.
22721 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22723 * config.gcc: (arm*-*-*): When building a-profile libraries, force
22724 the driver to pass through the default setting of -mfloat-abi.
22725 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
22727 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
22728 (all_feat_combs): New rule.
22729 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
22731 * config/arm/t-aprofile: Rewrite.
22733 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22735 * config/arm/arm.h (FPUTYPE_AUTO): Define.
22736 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
22737 fpu is not specified by the user/command-line.
22738 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
22739 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
22740 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
22741 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
22742 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
22743 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
22745 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22747 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
22748 * config/arm/t-arm-elf: Rewritten.
22750 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22752 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
22753 have some floating-point instructions.
22754 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
22755 (TARGET_MAYBE_HARD_FLOAT): New macro.
22756 * config/arm/arm-builtins.c (arm_init_builtins): Use
22757 TARGET_MAYBE_HARD_FLOAT.
22758 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
22760 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22762 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
22763 (configargs.h): Include it.
22764 (arm_print_hint_for_fpu_option): New function.
22765 (arm_parse_fpu_option): New function.
22766 (candidate_extension): New class.
22767 (arm_canon_for_multilib): New function.
22768 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
22769 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
22770 (ARCH_CANONICAL_SPECS): New macro.
22771 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
22773 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22775 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
22776 are set after handling multilib fragments. Set target_cpu_default2
22779 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22781 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
22783 (arm*-*-*): Set target_cpu_default2 to a quoted string.
22784 * config/arm/parsecpu.awk (check_cpu): Validate any extension
22786 (check_arch): Likewise.
22787 * config/arm/arm.c (arm_configure_build_target): Handle
22788 TARGET_CPU_DEFAULT being a string constant. Scan any feature
22789 options in the default.
22791 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22793 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
22794 when an option is an alias of another.
22795 * config/arm/parsecpu.awk (optalias): New parser token.
22796 (gen_comm_data): Mark non-alias options as such. Emit entries
22797 for extension aliases.
22798 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
22799 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
22800 (armv6kz, armv6zk, armv6t2): Likewise.
22801 (armv7): Make vfpv3-d16 an alias.
22802 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
22804 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
22805 Sort in canonical order.
22806 (armv8-a): Sort in canonical order.
22807 (armv8.1-a, armv8.2-a): Likewise.
22808 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
22810 (cortex-a9): Sort in canonical order.
22811 * config/arm/arm.c (selftests.h): Include it.
22812 (arm_test_cpu_arch_data): New function.
22813 (arm_run_self_tests): New function.
22814 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
22815 (targetm): Move declaration to the end of the file.
22816 * arm-cpu-cdata.h: Regenerated.
22818 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22820 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
22821 call to target_mode_check describing the type of option passed.
22822 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
22823 (arm_target_thumb_only): Use arm_parse_arch_option_name or
22824 arm_parse_cpu_option_name to match parameters against list of
22826 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
22827 arm_arch_core_flags data structure.
22828 * config/arm/arm-cpu_cdata.h: Regenerated.
22830 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22832 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
22834 (arm_print_hint_for_cpu_option): Likewise.
22835 (arm_print_hint_for_arch_option): Likewise.
22836 (arm_parse_cpu_option_name): Likewise.
22837 (arm_parse_arch_option_name): Likewise.
22838 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
22839 of entries in the all_fpus list.
22840 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
22841 (arm_parse_cpu_option_name): Declare.
22842 (arm_parse_arch_option_name): Declare.
22843 (arm_parse_option_features): Declare.
22844 (arm_intialize_isa): Declare.
22845 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
22847 (gen_comm_data): ... here. Make definitions non-static.
22848 * config/arm/arm-cpu-data.h: Regenerated.
22849 * config/arm/arm-cpu-cdata.h: Regenerated.
22851 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22853 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
22854 (cpu_arch_extension): New structure.
22855 (cpu_arch_option, arch_option, cpu_option): New structures.
22856 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
22857 architecture types.
22858 (gen_data): Generate new format data tables.
22859 * config/arm/arm.c (cpu_tune): New structure.
22860 (cpu_option, processors): Delete.
22861 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
22862 (arm_print_hint_for_cpu_option): ... this and ...
22863 (arm_print_hint_for_arch_option): ... this.
22864 (arm_parse_arch_cpu_name): Delete. Replace with ...
22865 (arm_parse_cpu_option_name): ... this and ...
22866 (arm_parse_arch_option_name): ... this.
22867 (arm_unrecognized_feature): Change type of target parameter to
22869 (arm_parse_arch_cpu_features): Delete. Replace with ...
22870 (arm_parse_option_features): ... this.
22871 (arm_configure_build_target): Rework to use new configuration data
22873 (arm_print_tune_info): Rework for new configuration data tables.
22874 * config/arm/arm-cpu-data.h: Regenerated.
22875 * config/arm/arm-cpu.h: Regenerated.
22877 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22879 * Makefile.in (OBJS): Move sbitmap.o from here ...
22880 (OBJS-libcommon): ... to here.
22882 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22884 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
22885 (ISA_ALL_CRYPTO): New macro.
22886 (ISA_ALL_SIMD): New macro
22887 (ISA_ALL_FP): New macro.
22888 * config/arm/arm.c (fpu_bitlist): Update initializer.
22889 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
22891 (arm9e): Add fpu. Add option for nofp
22892 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
22893 (arm926ej-s, arm1026ej-s): Likewise.
22894 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
22895 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
22896 neon-fp16, neon-vfpv4, nofp and nosimd.
22897 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
22898 (cortex-a8): Add fpu. Add option for nofp.
22899 (cortex-a9): Add fpu. Add options for nosimd and nofp.
22900 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
22901 (cortex-r4f): Add fpu.
22902 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
22903 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
22905 (cortex-r8): Likewise.
22906 (cortex-m4): Add fpu. Add option for nofp.
22907 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
22908 (cortex-a17.cortex-a7): Likewise.
22909 (cortex-a32): Add fpu. Add options for crypto and nofp.
22910 (cortex-a35, cortex-a53): Likewise.
22911 (cortex-a57): Add fpu. Add option for crypto.
22912 (cortex-a72, cortex-a73): Likewise.
22913 (exynos-m1): Likewise.
22914 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
22915 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
22916 (cortex-m33): Add fpu. Add option for nofp.
22917 * config/arm/arm-cpu-cdata.h: Regenerated
22918 * config/arm/arm-cpu-data.h: Regenerated.
22920 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22922 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
22923 (armv5te, armv5tej): Likewise.
22924 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
22925 (armv7): Add options fp and vfpv3-d16.
22926 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
22927 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
22929 (armv7ve): Likewise.
22930 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
22931 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
22932 (armv8-a): Add nocrypto option.
22933 (armv8.1-a, armv8.2-a): Likewise.
22934 (armv8-m.main): add options fp, fp.dp and nofp.
22936 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22938 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
22940 (armv8-a+crc): Delete.
22941 (armv8.1-a): Add options simd, crypto and nofp.
22942 (armv8.2-a): Add options fp16, simd, crypto and nofp.
22943 (armv8.2-a+fp16): Delete.
22944 (armv8-m.main): Add option dsp.
22945 (armv8-m.main+dsp): Delete.
22946 (cortex-a8): Add fpu. Add nofp option.
22947 (cortex-a9): Add fpu. Add nofp and nosimd options.
22948 * config/arm/parsecpu.awk (gen_data): Generate option tables and
22949 link to main cpu and architecture data structures.
22950 (gen_comm_data): Only put isa attributes from the main architecture
22952 (option): New statement for architecture and CPU entries.
22953 * arm.c (struct cpu_option): New structure.
22954 (struct processors): Add entry for options.
22955 (arm_unrecognized_feature): New function.
22956 (arm_parse_arch_cpu_name): Ignore any characters after the first
22958 (arm_parse_arch_cpu_feature): New function.
22959 (arm_configure_build_target): Separate out any CPU and architecture
22960 features and parse separately. Don't error out if -mfpu=auto is
22961 used with only an architecture string.
22962 (arm_print_asm_arch_directives): New function.
22963 (arm_file_start): Call it.
22964 * config/arm/arm-cpu-cdata.h: Regenerated.
22965 * config/arm/arm-cpu-data.h: Likewise.
22966 * config/arm/arm-tables.opt: Likewise.
22968 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22970 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
22971 assembler when it is not -mfpu=auto.
22973 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22975 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
22976 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
22977 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
22978 (ASM_CPU_SPEC): Rewrite.
22979 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
22980 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
22981 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
22982 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
22983 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
22984 copied string is NUL-terminated. Also strip any characters prefixed
22986 (arm_rewrite_selected_arch): New function.
22987 (arm_rewrite_march): New function.
22989 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22991 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
22992 (x_arm_cpu_string, x_arm_tune_string): Likewise.
22993 (march, mcpu, mtune): Convert to string-based options.
22994 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
22995 (arm_parse_arch_cpu_name): New function.
22996 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
22997 identify selected architecture or CPU.
22998 (arm_option_save): New function.
22999 (TARGET_OPTION_SAVE): Redefine.
23000 (arm_option_restore): Restore string options.
23001 (arm_option_print): Print string options.
23003 2017-06-16 Martin Sebor <msebor@redhat.com>
23005 PR tree-optimization/80933
23006 PR tree-optimization/80934
23007 * builtins.c (fold_builtin_3): Do not handle bcmp here.
23008 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
23009 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
23010 (gimple_fold_builtin): Call them.
23012 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23014 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
23015 as unlikely; update profile.
23017 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23019 * predict.c (force_edge_cold): Handle declaring edges impossible
23022 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23024 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
23026 (try_unroll_loop_completely): Fix reporting.
23028 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23030 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
23032 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
23035 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
23036 if given a non-constant argument for an intrinsic which requires a
23039 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23041 * profile.c (compare_freqs): New function.
23042 (branch_prob): Sort edge list.
23043 (find_spanning_tree): Assume that the list is priority sorted.
23045 2017-06-16 Richard Biener <rguenther@suse.de>
23047 PR tree-optimization/81090
23048 * passes.def (pass_record_bounds): Remove.
23049 * tree-pass.h (make_pass_record_bounds): Likewise.
23050 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
23051 make_pass_record_bounds): Likewise.
23052 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
23053 not free niter estimates at the beginning but at the end.
23054 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
23056 2017-06-16 Richard Biener <rguenther@suse.de>
23058 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
23059 initializer to workaround ICE in host GCC 4.8.
23061 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23063 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
23065 (clone_inlined_nodes): Update.
23067 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23069 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
23070 prefetch settings, and enable prefetching by default at -O3.
23072 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23074 * config/aarch64/aarch64.c (aarch64_override_options_internal):
23075 Set flag_prefetch_loop_arrays according to tuning data.
23077 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23079 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
23080 New tune structure.
23081 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
23082 [Unrelated to main purpose of the patch] Place the pointer field last
23083 to enable type checking errors when tune structure are wrongly merged.
23084 * config/aarch64/aarch64.c (generic_prefetch_tune,)
23085 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
23086 (thunderx2t99_prefetch_tune): New tune constants.
23087 (tune_params *_tunings): Update all tunings (no functional change).
23088 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
23089 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
23090 from tunings structures.
23092 2017-06-16 Jakub Jelinek <jakub@redhat.com>
23095 * ubsan.c (instrument_null): Add T argument, use it instead
23096 of computing it based on IS_LHS.
23097 (instrument_object_size): Likewise.
23098 (pass_ubsan::execute): Adjust instrument_null and
23099 instrument_object_size callers to pass gimple_get_lhs or
23100 gimple_assign_rhs1 result to it. Use instrument_null instead of
23101 calling get_base_address and instrument_mem_ref. Handle
23102 aggregate call arguments for object-size sanitization.
23104 2017-06-16 Yury Gribov <tetra2005@gmail.com>
23106 PR tree-optimization/81089
23107 * tree-vrp.c (is_masked_range_test): Validate operands of
23110 2017-06-15 Martin Sebor <msebor@redhat.com>
23113 * dumpfile.c (dump_register): Avoid calling memset to initialize
23114 a class with a default ctor.
23115 * gcc.c (struct compiler): Remove const qualification.
23116 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
23117 * hash-table.h: Ditto.
23118 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
23120 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
23121 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
23123 * params.h (struct param_info): Make struct members non-const.
23124 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
23125 with default initialization.
23126 * vec.h (vec_copy_construct, vec_default_construct): New helper
23128 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
23129 with vec_copy_construct.
23130 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
23131 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
23132 * doc/invoke.texi (-Wclass-memaccess): Document.
23134 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23136 * emit-rtl.h (is_leaf): Update comment about local
23137 register allocator.
23139 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23142 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
23143 for a variable to have a section before checking if the section has a
23145 Set section to.persistent if persistent attribute is set.
23146 Warn if .persistent attribute is used on an automatic variable.
23148 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
23150 PR rtl-optimization/80474
23151 * reorg.c (update_block): Do not ignore instructions in a delay slot.
23153 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
23155 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
23158 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
23160 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
23161 (casesi): Emit bounds checking as RTL.
23162 (casesi_internal_mips16_<mode>): Remove bounds checking.
23164 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
23166 * config/xtensa/xtensa.c (xtensa_option_override): Append
23167 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
23168 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
23169 xtensa_doloop_hooks): Define unconditionally.
23170 (xtensa_reorg_loops): Only call reorg_loops in the presence of
23172 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
23173 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
23174 for it in xtensa_option_override.
23175 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
23176 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
23178 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
23180 * doc/cppopts.texi: Document '-' special value to -MF.
23182 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
23184 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
23185 (cortex_a53_fconst): Likewise.
23186 (cortex_a53_fpmul): Likewise.
23187 (cortex_a53_f_load_64): Likewise.
23188 (cortex_a53_f_load_many): Likewise.
23189 (cortex_a53_advsimd_alu): Likewise.
23190 (cortex_a53_advsimd_alu_q): Likewise.
23191 (cortex_a53_advsimd_mul): Likewise.
23192 (cortex_a53_advsimd_mul_q): Likewise.
23193 (fpmac bypass): Add new bypass for fpmac-fpmac case.
23194 Add missing fmul, r2f_cvt and fconst cases.
23196 2017-06-14 Richard Biener <rguenther@suse.de>
23198 PR middle-end/81088
23199 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
23201 (fold_binary_loc): When associating do not treat pre-existing
23202 TREE_OVERFLOW on literal constants as a reason to allow
23203 TREE_OVERFLOW on associated literal constants.
23205 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23207 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
23208 (MASK_FEATURES): New macro.
23209 * config/sparc/sparc.c (sparc_option_override): Remove the special
23210 handling of -mfpu and generalize it to all MASK_FEATURES switches.
23212 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23214 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
23215 a division of 0 if non-call exceptions are enabled.
23217 2017-06-14 Andrew Pinski <apinski@cavium.com>
23218 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23221 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
23222 Improve vector initialization code gen for only variable case.
23224 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23226 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
23228 2017-06-14 Richard Biener <rguenther@suse.de>
23230 PR tree-optimization/81083
23231 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
23234 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23236 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
23237 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
23238 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
23239 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
23240 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
23241 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
23243 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23245 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
23246 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
23248 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23250 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
23252 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23254 * config/rs6000/t-rtems: Don't handle SPE.
23256 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23258 * config/rs6000/t-linux: Don't handle SPE.
23260 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23262 * config/rs6000/eabispe.h: Delete file.
23264 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23266 * config/rs6000/t-spe: Delete file.
23268 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23270 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
23271 (rs6000_legitimate_offset_address_p): Return false for anything in
23272 V2SImode or V2SFmode.
23274 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23276 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
23277 except V2SF and V2SI. Rearrange the vector modes, and add comments.
23278 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
23280 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
23281 (rs6000_legitimate_offset_address_p): Ditto.
23282 (rs6000_emit_move): Ditto.
23283 (rs6000_init_builtins): Remove V4HI_type_node.
23285 2017-06-13 Martin Liska <mliska@suse.cz>
23288 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
23289 (gate_asan): Likewise.
23290 * asan.h (asan_no_sanitize_address_p): Remove the function.
23291 (sanitize_flags_p): New function.
23292 * builtins.def: Fix coding style.
23293 * common.opt: Use renamed enum value.
23294 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
23295 * doc/extend.texi: Document no_sanitize attribute.
23296 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
23297 to SANITIZE_UNDEFINED_NONDEFAULT.
23298 * gcc.c (sanitize_spec_function): Use the renamed enum value.
23299 * gimple-fold.c (optimize_atomic_compare_exchange_p):
23300 Use sanitize_flags_p.
23301 * gimplify.c (gimplify_function_tree): Likewise.
23302 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
23303 * opts.c (parse_no_sanitize_attribute): New function.
23304 (common_handle_option): Use renamed enum value.
23305 * opts.h (parse_no_sanitize_attribute): Declare.
23306 * tree.c (sanitize_flags_p): New function.
23307 * tree.h: Declared here.
23308 * tsan.c: Use sanitize_flags_p.
23309 * ubsan.c (ubsan_expand_null_ifn): Likewise.
23310 (instrument_mem_ref): Likewise.
23311 (instrument_bool_enum_load): Likewise.
23312 (do_ubsan_in_current_function): Remove the function.
23313 (pass_ubsan::execute): Use sanitize_flags_p.
23314 * ubsan.h: Remove do_ubsan_in_current_function
23315 * tree-cfg.c (print_no_sanitize_attr_value): New function.
23316 (dump_function_to_file): Use it here.
23318 2017-06-13 Martin Jambor <mjambor@suse.cz>
23320 PR tree-optimization/80803
23321 PR tree-optimization/81063
23322 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
23323 (propagate_subaccesses_across_link): Enqueue subtree whenever
23324 necessary instead of relying on the caller.
23326 2017-06-13 Martin Jambor <mjambor@suse.cz>
23328 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
23329 that have a first_link.
23330 (sort_and_splice_var_accesses): Do not check first_link before
23332 (subtree_mark_written_and_enqueue): Likewise.
23333 (propagate_all_subaccesses): Likewise and do not stop at first
23334 parent with a first_link.
23336 2017-06-13 Martin Jambor <mjambor@suse.cz>
23338 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
23341 2017-06-13 Yury Gribov <tetra2005@gmail.com>
23343 * match.pd: New pattern.
23345 2017-06-13 Yury Gribov <tetra2005@gmail.com>
23347 * tree-vrp.c (is_masked_range_test): New function.
23348 (register_edge_assert_for): Determine ranges for
23351 2017-06-13 Yury Gribov <tetra2005@gmail.com>
23353 PR tree-optimization/67328
23354 * fold-const.c (maskable_range_p): New function.
23355 (build_range_check): Generate bittests if possible.
23357 2017-06-13 Martin Liska <mliska@suse.cz>
23359 * gimple-pretty-print.c (dump_probability): Add new argument.
23360 (dump_edge_probability): Dump both probability and count.
23361 (dump_gimple_label): Likewise.
23362 (dump_gimple_bb_header): Likewise.
23364 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
23367 * config/avr/avr-devices.c: Fix indentation.
23368 * config/avr/gen-avr-mmcu-specs.c: Dito.
23370 2017-06-13 Richard Biener <rguenther@suse.de>
23372 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
23373 instead get vector type from stmt_info.
23374 (vectorizable_reduction): Adjust. Remove dead code.
23376 2017-06-13 Richard Biener <rguenther@suse.de>
23378 PR middle-end/81065
23379 * fold-const.c (extract_muldiv_1): Remove bogus distribution
23380 case of C * (x * C2 + C3).
23381 (fold_addr_of_array_ref_difference): Properly fold index difference.
23383 2017-06-12 David S. Miller <davem@davemloft.net>
23386 * config/sparc/sparc.md (return expander): Emit frame blockage if
23387 function uses alloca.
23389 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
23391 * combine.c (make_field_assignment): Check len rather than the mode
23392 precision when calling force_to_mode.
23394 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
23396 Support multilibs and devices that see flash in RAM address range.
23399 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
23400 (avr_mcu_t) <flash_pm_offset>: New field.
23401 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
23402 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
23403 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
23404 (AVR_TINY_PM_OFFSET): Remove macro.
23405 * config/avr/avr.opt (-mshort-calls): New option.
23406 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23407 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
23408 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
23409 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
23410 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
23411 instead of avr_arch->have_jmp_call.
23412 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
23413 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
23414 avr_arch->flash_pm_offset to define.
23415 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
23416 new field flash_pm_offset. Add entry for avrxmega3.
23417 (avr_texinfo): Add entry for avrxmega3.
23418 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
23419 attiny212, attiny214,
23420 attiny412, attiny414, attiny416, attiny417,
23421 attiny814, attiny816, attiny817,
23422 attiny1614, attiny1616, attiny1617,
23423 attiny3214, attiny3216, attiny3217.
23424 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
23425 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
23426 (avr_print_operand_address) [AVR_TINY]: Same.
23427 (avr_asm_init_sections) <readonly_data_section>: Only patch
23428 callback if avr_arch->flash_pm_offset = 0.
23429 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
23430 for rodata if avr_arch->flash_pm_offset != 0.
23431 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
23432 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
23433 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
23434 (m_options): Append opt_rcall.
23435 (m_dirnames): Append dir_rcall.
23436 * config/avr/t-multilib: Regenerate.
23438 * configure.ac [target=avr]: Check whether avrxmega3 default
23439 linker description file works as needed.
23440 * configure: Regenerate.
23441 * doc/avr-mmcu.texi: Regenerate.
23442 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
23443 <__AVR_ARCH__>: Document avrxmega3 and 103.
23444 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
23445 <__AVR_SHORT_CALLS__>: Document it.
23446 <__AVR_PM_BASE_ADDRESS__>: Document it.
23447 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
23448 (AVR Variable Attributes) <progmem>: Document this is
23449 not needed for avrxmega3.
23450 (AVR Named Address Spaces) <__flash>: Dito.
23452 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
23454 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
23456 2017-06-12 Doug Rupp <rupp@adacore.com>
23458 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
23459 Append vxworks-stdint.h to the tm_file list.
23460 * config/vxworks-stdint.h: New file.
23462 2017-06-12 Martin Liska <mliska@suse.cz>
23464 PR tree-optimization/81041
23465 * tree-profile.c (gimple_gen_ic_func_profiler):
23466 Create an extra BB in profile-generate
23467 (gimple_gen_time_profiler): Likewise.
23469 2017-06-12 Jakub Jelinek <jakub@redhat.com>
23471 PR tree-optimization/81003
23472 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
23473 (update_range_test): Use it instead of force_gimple_operand_gsi.
23475 2017-06-12 Richard Biener <rguenther@suse.de>
23477 PR tree-optimization/81053
23478 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
23479 with backedge value not defined in loop. Simplify def stmt
23482 2017-06-11 Tom de Vries <tom@codesourcery.com>
23485 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
23487 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
23488 nvptx_cannot_force_const_mem.
23490 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23492 * opts.c (finish_options): Move test for flag_split_stack after
23493 it has been initialized.
23495 2017-06-11 Jason Merrill <jason@redhat.com>
23497 * tree.h (id_equal): New.
23498 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
23499 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
23500 instead of strcmp of IDENTIFIER_POINTER.
23502 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23504 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
23505 (mark_all_inlined_calls_cdtor): Fix formating.
23506 (inline_transform): Rescale profile before inlining.
23508 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23510 * cgraph.h (cgraph_edge::clone): Update prototype.
23511 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
23512 (cgraph_node::create_clone): Update.
23513 (cgraph_node::create_version_clone): Update.
23514 * tree-inline.c (copy_bb): Update.
23515 (expand_call_inline): Update.
23517 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
23519 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
23520 factored out from ...
23521 (rs6000_emit_prologue): ... here.
23523 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
23525 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
23526 factored out from ...
23527 (rs6000_emit_prologue): ... here.
23529 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23531 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
23533 (handle_missing_profiles): Fix computation of tp_first_run.
23534 (counts_to_freqs): Do not touch freqs when count is 0.
23536 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23538 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
23541 2017-06-10 Tom de Vries <tom@codesourcery.com>
23543 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
23544 attributes): Document signal effective target.
23546 2017-06-10 Tom de Vries <tom@codesourcery.com>
23548 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
23549 Document effective target stack_size.
23551 2017-06-09 David Malcolm <dmalcolm@redhat.com>
23553 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
23554 to the edit_context if they can be auto-applied.
23556 2017-06-9 Ian Lance Taylor <iant@golang.org>
23558 * opts.c (finish_options): If -fsplit-stack, disable implicit
23559 -forder-blocks-and-partition.
23560 * doc/invoke.texi (Optimize Options): Document that when using
23561 -fsplit-stack -forder-blocks-and-partition is not implicitly
23564 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23566 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
23567 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
23568 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
23569 * builtins.def (abort, trap, unreachable): Declare cold.
23570 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
23571 * tree-core.h (ECF_COLD): New.
23572 * tree.c (set_call_expr_flags): Handle ECF_COLD.
23573 (build_common_builtin_nodes): Mark unreachable and abort as cold.
23575 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23577 * predict.c (unlikely_executed_stmt_p): Cleanup.
23579 2017-06-09 Richard Biener <rguenther@suse.de>
23581 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
23582 model if the ref is always written to.
23584 2017-06-09 Tamar Christina <tamar.christina@arm.com>
23586 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
23588 2017-06-09 Tamar Christina <tamar.christina@arm.com>
23590 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
23593 2017-06-09 Tom de Vries <tom@codesourcery.com>
23596 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
23597 "target cannot support label values" when encountering LABEL_REF.
23599 2017-06-09 Martin Liska <mliska@suse.cz>
23601 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
23602 (gimple_gen_ic_func_profiler): Emit direct comparison
23603 of __gcov_indirect_call_callee with NULL.
23604 (gimple_gen_time_profiler): Change probability from
23605 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
23607 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23609 * profile.c (edge_gcov_counts): Turn to pointer.
23610 (compute_branch_probabilities, compute_branch_probabilities): Update.
23611 (branch_prob): Do not clear edge_gcov_count.
23612 * profile.h (edge_gcov_counts): Turn to pointer.
23613 (edge_gcov_count): Update.
23615 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23617 * gimple.h (gimple_check_failed): Mark cold.
23619 2017-06-09 Richard Biener <rguenther@suse.de>
23621 PR tree-optimization/66623
23622 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
23623 refactor check_reduction into two parts, properly computing
23624 whether we have to check reduction validity for outer loop
23627 2017-06-09 Richard Biener <rguenther@suse.de>
23629 PR tree-optimization/79483
23630 * graphite-scop-detection.c (order): New global.
23631 (get_order): Compute bb to order mapping that satisfies code
23632 generation constraints.
23633 (cmp_pbbs): New helper.
23634 (build_scops): Start domwalk at entry block, sort generated
23637 2017-06-09 Richard Biener <rguenther@suse.de>
23639 PR middle-end/81007
23640 * ipa-polymorphic-call.c
23641 (ipa_polymorphic_call_context::restrict_to_inner_class):
23642 Skip FIELD_DECLs with error_mark_node type.
23643 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
23646 2017-06-09 Martin Liska <mliska@suse.cz>
23648 * predict.c (struct branch_predictor): New struct.
23649 (test_prediction_value_range): New test.
23650 (predict_c_tests): New function.
23651 * selftest-run-tests.c (selftest::run_tests): Run the function.
23652 * selftest.h: Declare new tests.
23654 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
23657 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
23658 gen_add3_insn did not fail.
23659 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
23660 r0, construct that number in a temporary reg and add that reg to r0.
23661 If asked to put the result in r0 as well, fail.
23663 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
23665 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23666 for early expansion of vec_eqv.
23668 2017-06-08 Jakub Jelinek <jakub@redhat.com>
23670 PR middle-end/81005
23671 * ubsan.c (instrument_null): Avoid pointless code temporary.
23672 (pass_ubsan::execute): Instrument aggregate arguments of calls.
23674 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
23678 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
23681 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
23684 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23686 * predict.c (unlikely_executed_edge_p): Move ahead.
23687 (probably_never_executed_edge_p): Use it.
23689 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
23691 PR middle-end/79988
23692 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
23693 gimple_call_builtin_p call.
23695 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23697 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
23698 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
23699 rtl_check_failed_type2, rtl_check_failed_code1,
23700 rtl_check_failed_code2, rtl_check_failed_code_mode,
23701 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
23702 rtvec_check_failed_bounds, rtl_check_failed_flag,
23703 _fatal_insn_not_found, _fatal_insn): Likewise.
23704 * tree.h (tree_contains_struct_check_failed,
23705 tree_check_failed, tree_not_check_failed,
23706 tree_class_check_failed, tree_range_check_failed,
23707 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
23708 tree_vec_elt_check_failed, phi_node_elt_check_failed,
23709 tree_operand_check_failed, omp_clause_check_failed,
23710 omp_clause_operand_check_failed, omp_clause_range_check_failed):
23713 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23715 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
23716 flag_branch_probabilities.
23717 * ipa-inline.c (edge_badness): Likewise.
23718 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
23719 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
23720 * predict.c (maybe_hot_frequency_p): Likewise.
23721 (probably_never_executed): Likewise.
23722 * sched-ebb.c (schedule_ebbs): Likewise.
23723 * sched-rgn.c (find_single_block_region): Likewise.
23724 * tracer.c (tail_duplicate): Likewise.
23726 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23728 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
23729 longer requires x_flag_profile_use.
23731 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23733 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
23734 instead of flag_reorder_blocks_and_partition.
23735 * dbxout.c (dbxout_function_end): Likewise.
23736 * dwarf2out.c (gen_subprogram_die): Likewise.
23737 * haifa-sched.c (sched_create_recovery_edges): Likewise.
23738 * hw-doloop.c (reorg_loops): Likewise.
23739 * varasm.c (assemble_start_function,
23740 assemble_end_function): Likewise.
23741 (decide_function_section): Do not check for
23742 flag_reorder_blocks_and_partition.
23744 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
23746 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
23748 (chkp_get_hard_register_fake_addr_expr): Ditto.
23749 (chkp_build_addr_expr): Add check for hard reg case.
23750 (chkp_parse_array_and_component_ref): Ditto.
23751 (chkp_find_bounds_1): Ditto.
23752 (chkp_process_stmt): Don't generate bounds store for
23755 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23757 * predict.c (maybe_hot_bb_p): Do not check profile status.
23758 (maybe_hot_edge_p): Likewise.
23759 (probably_never_executed): Check for zero counts even if profile
23761 (unlikely_executed_edge_p): New function.
23762 (unlikely_executed_stmt_p): New function.
23763 (unlikely_executed_bb_p): New function.
23764 (set_even_probabilities): Use unlikely predicates.
23765 (combine_predictions_for_bb): Likewise.
23766 (predict_paths_for_bb): Likewise.
23767 (predict_paths_leading_to_edge): Likewise.
23768 (determine_unlikely_bbs): New function.
23769 (estimate_bb_frequencies): Use it.
23770 (compute_function_frequency): Use zero counts even if profile is
23772 * profile-count.h: Fix typo.
23774 2017-08-08 Julia Koval <julia.koval@intel.com>
23776 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
23777 _mm512_mask_cvtsepi16_storeu_epi8,
23778 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
23779 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
23780 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
23781 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
23782 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
23783 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
23784 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
23785 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
23786 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
23787 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
23788 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
23789 __builtin_ia32_pmovuswb256mem_mask,
23790 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
23791 __builtin_ia32_pmovwb512mem_mask): New builtins.
23793 2017-08-08 Julia Koval <julia.koval@intel.com>
23795 PR target/73350,80862
23796 * config/i386/subst.md (round): Fix round pattern.
23797 * config/i386/i386.c (ix86_erase_embedded_rounding):
23798 Fix erasing rounding for the fixed pattern.
23800 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23802 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
23804 2017-06-08 Martin Liska <mliska@suse.cz>
23806 PR gcov-profile/80911
23807 * gcov.c (block_info::block_info): New constructor.
23809 2017-06-07 Carl Love <cel@us.ibm.com>
23811 * config/rs6000/rs6000-c: The return type of the following
23812 built-in functions was implemented as int not long long. Fix sign
23813 of return value for the unsigned version of vec_mulo and vec_mule.
23814 vector unsigned long long vec_bperm (vector unsigned long long,
23815 vector unsigned char)
23816 vector signed long long vec_mule (vector signed int,
23818 vector unsigned long long vec_mule (vector unsigned int,
23819 vector unsigned int)
23820 vector signed long long vec_mulo (vector signed int,
23822 vector unsigned long long vec_mulo (vector unsigned int,
23823 vector unsigned int)
23824 * doc/extend.texi: Fix the documentation for the built-in
23827 2017-06-07 Carl Love <cel@us.ibm.com>
23830 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
23833 2017-06-07 Carl Love <cel@us.ibm.com>
23835 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
23836 support, Generate doublehv for signed int/float for BE case only.
23838 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
23840 * doc/invoke.texi (mcx16): Rewrite.
23842 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23844 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
23845 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
23846 *mov<mode>_softfloat, and an anonymous splitter): Use
23847 nonimmediate_operand instead of rs6000_nonimmediate_operand.
23849 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23851 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
23853 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
23854 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
23855 (rs6000_debug_reg_global): Adjust.
23856 (rs6000_init_hard_regno_mode_ok): Adjust.
23857 (rs6000_dbx_register_number): Adjust.
23858 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
23859 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
23860 Remove SPE_ACC and SPEFSCR.
23861 (REG_ALLOC_ORDER): Ditto.
23862 (FRAME_POINTER_REGNUM): Change to 111.
23863 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
23864 (REG_CLASS_NAMES): Ditto.
23865 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
23866 (REGISTER_NAMES): Ditto.
23867 (ADDITIONAL_REG_NAMES): Ditto.
23868 (rs6000_reg_names): Ditto.
23869 * config/rs6000/rs6000.md: Renumber some register number
23872 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23874 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
23876 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
23877 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
23879 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
23880 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
23881 Delete the SPE high registers.
23882 (REG_ALLOC_ORDER): Ditto.
23883 (enum reg_class): Remove SPE_HIGH_REGS.
23884 (REG_CLASS_NAMES): Ditto.
23885 (REG_CLASS_CONTENTS): Delete the SPE high registers.
23886 (REGISTER_NAMES): Ditto.
23887 (rs6000_reg_names): Ditto.
23888 * doc/tm.texi.in: Remove SPE as example.
23889 * doc/tm.texi: Regenerate.
23891 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23893 * config/rs6000/8540.md (ppc8540_brinc): Delete.
23894 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
23895 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
23896 * config/rs6000/rs6000.md (type): Remove "brinc".
23898 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23900 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
23901 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
23902 * config/rs6000/linuxspe.h: Delete file.
23903 * config/rs6000/rs6000.md: Don't include spe.md.
23904 * config/rs6000/spe.h: Delete file.
23905 * config/rs6000/spe.md: Delete file.
23906 * config/rs6000/t-rs6000: Remove spe.md.
23908 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23910 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
23911 (reg_or_none500mem_operand): Delete.
23912 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
23913 instead of reg_or_none500mem_operand.
23915 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23917 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23918 handling of SPE flags.
23919 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
23921 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23923 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
23925 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
23926 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
23927 paired_divv2sf3): Similar.
23928 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
23929 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
23930 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
23932 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
23933 Rename the paired_* instruction patterns.
23934 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
23936 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
23937 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
23938 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
23939 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
23940 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
23941 PAIRED_VECTOR_MODE.
23942 (struct machine_function): Delete field spe_insn_chain_scanned_p.
23943 (spe_func_has_64bit_regs_p): Delete.
23944 (spe_expand_predicate_builtin): Delete.
23945 (spe_expand_evsel_builtin): Delete.
23946 (TARGET_DWARF_REGISTER_SPAN): Do not define.
23947 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
23948 (invalid_e500_subreg): Delete.
23949 (rs6000_legitimize_address): Always force_reg op2 as well, for
23950 paired single memory accesses.
23951 (rs6000_member_type_forces_blk): Delete.
23952 (rs6000_spe_function_arg): Delete.
23953 (rs6000_expand_unop_builtin): Delete SPE handling.
23954 (rs6000_expand_binop_builtin): Ditto.
23955 (spe_expand_stv_builtin): Delete.
23956 (bdesc_2arg_spe): Delete.
23957 (spe_expand_builtin): Delete.
23958 (spe_expand_predicate_builtin): Delete.
23959 (spe_expand_evsel_builtin): Delete.
23960 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
23961 (spe_init_builtins): Delete.
23962 (spe_func_has_64bit_regs_p): Delete.
23963 (savres_routine_name): Delete "info" parameter. Adjust callers.
23964 (rs6000_emit_stack_reset): Ditto.
23965 (rs6000_dwarf_register_span): Delete.
23966 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
23967 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
23968 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
23970 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
23972 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
23973 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
23974 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
23975 mulv2sf3, divv2sf3): Delete expanders.
23977 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23979 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
23981 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23983 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
23984 * config/rs6000/rs6000.c: Ditto.
23986 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23988 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
23989 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
23990 comparison_operator.
23992 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23994 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
23995 * config/rs6000/rs6000.opt: Ditto.
23996 * config/rs6000/t-rtems: Ditto.
23998 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24000 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
24001 TARGET_E500_SINGLE by 0, simplify.
24002 * config/rs6000/rs6000.c: Ditto.
24003 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
24004 (spe_build_register_parallel): Delete.
24005 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
24006 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
24007 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
24008 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
24009 (E500_CONVERT): Delete.
24010 * config/rs6000/spe.md: Remove many patterns and all define_constants.
24012 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24014 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
24015 * config/rs6000/dfp.md: Ditto.
24016 (negdd2, *negdd2_fpr): Merge.
24017 (absdd2, *absdd2_fpr): Merge.
24018 (negtd2, *negtd2_fpr): Merge.
24019 (abstd2, *abstd2_fpr): Merge.
24020 * config/rs6000/e500.h: Delete file.
24021 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
24022 TARGET_FPRS by 1 and simplify.
24023 * config/rs6000/rs6000-c.c: Ditto.
24024 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
24025 TARGET_DF_SPE by 0.
24026 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
24028 * config/rs6000/rs6000.md: Ditto.
24029 (floatdidf2, *floatdidf2_fpr): Merge.
24030 (move_from_CR_gt_bit): Delete.
24031 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
24032 (E500_CR_IOR_COMPARE): Delete.
24033 (All patterns that require !TARGET_FPRS): Delete.
24034 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
24036 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24038 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
24040 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24042 * graphds.c (add_edge): Intitialize edge's attached data.
24043 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
24044 pointer parameter. Call pointed function on each edge during
24045 graph traversing. Skip traversing the edge when the function
24047 (graphds_dfs, graphds_scc): Ditto.
24048 (for_each_edge): New parameter. Pass the new parameter to callback
24050 * graphds.h (skip_edge_callback): New function pointer type.
24051 (graphds_dfs, graphds_scc): New function pointer parameter.
24052 (graphds_edge_callback, for_each_edge): New parameter.
24054 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24056 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
24057 out code checking if runtime alias check is possible to below ...
24058 Call the new function.
24059 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
24060 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
24062 2017-06-07 Marek Polacek <polacek@redhat.com>
24065 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
24066 TYPE_OVERFLOW_WRAPS check.
24068 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24070 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
24071 if versioning is required.
24072 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
24073 peeling with the check for versioning.
24075 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24077 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
24078 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
24079 Set true to new parameter if new ssa variable is defined.
24080 (vect_gen_vector_loop_niters): Refactor. Set range information
24081 for the new vector loop bound variable.
24082 (vect_do_peeling): Ditto.
24084 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24086 * tree-affine.c (ssa.h): Include header file.
24087 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
24088 has wrapping overflow behavior.
24090 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24092 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
24094 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24096 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
24097 (tree_to_aff_combination): ... here.
24099 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24101 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
24102 reg_pressure model function.
24103 (ivopts_global_cost_for_size): Delete.
24104 (determine_set_costs, iv_ca_recount_cost): Call new model function
24105 ivopts_estimate_reg_pressure.
24107 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24109 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
24110 expensive than udiv. Remove floating point cases from mod.
24112 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24114 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
24115 Increase idiv cost.
24117 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24119 * config/aarch64/aarch64.md
24120 (copysignsf3): Fix mask generation.
24122 2017-06-07 Jakub Jelinek <jakub@redhat.com>
24124 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
24126 (class dump_manager): Add register_dumps method.
24127 * dumpfile.c: Include langhooks.h.
24128 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
24129 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
24130 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
24131 (dump_manager::dump_register): Start with 512 entries instead of 32.
24132 (dump_manager::register_dumps): New method.
24133 * toplev.c (general_init): Instead of invoking register_dumps
24134 langhook, invoke register_dumps method on the dump manager.
24135 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
24138 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
24140 * doc/md.texi: Clarify the restrictions on a define_insn condition.
24141 Say that # requires an associated define_split to exist, and that
24142 the define_split must be suitable for use after register allocation.
24144 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24146 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
24147 (compute_outgoing_frequencies): Also initialize zero counts.
24148 (find_many_sub_basic_blocks): Do not produce uninitialized profile
24149 around loops; preserve more of profile when nothing changes.
24151 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
24153 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
24155 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
24156 * config/arm/arm-cpu-cdata.h: Regenerate.
24157 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
24158 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
24159 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
24160 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
24161 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
24163 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
24165 * config/arm/t-rmprofile: Likewise.
24166 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
24168 2017-06-06 David S. Miller <davem@davemloft.net>
24171 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
24172 blockage if function uses alloca.
24174 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24176 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
24177 New "uid" fields to hold pretty-print IDs of group and ref.
24178 Memory references are now identified as <group_id>:<ref_id>
24179 instead of using [random] addresses.
24180 (dump_mem_details): Simplify, no functional change.
24181 (dump_mem_ref): Simplify and make output more concise.
24182 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
24183 (find_or_create_group): Initialize group uid.
24184 (record_ref): Initialize ref uid. Improve debug output.
24185 (prune_group_by_reuse, should_issue_prefetch_p,)
24186 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
24187 (mark_nontemporal_store, determine_loop_nest_reuse):
24188 Improve debug output.
24190 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24192 * dbgcnt.def (prefetch): New debug counter.
24193 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
24194 (schedule_prefetches): Stop issueing prefetches if debug counter
24197 2017-06-06 Tom de Vries <tom@codesourcery.com>
24199 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
24200 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
24202 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24204 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
24205 Use aarch64_reg_or_zero predicate for operand 4.
24206 (aarch64_compare_and_swap<mode> define_insn_and_split):
24207 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
24208 (aarch64_store_exclusive<mode>): Likewise for operand 2.
24210 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24212 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
24213 (arm_compute_save_core_reg_mask): This.
24214 (thumb1_compute_save_reg_mask): Rename into ...
24215 (thumb1_compute_save_core_reg_mask): This.
24216 (arm_compute_save_reg0_reg12_mask): Adapt comment.
24217 (arm_compute_frame_layout): Likewise.
24219 2017-06-06 Richard Biener <rguenther@suse.de>
24221 PR tree-optimization/80974
24222 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
24223 keep or clear leaders SSA info.
24225 2017-06-06 Tom de Vries <tom@codesourcery.com>
24227 * config/nvptx/nvptx.c (split_mode_p): New function.
24228 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
24230 2017-06-06 Tom de Vries <tom@codesourcery.com>
24232 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
24234 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24237 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
24240 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24242 * shrink-wrap.c (handle_simple_exit): Update profile.
24243 (try_shrink_wrapping): Upate profile.
24245 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24247 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
24248 (tree_guess_outgoing_edge_probabilities): New.
24249 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
24250 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
24252 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24254 * ipa-split.c (split_function): Initialize return bb profile.
24256 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24258 * profile.c (compute_branch_probabilities): Also initialize
24259 EXIT_BLOCK profile.
24261 2017-06-06 Richard Biener <rguenther@suse.de>
24263 PR tree-optimization/80928
24264 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
24265 (vect_analyze_loop_operations): Properly guard analysis for
24267 (vect_transform_loop): Likewise.
24268 (vect_analyze_loop_2): Also reset SLP type on PHIs.
24269 (vect_model_induction_cost): Do not cost for pure SLP.
24270 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
24271 of induction in inner loop vectorization.
24272 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
24273 (vect_get_and_check_slp_defs): Handle vect_induction_def.
24274 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
24276 (vect_analyze_slp_cost_1): Cost induction.
24277 (vect_detect_hybrid_slp_stmts): Handle PHIs.
24278 (vect_get_slp_vect_defs): Likewise.
24279 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
24280 (vect_transform_stmt): Handle SLP reductions.
24281 * tree-vectorizer.h (vectorizable_induction): Adjust.
24283 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24285 * config/rs6000/rs6000.c (make_resolver_func): Update
24286 init_lowered_empty_function call.
24288 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
24290 * doc/invoke.texi: Document the -fprofile-abs-path option.
24291 * common.opt (fprofile-abs-path): New option.
24292 * gcov-io.h (gcov_write_filename): Declare.
24293 * gcov-io.c (gcov_write_filename): New function.
24294 * coverage.c (coverage_begin_function): Use gcov_write_filename.
24295 * profile.c (output_location): Likewise.
24297 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24299 * shring-wrap.c: Revert accidental commit.
24301 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
24303 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
24305 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24307 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
24309 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
24310 profile in callgraph edge.
24311 * profile-count.h (apply_probability): If THIS is 0, then result is 0
24312 (apply_scale): Likewise.
24313 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
24314 Also scale profile when inlining function with zero profile.
24315 (initialize_cfun): Update exit block profile even when it is zero.
24316 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
24317 when profile is read.
24319 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24321 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
24322 (CLONE_*): New constants to define the processors we can generate
24323 code for with the target_clone attribute.
24324 (rs6000_clone_map): New array to identify which clone processors
24325 the current program is running on.
24326 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
24327 target_clone attribute.
24328 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
24329 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
24330 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
24331 (cpu_expand_builtin): Add support for target_clone attribute.
24332 (rs6000_valid_attribute_p): Allow "default" attribute.
24333 (get_decl_name): New debug function to simplify printing the
24334 current function name in debugging statements.
24335 (rs6000_clone_priority): New functions to support the target_clone
24336 attribute, and be able to generate code to switch between ISA 2.05
24337 through ISA 3.0 (power6 through power9).
24338 (rs6000_compare_version_priority): Likewise.
24339 (rs6000_get_function_versions_dispatcher): Likewise.
24340 (make_resolver_func): Likewise.
24341 (add_condition_to_bb): Likewise.
24342 (dispatch_function_versions): Likewise.
24343 (rs6000_generate_version_dispatcher_body): Likewise.
24344 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
24345 (fusion_gpr_load_p): Fix a spacing issue.
24346 * doc/extend.texi (Common Function Attributes): Document that the
24347 PowerPC supports the target_clone attribute.
24349 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24351 * config/arm/arm.h: explain F symbol found in description of ARM
24352 register allocation in its legend.
24354 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24356 * config/mips/frame-header-opt.c: Include profile-count.h.
24357 * config/riscv/riscv.c: Include profile-count.h
24359 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24361 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
24363 (sm_set_flag_if_changed): Add bbs field.
24364 (execute_sm_if_changed_flag_set): Pass BBS.
24365 (execute_sm): Update.
24367 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24369 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
24372 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24374 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
24375 (peephole2): New peephole2 to emit the above.
24376 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
24378 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24380 * config/aarch64/aarch64.c (define_peephole2 above
24381 *sub_<shift>_<mode>): New peephole.
24383 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24385 * config/i386/i386.c (make_resolver_func): Update.
24386 * Makefile.in: Add profile-count.h and profile-count.o
24387 * auto-profile.c (afdo_indirect_call): Update to new API.
24388 (afdo_set_bb_count): Update.
24389 (afdo_propagate_edge): Update.
24390 (afdo_propagate_circuit): Update.
24391 (afdo_calculate_branch_prob): Update.
24392 (afdo_annotate_cfg): Update.
24393 * basic-block.h: Include profile-count.h
24394 (struct edge_def): Turn count to profile_count.
24395 (struct basic_block_def): Likewie.
24396 (REG_BR_PROB_BASE): Move to profile-count.h
24397 (RDIV): Move to profile-count.h
24398 * bb-reorder.c (max_entry_count): Turn to profile_count.
24399 (find_traces): Update.
24400 (rotate_loop):Update.
24401 (connect_traces):Update.
24402 (sanitize_hot_paths):Update.
24403 * bt-load.c (migrate_btr_defs): Update.
24404 * cfg.c (RDIV): Remove.
24405 (init_flow): Use alloc_block.
24406 (alloc_block): Uninitialize count.
24407 (unchecked_make_edge): Uninitialize count.
24408 (check_bb_profile): Update.
24409 (dump_edge_info): Update.
24410 (dump_bb_info): Update.
24411 (update_bb_profile_for_threading): Update.
24412 (scale_bbs_frequencies_int): Update.
24413 (scale_bbs_frequencies_gcov_type): Update.
24414 (scale_bbs_frequencies_profile_count): New.
24415 * cfg.h (update_bb_profile_for_threading): Update.
24416 (scale_bbs_frequencies_profile_count): Declare.
24417 * cfgbuild.c (compute_outgoing_frequencies): Update.
24418 (find_many_sub_basic_blocks): Update.
24419 * cfgcleanup.c (try_forward_edges): Update.
24420 (try_crossjump_to_edge): Update.
24421 * cfgexpand.c (expand_gimple_tailcall): Update.
24422 (construct_exit_block): Update.
24423 * cfghooks.c (verify_flow_info): Update.
24424 (dump_bb_for_graph): Update.
24425 (split_edge): Update.
24426 (make_forwarder_block): Update.
24427 (duplicate_block): Update.
24428 (account_profile_record): Update.
24429 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
24430 (get_estimated_loop_iterations): Update.
24431 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
24432 (single_likely_exit): Update.
24433 * cfgloopmanip.c (scale_loop_profile): Update.
24435 (set_zero_probability): Update.
24436 (lv_adjust_loop_entry_edge): Update.
24437 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
24438 (purge_dead_edges): Update.
24439 (rtl_account_profile_record): Update.
24440 * cgraph.c (cgraph_node::create): Uninitialize count.
24441 (symbol_table::create_edge): Uninitialize count.
24442 (cgraph_update_edges_for_call_stmt_node): Update.
24443 (cgraph_edge::dump_edge_flags): Update.
24444 (cgraph_node::dump): Update.
24445 (cgraph_edge::maybe_hot_p): Update.
24446 * cgraph.h: Include profile-count.h
24447 (create_clone), create_edge, create_indirect_edge): Update.
24448 (cgraph_node): Turn count to profile_count.
24449 (cgraph_edge0: Likewise.
24450 (make_speculative, clone): Update.
24451 (create_edge): Update.
24452 (init_lowered_empty_function): Update.
24453 * cgraphclones.c (cgraph_edge::clone): Update.
24454 (duplicate_thunk_for_node): Update.
24455 (cgraph_node::create_clone): Update.
24456 * cgraphunit.c (cgraph_node::analyze): Update.
24457 (cgraph_node::expand_thunk): Update.
24458 * final.c (dump_basic_block_info): Update.
24459 * gimple-streamer-in.c (input_bb): Update.
24460 * gimple-streamer-out.c (output_bb): Update.
24461 * graphite.c (print_global_statistics): Update.
24462 (print_graphite_scop_statistics): Update.
24463 * hsa-brig.c: Include basic-block.h.
24464 * hsa-dump.c: Include basic-block.h.
24465 * hsa-gen.c (T sum_slice): Update.
24466 (convert_switch_statements):Update.
24467 * hsa-regalloc.c: Include basic-block.h.
24468 * ipa-chkp.c (chkp_produce_thunks): Update.
24469 * ipa-cp.c (struct caller_statistics): Update.
24470 (init_caller_stats): Update.
24471 (gather_caller_stats): Update.
24472 (ipcp_cloning_candidate_p): Update.
24473 (good_cloning_opportunity_p): Update.
24474 (get_info_about_necessary_edges): Update.
24475 (dump_profile_updates): Update.
24476 (update_profiling_info): Update.
24477 (update_specialized_profile): Update.
24478 (perhaps_add_new_callers): Update.
24479 (decide_about_value): Update.
24480 (ipa_cp_c_finalize): Update.
24481 * ipa-devirt.c (struct odr_type_warn_count): Update.
24482 (struct decl_warn_count): Update.
24483 (struct final_warning_record): Update.
24484 (possible_polymorphic_call_targets): Update.
24485 (ipa_devirt): Update.
24486 * ipa-fnsummary.c (redirect_to_unreachable): Update.
24487 * ipa-icf.c (sem_function::merge): Update.
24488 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
24489 * ipa-inline.c (compute_uninlined_call_time): Update.
24490 (compute_inlined_call_time): Update.
24491 (want_inline_small_function_p): Update.
24492 (want_inline_self_recursive_call_p): Update.
24493 (edge_badness): Update.
24494 (lookup_recursive_calls): Update.
24495 (recursive_inlining): Update.
24496 (inline_small_functions): Update.
24497 (dump_overall_stats): Update.
24498 (dump_inline_stats): Update.
24499 * ipa-profile.c (ipa_profile_generate_summary): Update.
24500 (ipa_propagate_frequency): Update.
24501 (ipa_profile): Update.
24502 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24503 * ipa-utils.c (ipa_merge_profiles): Update.
24504 * loop-doloop.c (doloop_modify): Update.
24505 * loop-unroll.c (report_unroll): Update.
24506 (unroll_loop_runtime_iterations): Update.
24507 * lto-cgraph.c (lto_output_edge): Update.
24508 (lto_output_node): Update.
24509 (input_node): Update.
24510 (input_edge): Update.
24511 (merge_profile_summaries): Update.
24512 * lto-streamer-in.c (input_cfg): Update.
24513 * lto-streamer-out.c (output_cfg): Update.
24514 * mcf.c (create_fixup_graph): Update.
24515 (adjust_cfg_counts): Update.
24516 (sum_edge_counts): Update.
24517 * modulo-sched.c (sms_schedule): Update.
24518 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
24519 * predict.c (maybe_hot_count_p): Update.
24520 (probably_never_executed): Update.
24521 (dump_prediction): Update.
24522 (combine_predictions_for_bb): Update.
24523 (propagate_freq): Update.
24524 (handle_missing_profiles): Update.
24525 (counts_to_freqs): Update.
24526 (rebuild_frequencies): Update.
24527 (force_edge_cold): Update.
24528 * predict.h: Include profile-count.h
24529 (maybe_hot_count_p, counts_to_freqs): UPdate.
24530 * print-rtl-function.c: Do not include cfg.h
24531 * print-rtl.c: Include basic-block.h
24532 * profile-count.c: New file.
24533 * profile-count.h: New file.
24534 * profile.c (is_edge_inconsistent): Update.
24535 (correct_negative_edge_counts): Update.
24536 (is_inconsistent): Update.
24537 (set_bb_counts): Update.
24538 (read_profile_edge_counts): Update.
24539 (compute_frequency_overlap): Update.
24540 (compute_branch_probabilities): Update; Initialize and deinitialize
24542 (branch_prob): Update.
24543 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
24544 (edge_gcov_count): New.
24545 (bb_gcov_count): New.
24546 * shrink-wrap.c (try_shrink_wrapping): Update.
24547 * tracer.c (better_p): Update.
24548 * trans-mem.c (expand_transaction): Update.
24549 (ipa_tm_insert_irr_call): Update.
24550 (ipa_tm_insert_gettmclone_call): Update.
24551 * tree-call-cdce.c: Update.
24552 * tree-cfg.c (gimple_duplicate_sese_region): Update.
24553 (gimple_duplicate_sese_tail): Update.
24554 (gimple_account_profile_record): Update.
24555 (execute_fixup_cfg): Update.
24556 * tree-inline.c (copy_bb): Update.
24557 (copy_edges_for_bb): Update.
24558 (initialize_cfun): Update.
24559 (freqs_to_counts): Update.
24560 (copy_cfg_body): Update.
24561 (expand_call_inline): Update.
24562 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
24563 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
24564 (try_unroll_loop_completely): Update.
24565 (try_peel_loop): Update.
24566 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24567 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
24568 * tree-ssa-loop-split.c (connect_loops): Update.
24569 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
24570 * tree-ssa-reassoc.c (branch_fixup): Update.
24571 * tree-ssa-tail-merge.c (replace_block_by): Update.
24572 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
24573 (compute_path_counts): Update.
24574 (update_profile): Update.
24575 (recompute_probabilities): Update.
24576 (update_joiner_offpath_counts): Update.
24577 (estimated_freqs_path): Update.
24578 (freqs_to_counts_path): Update.
24579 (clear_counts_path): Update.
24580 (ssa_fix_duplicate_block_edges): Update.
24581 (duplicate_thread_path): Update.
24582 * tree-switch-conversion.c (case_bit_test_cmp): Update.
24583 (struct switch_conv_info): Update.
24584 * tree-tailcall.c (decrease_profile): Update.
24585 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
24586 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24587 * value-prof.c (check_counter): Update.
24588 (gimple_divmod_fixed_value): Update.
24589 (gimple_mod_pow2): Update.
24590 (gimple_mod_subtract): Update.
24591 (gimple_ic_transform): Update.
24592 (gimple_stringop_fixed_value): Update.
24593 * value-prof.h (gimple_ic): Update.
24595 2017-06-02 Carl Love <cel@us.ibm.com>
24597 * config/rs6000/rs6000-c: Add support for built-in functions
24598 vector double vec_doublee (vector signed int);
24599 vector double vec_doublee (vector unsigned int);
24600 vector double vec_doublee (vector float);
24601 vector double vec_doubleh (vector signed int);
24602 vector double vec_doubleh (vector unsigned int);
24603 vector double vec_doubleh (vector float);
24604 vector double vec_doublel (vector signed int);
24605 vector double vec_doublel (vector unsigned int);
24606 vector double vec_doublel (vector float);
24607 vector double vec_doubleo (vector signed int);
24608 vector double vec_doubleo (vector unsigned int);
24609 vector double vec_doubleo (vector float);.
24610 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
24611 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
24613 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
24614 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
24615 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
24617 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
24618 vec_doublel, vec_doubleh.
24619 * doc/extend.texi: Update the built-in documentation file for the
24620 new built-in functions.
24622 2017-06-02 David Malcolm <dmalcolm@redhat.com>
24625 * ipa-inline-analysis.c (free_growth_caches): Set
24626 edge_removal_hook_holder to NULL after removing it.
24628 2017-06-02 Sudakshina Das <sudi.das@arm.com>
24630 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
24631 comparision with zero.
24633 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
24634 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
24635 for early expansion of vec_min and vec_max builtins.
24636 (builtin_function_type): Add min/max unsigned variants to those
24637 identified as having unsigned arguments.
24639 2017-06-02 Olivier Hainque <hainque@adacore.com>
24641 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
24643 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24645 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
24646 Use VALL_F16 iterator rather than VALL.
24648 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24650 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
24651 Emit CBNZ inside loop when doing a strong exchange and comparing
24652 against zero. Generate the CC flags after the loop.
24654 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
24656 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
24657 (dl_section_ref): New.
24658 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
24659 On AIX, append an expression to subtract the size of the
24660 section length to dl_section_ref.
24662 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
24664 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
24665 for early expansion of vector absolute builtins.
24667 2017-06-02 Richard Biener <rguenther@suse.de>
24669 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
24670 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
24672 2017-06-02 Richard Biener <rguenther@suse.de>
24674 PR tree-optimization/80948
24675 * tree-tailcall.c (find_tail_calls): Track stmts to move in
24676 stmt order as well.
24678 2017-06-02 Richard Biener <rguenther@suse.de>
24680 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
24682 * tree-vect-stmts.c (process_use): Do not mark backedge defs
24683 for inductions as relevant.
24685 2017-06-02 Richard Biener <rguenther@suse.de>
24687 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
24688 (vectorizable_induction): ... this. Remove dead code.
24690 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
24692 * builtins. (expand_builtin_alloca): Remove second parameter and
24693 infer its value from the first parameter instead.
24694 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
24696 2017-06-02 Jakub Jelinek <jakub@redhat.com>
24698 PR rtl-optimization/80903
24699 * loop-doloop.c (add_test): Unshare sequence.
24701 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
24703 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
24705 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
24707 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
24709 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
24710 xlogue_layout::get_instance, logue_layout::xlogue_layout,
24711 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
24712 (xlogue_layout::get_stub_rtx): Make static.
24713 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
24714 (xlogue_layout::compute_stub_managed_regs): Rename to...
24715 (xlogue_layout::count_stub_managed_regs): ...this.
24716 (xlogue_layout::is_stub_managed_reg): New function.
24717 (xlogue_layout::m_stub_names): Rename to...
24718 (xlogue_layout::s_stub_names): ...this, make static.
24719 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
24720 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
24721 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
24722 xlogue_layout::s_stub_names): Instantiate statics.
24723 (stub_managed_regs): Remove.
24724 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
24725 (disable_call_ms2sysv_xlogues): Rename to...
24726 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
24727 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
24729 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
24730 change after reload_completed.
24731 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
24733 (ix86_expand_prologue): Likewise.
24734 (ix86_expand_epilogue): Likewise.
24735 (ix86_expand_split_stack_prologue): Likewise.
24736 (ix86_compute_frame_layout): Remove frame parameter ...
24737 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
24738 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
24740 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
24741 (ix86_frame): Move from here ...
24742 * config/i386/i386.h (ix86_frame): ... to here.
24743 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
24744 complete ix86_frame data structure instead. Remove some_ld_name.
24746 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
24748 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
24749 symbols that hold a DECL_VALUE_EXPR.
24751 2017-06-01 Martin Jambor <mjambor@suse.cz>
24753 PR tree-optimization/80898
24754 * tree-sra.c (process_subtree_disqualification): Removed.
24755 (disqualify_candidate): Do not acll
24756 process_subtree_disqualification.
24757 (subtree_mark_written_and_enqueue): New function.
24758 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
24759 RHS has been disqualified and re-queue LHS if necessary. Apart
24760 from that, ignore disqualified RHS.
24762 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24764 * config/s390/s390.c (s390_emit_epilogue): Disable early return
24765 address fetch for z10 or later.
24767 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24769 * config/arc/arc.md (tst_movb): Add guard when splitting.
24771 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24773 * config/arc/arc.c (arc_can_eliminate): Test against
24774 arc_frame_pointer_needed.
24776 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24778 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
24779 to prevent store reordering.
24780 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
24781 (type): Add block type.
24782 (stack_tie): Define special instruction to be used in
24785 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24787 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
24788 constraint. It is not valid for the pattern.
24789 (noncommutative_binary_comparison): Likewise.
24791 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24793 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
24796 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24798 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
24799 be used by the reg-alloc.
24801 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24803 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
24804 reg-alloc when having mul64 or mul32x16 instructions.
24805 (mulsidi3): Likewise.
24806 (umulsidi3): Likewise.
24807 (mulsi32x16): New pattern.
24808 (mulsi64): Likewise.
24809 (mulsidi64): Likewise.
24810 (umulsidi64): Likewise.
24811 (MUL32x16_REG): Define.
24812 (mul64_600): Use MUL32x16_REG.
24813 (mac64_600): Likewise.
24814 (umul64_600): Likewise.
24815 (umac64_600): Likewise.
24817 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24819 * config/arc/arc.md (mulsi3_700): Make it commutative.
24821 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
24823 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
24825 (*sign_extendsidi2_insn): Likewise for movstosw.
24827 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
24829 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
24830 the type of the input discriminant value. Convert the
24831 discriminant value of signedness vary.
24833 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
24835 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
24836 Add to -Wall section.
24838 2017-06-01 Richard Biener <rguenther@suse.de>
24840 PR middle-end/66313
24841 * fold-const.c (fold_plusminus_mult_expr): If the factored
24842 factor may be zero use a wrapping type for the inner operation.
24843 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
24844 and handle moved defs.
24845 (process_assignment): Properly guard the unary op case. Return a
24846 tri-state indicating that moving the stmt before the call may allow
24847 to continue. Pass through to_move.
24848 (find_tail_calls): Handle moving unrelated defs before
24851 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
24854 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
24855 splitter result in the canonical way.
24857 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
24859 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
24860 also for 32bit target. Update insn attributes.
24861 (zero-extendsidi2 splitter): Allow all registers for operand 1.
24863 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
24865 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
24866 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
24867 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
24868 (_mm_maskz_min_ss): New intrinsics.
24870 2017-05-31 Martin Liska <mliska@suse.cz>
24872 * tree-vect-loop.c (vect_create_epilog_for_reduction):
24873 Change comment style to one we normally use.
24874 (vectorizable_reduction): Likewise.
24875 (vectorizable_induction): Likewise.
24876 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
24877 (vectorizable_call): Likewise.
24878 (vectorizable_simd_clone_call): Likewise.
24879 (vectorizable_conversion): Likewise.
24880 (vectorizable_assignment): Likewise.
24881 (vectorizable_shift): Likewise.
24882 (vectorizable_operation): Likewise.
24883 (vectorizable_store): Likewise.
24884 (vectorizable_load): Likewise.
24885 * tree-vectorizer.h: Likewise.
24887 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
24889 * passes.c (emergency_dump_function): New.
24890 * tree-pass.h (emergency_dump_function): Declare.
24891 * plugin.c (plugins_internal_error_function): Remove.
24892 * plugin.h (plugins_internal_error_function): Remove declaration.
24893 * toplev.c (internal_error_function): New static function. Use it...
24894 (general_init): ...here.
24896 2017-05-31 Graham Markall <graham.markall@embecosm.com>
24898 * config/arc/arc.c (arc_print_operand): Handle constant operands.
24899 (arc_rtx_costs): Add costs for new patterns.
24900 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
24901 * config/arc/predicates.md: Add _1_2_3_operand predicate.
24903 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
24905 * tree-ssa-strlen.c (get_next_strinfo): New function.
24906 (get_stridx_plus_constant): Use it.
24907 (zero_length_string): Likewise.
24908 (adjust_related_strinfos): Likewise.
24909 (adjust_last_stmt): Likewise.
24911 2017-05-31 Richard Biener <rguenther@suse.de>
24914 * config/i386/i386.c (ix86_expand_builtin): Remove assert
24915 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
24917 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
24919 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
24920 loop_vinfo argument and use of dependence distance vectors.
24921 Check instead whether the two references differ only in their
24922 initial value and assume that they have the same alignment if the
24923 difference is a multiple of the vector alignment.
24924 (vect_analyze_data_refs_alignment): Update call accordingly.
24926 2017-05-31 Martin Liska <mliska@suse.cz>
24929 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
24931 2017-05-31 Bin Cheng <bin.cheng@arm.com>
24933 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
24934 (create_intersect_range_checks): Move from ...
24935 * tree-data-ref.c (create_intersect_range_checks_index)
24936 (create_intersect_range_checks): ... to here.
24937 (create_runtime_alias_checks): New function factored from ...
24938 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
24939 here. Call above function.
24940 * tree-data-ref.h (create_runtime_alias_checks): New function.
24942 2017-05-31 Bin Cheng <bin.cheng@arm.com>
24944 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
24945 segment length for dr_b and compute it in wide_int.
24947 2017-05-31 Richard Biener <rguenther@suse.de>
24949 PR tree-optimization/80906
24950 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
24951 and pass through iv_map.
24952 (copy_bb_and_scalar_dependences): Adjust.
24953 (translate_pending_phi_nodes): Likewise.
24954 (copy_loop_close_phi_args): Handle code-generating IVs instead
24957 2017-05-30 David Malcolm <dmalcolm@redhat.com>
24959 * diagnostic-color.c (color_dict): Add "type-diff".
24960 (parse_gcc_colors): Update comment.
24961 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
24962 -fdiagnostics-show-template-tree and -fno-elide-type.
24963 (GCC_COLORS): Add type-diff to example.
24965 (-fdiagnostics-show-template-tree): New.
24966 (-fno-elide-type): New.
24967 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
24968 the pp_format_decoder callback. Call any m_format_postprocessor's
24970 (pretty_printer::pretty_printer): Initialize
24971 m_format_postprocessor.
24972 (pretty_printer::~pretty_printer): Delete any
24973 m_format_postprocessor.
24974 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
24975 (class format_postprocessor): New class.
24976 (struct pretty_printer::format_decoder): Document the new parameters.
24977 (struct pretty_printer::m_format_postprocessor): New field.
24978 * tree-diagnostic.c (default_tree_printer): Update for new
24979 bool and const char ** params.
24980 * tree-diagnostic.h (default_tree_printer): Likewise.
24982 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
24984 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
24985 (lwa_operand): Delete rs6000_gen_cell_microcode test.
24986 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
24987 rs6000_gen_cell_microcode code.
24988 (rs6000_final_prescan_insn): Delete.
24989 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
24990 "warn-cell-microcode" entries.
24991 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
24992 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
24993 throughout. Change cc_reg_not_micro_cr0_operand to
24994 cc_reg_not_cr0_operand throughout.
24995 (*extendhi<mode>2_noload): Delete.
24996 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
24997 (mwarn-cell-microcode): Delete.
24998 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
24999 -mgen-cell-microcode and -mwarn-cell-microcode.
25001 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
25004 * config/i386/constraints.md (Yd): New constraint.
25006 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
25007 and (?Yd, r) alternatives. Update insn attributes.
25008 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
25009 and (?*Yd, r) alternatives. Update insn attributes.
25010 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
25012 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
25014 * gimplify.c (gimplify_modify_expr): Don't create a
25015 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
25018 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
25020 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
25022 2017-05-30 Richard Biener <rguenther@suse.de>
25024 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
25025 and reduc_def fields.
25026 (STMT_VINFO_REDUC_TYPE): New define.
25027 (STMT_VINFO_REDUC_DEF): Likewise.
25028 (vect_force_simple_reduction): Adjust prototype.
25029 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
25030 (vect_is_simple_reduction): Remove check_reduction argument.
25031 (vect_force_simple_reduction): Adjust and set
25032 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
25033 (vectorizable_reduction): Do not re-do reduction analysis
25034 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
25035 * tree-parloops.c (gather_scalar_reductions): Adjust.
25037 2017-05-30 Richard Biener <rguenther@suse.de>
25039 PR middle-end/80901
25040 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
25043 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
25045 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
25046 Introduce unknown_misalignment parameter and remove vf.
25047 (vect_peeling_hash_get_lowest_cost):
25048 Pass unknown_misalignment parameter.
25049 (vect_enhance_data_refs_alignment):
25050 Fix unsupportable data ref treatment.
25052 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25054 * tree-vect-data-refs.c (vect_get_data_access_cost):
25055 Workaround for SLP handling.
25056 (vect_enhance_data_refs_alignment):
25057 Compute costs for doing no peeling at all, compare to the best
25058 peeling costs so far and avoid peeling if cheaper.
25060 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25062 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
25063 Return peeling info and set costs to zero for unlimited cost
25065 (vect_enhance_data_refs_alignment): Also inspect all datarefs
25066 with unknown misalignment. Compute and costs for unknown
25067 misalignment, compare them to the costs for known misalignment
25068 and choose the cheapest for peeling.
25070 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25072 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
25073 (vect_get_peeling_costs_all_drs): Create function.
25074 (vect_peeling_hash_get_lowest_cost):
25075 Use vect_get_peeling_costs_all_drs.
25076 (vect_peeling_supportable): Create function.
25077 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
25079 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25081 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
25082 DR_HAS_NEGATIVE_STEP.
25083 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
25084 (vect_enhance_data_refs_alignment): Use.
25085 (vect_duplicate_ssa_name_ptr_info): Use.
25086 * tree-vectorizer.h (dr_misalignment): Use.
25087 (known_alignment_for_access_p): Use.
25089 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
25092 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
25094 (has_section_name): New function.
25096 2017-05-30 Martin Liska <mliska@suse.cz>
25099 * auto-profile.c (get_function_decl_from_block): Fix
25102 2017-05-30 Richard Biener <rguenther@suse.de>
25104 PR middle-end/80876
25105 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
25107 2017-05-30 Martin Liska <mliska@suse.cz>
25109 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
25110 * dumpfile.h (struct dump_file_info): Remove ctors.
25112 2017-05-30 Martin Liska <mliska@suse.cz>
25114 * predict.def: Fix GNU coding style.
25116 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
25118 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25119 Mark 'to' argument with ATTRIBUTE_UNUSED.
25121 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
25123 * config/xtensa/xtensa.c (xtensa_emit_call): Use
25124 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
25125 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
25128 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
25130 * doc/install.texi (Options specification): Restore entry of
25131 --enable-sjlj-exceptions.
25133 2017-05-27 Michael Eager <eager@eagercon.com>
25136 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
25138 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
25140 * config/microblaze/microblaze.h
25141 (FIXED_REGISTERS): Update in macro.
25142 (CALL_USED_REGISTERS): Update in macro.
25144 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
25146 * doc/install.texi: Add links to macOS binary distributions.
25148 2017-05-27 Jakub Jelinek <jakub@redhat.com>
25152 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25154 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
25156 2017-05-26 Martin Liska <mliska@suse.cz>
25158 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
25160 2017-05-26 Martin Liska <mliska@suse.cz>
25162 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
25163 always leading ';; '.
25164 (dump_bb_info): Likewise.
25165 (brief_dump_cfg): Likewise.
25166 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
25167 * dumpfile.c: Remove usage of TDF_VERBOSE.
25168 * dumpfile.h (enum dump_kind): Likewise.
25169 (dump_gimple_bb_header): Do not use TDF_COMMENT.
25170 * print-tree.c (debug_verbose): Remove.
25171 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
25172 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
25173 * tree-diagnostic.c (default_tree_printer): Replace
25174 TDF_DIAGNOSTIC with TDF_SLIM.
25176 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25178 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
25179 in parameter loop, rather than loop_vinfo.
25180 (create_intersect_range_checks): Ditto.
25181 (vect_create_cond_for_alias_checks): Update call to above functions.
25183 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25185 PR tree-optimization/80815
25186 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
25187 for merging runtime alias checks. Handle negative DR_STEPs.
25189 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25191 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
25193 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
25194 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
25195 out code pruning runtime alias checks.
25196 * tree-data-ref.c (prune_runtime_alias_test_list): New function
25197 factored out from above.
25198 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
25200 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
25202 (prune_runtime_alias_test_list): New decalaration.
25204 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25206 * tree-vect-data-refs.c (compare_tree): Rename and move ...
25207 * tree-data-ref.c (data_ref_compare_tree): ... to here.
25208 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
25209 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
25210 (operator==, comp_dr_with_seg_len_pair): Ditto.
25211 (vect_prune_runtime_alias_test_list): Ditto.
25213 2017-05-26 Martin Liska <mliska@suse.cz>
25216 * params.def: Bound partial-inlining-entry-probability param.
25218 2017-05-26 Marek Polacek <polacek@redhat.com>
25221 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
25224 2017-05-26 Richard Biener <rguenther@suse.de>
25226 PR tree-optimization/80842
25227 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
25230 2017-05-26 Richard Biener <rguenther@suse.de>
25232 PR tree-optimization/80844
25233 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
25235 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
25237 * doc/md.texi (Machine Constraints): Update x86 family
25238 machine constraints section to match 'config/i386/constraints.md'.
25240 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
25242 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
25244 2017-05-25 Nathan Sidwell <nathan@acm.org>
25246 * doc/invoke.texi (--enable-languages): Update documentation.
25248 2017-05-25 Martin Liska <mliska@suse.cz>
25250 * dumpfile.c: Add TDF_FOLDING.
25251 * dumpfile.h (enum dump_kind): Likewise.
25252 * genmatch.c (dt_simplify::gen_1): Use it.
25254 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25256 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
25258 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25260 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
25261 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
25263 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25265 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
25266 * match.pd (X == C): Rewrite it here.
25267 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
25268 with_certain_nonzero_bits2): New predicates.
25269 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
25271 2017-05-24 Nathan Sidwell <nathan@acm.org>
25273 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
25276 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
25279 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
25281 * config/powerpcspe: New port. Files are copied from the rs6000
25282 port, with "rs6000" in filenames replaced by "powerpcspe".
25284 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
25286 PR rtl-optimization/80754
25287 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
25289 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
25291 * config/sparc/sparc.md (length): Return the correct value for -mflat
25292 sibcalls to match output_sibcall.
25294 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
25298 * config/rs6000/rs6000.c (struct machine_function): Add new field
25300 (rs6000_get_separate_components): Init that field, use it.
25301 (rs6000_components_for_bb): Use the field.
25303 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
25305 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
25307 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
25309 PR middle-end/80823
25310 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
25312 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25315 * config/s390/s390.c (s390_check_qrst_address): Check incoming
25316 address against address_operand predicate.
25317 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
25319 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
25321 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
25322 parameters passed indirectly.
25324 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
25326 * config/i386/i386.md (*movdi_internal): Remove SSE4
25327 alternative 18 (?r, *v). Update insn attributes.
25328 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
25329 Update insn attributes.
25330 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
25331 Update insn attributes.
25332 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
25333 alternative 1 (r, v). Remove isa attribute.
25334 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
25335 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
25336 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
25338 2017-05-23 Tom de Vries <tom@codesourcery.com>
25340 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
25343 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25345 * cgraphunit.c (symbol_table::process_new_functions): Update.
25346 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
25347 (inline_generate_summary): Rename to ...
25348 (ipa_fn_summary_generate): ... this one.
25349 (inline_read_summary): Rename to ...
25350 (ipa_fn_summary_read): ... this one.
25351 (inline_write_summary): Rename to ...
25352 (ipa_fn_summary_write): ... this one.
25353 (inline_free_summary): Rename to ...
25354 (ipa_free_fn_summary): ... this one.
25355 (pass_data_local_fn_summary, pass_local_fn_summary,
25356 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
25357 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
25358 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
25359 make_pass_ipa_fn_summary): New.
25360 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
25361 inline_write_summary, inline_free_summary): Remove.
25362 (ipa_free_fn_summary) : New.
25363 * ipa-inline.c (ipa_inline): Update.
25364 (pass_ipa_inline): Do not generate summaries.
25365 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
25367 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
25368 and add pass_ipa_fn_summary.
25369 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
25371 (make_pass_inline_parameters): Remove.
25373 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
25375 * omp-low.c (struct omp_context): Remove "default_kind" member.
25378 * omp-offload.c (execute_oacc_device_lower): Remove the
25379 parallelism dimensions function attributes for unparallelized
25380 OpenACC kernels constructs.
25382 2017-05-23 Martin Liska <mliska@suse.cz>
25384 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
25386 (cgraph_edge::make_speculative): Likewise.
25387 (cgraph_edge::resolve_speculation): Likewise.
25388 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
25389 (cgraph_node::dump): Likewise.
25390 * cgraph.h: Likewise.
25391 * cgraphunit.c (analyze_functions): Likewise.
25392 (symbol_table::compile): Likewise.
25393 * ipa-cp.c (print_all_lattices): Likewise.
25394 (determine_versionability): Likewise.
25395 (initialize_node_lattices): Likewise.
25396 (ipcp_verify_propagated_values): Likewise.
25397 (estimate_local_effects): Likewise.
25398 (update_profiling_info): Likewise.
25399 (create_specialized_node): Likewise.
25400 (perhaps_add_new_callers): Likewise.
25401 (decide_about_value): Likewise.
25402 (decide_whether_version_node): Likewise.
25403 (identify_dead_nodes): Likewise.
25404 (ipcp_store_bits_results): Likewise.
25405 * ipa-devirt.c (dump_targets): Likewise.
25406 (ipa_devirt): Likewise.
25407 * ipa-icf.c (sem_item::dump): Likewise.
25408 (sem_function::equals): Likewise.
25409 (sem_variable::equals): Likewise.
25410 (sem_item_optimizer::read_section): Likewise.
25411 (sem_item_optimizer::execute): Likewise.
25412 (congruence_class::dump): Likewise.
25413 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
25414 (dump_inline_summary): Likewise.
25415 (estimate_node_size_and_time): Likewise.
25416 (inline_analyze_function): Likewise.
25417 * ipa-inline-transform.c (inline_call): Likewise.
25418 * ipa-inline.c (report_inline_failed_reason): Likewise.
25419 (want_early_inline_function_p): Likewise.
25420 (edge_badness): Likewise.
25421 (update_edge_key): Likewise.
25422 (inline_small_functions): Likewise.
25423 * ipa-profile.c (ipa_profile): Likewise.
25424 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
25425 (ipa_make_edge_direct_to_target): Likewise.
25426 (remove_described_reference): Likewise.
25427 (ipa_impossible_devirt_target): Likewise.
25428 (propagate_controlled_uses): Likewise.
25429 (ipa_print_node_params): Likewise.
25430 (ipcp_transform_function): Likewise.
25431 * ipa-pure-const.c (pure_const_read_summary): Likewise.
25432 (propagate_pure_const): Likewise.
25433 * ipa-reference.c (generate_summary): Likewise.
25434 (read_write_all_from_decl): Likewise.
25435 (propagate): Likewise.
25436 (ipa_reference_read_optimization_summary): Likewise.
25437 * ipa-utils.c (ipa_merge_profiles): Likewise.
25438 * ipa.c (walk_polymorphic_call_targets): Likewise.
25439 (symbol_table::remove_unreachable_nodes): Likewise.
25440 (ipa_single_use): Likewise.
25441 * passes.c (execute_todo): Likewise.
25442 * predict.c (drop_profile): Likewise.
25443 * symtab.c (symtab_node::get_dump_name): New function.
25444 (symtab_node::dump_name): Likewise.
25445 (symtab_node::dump_asm_name): Likewise.
25446 (symtab_node::dump_references): Likewise.
25447 (symtab_node::dump_referring): Likewise.
25448 (symtab_node::dump_base): Likewise.
25449 (symtab_node::debug_symtab): Likewise.
25450 * tree-sra.c (convert_callers_for_node): Likewise.
25451 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
25452 * value-prof.c (init_node_map): Likewise.
25454 2017-05-23 Martin Liska <mliska@suse.cz>
25456 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
25457 and symtab_node::debug_symtab to symbol_table::debug.
25458 * cgraphunit.c (analyze_functions): Use the renamed function.
25459 (symbol_table::compile): Likewise.
25460 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
25461 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
25462 * passes.c (execute_todo): Likewise.
25463 * symtab.c (symbol_table::dump): New function.
25464 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
25466 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25468 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
25469 that nonconst implies exec.
25471 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25473 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
25474 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
25475 (inline_edge_summary_vec): Turn into ...
25476 (ipa_call_summaries): ... this one.
25477 (redirect_to_unreachable, edge_set_predicate,
25478 evaluate_properties_for_edge, inline_summary_alloc,
25479 reset_ipa_call_summary, reset_inline_summary,
25480 inline_summary_t::duplicate): Update.
25481 (inline_edge_duplication_hook): Turn to ...
25482 (ipa_call_summary_t::duplicate): ... this one.
25483 (inline_edge_removal_hook): Turn to ...
25484 (ipa_call_summary_t::remove): ... this one.
25485 (dump_inline_edge_summary): Turn to ...
25486 (dump_ipa_call_summary): ... this one.
25487 (estimate_function_body_sizes): Update.
25488 (inline_update_callee_summaries): Update.
25489 (remap_edge_change_prob): Update.
25490 (remap_edge_summaries): Update.
25491 (inline_merge_summary): Update.
25492 (do_estimate_edge_time): Update.
25493 (inline_generate_summary): Update.
25494 (inline_read_section): Update.
25495 (inline_read_summary): Update.
25496 (inline_free_summary): Update.
25497 * ipa-inline.c (can_inline_edge_p): Update.
25498 (compute_inlined_call_time): Update.
25499 (want_inline_small_function_p): Update.
25500 (edge_badness): Update.
25501 (early_inliner): Update.
25502 * ipa-inline.h (inline_edge_summary): Turn to ...
25503 (ipa_call_summary): ... this one.
25504 (ipa_call_summary_t): New class.
25505 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
25506 (ipa_call_summaries): New.
25507 (inline_edge_summary): Remove.
25508 (estimate_edge_growth): Update.
25509 * ipa-profile.c (ipa_propagate_frequency_1): Update.
25510 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
25511 * ipa-split.c (execute_split_functions): Update.
25512 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
25514 2017-05-23 Tom de Vries <tom@codesourcery.com>
25516 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
25517 attributes): Document rdrand effective target.
25519 2017-05-23 Tom de Vries <tom@codesourcery.com>
25521 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
25522 attributes): Sort alphabetically.
25524 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
25526 * config/avr/genmultilib.awk: Use gsub instead of gensub.
25528 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
25531 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
25532 V2DF/V2DI splat into two separate patterns, one that handles
25533 registers, and the other that only handles memory. Drop support
25534 for splatting from a GPR on ISA 2.07 and then splitting the
25535 splat into direct move and splat.
25536 (vsx_splat_<mode>_reg): Likewise.
25537 (vsx_splat_<mode>_mem): Likewise.
25539 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
25541 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
25543 2017-05-22 Jakub Jelinek <jakub@redhat.com>
25545 PR middle-end/80809
25546 * omp-low.c (finish_taskreg_remap): New function.
25547 (finish_taskreg_scan): If unit size of ctx->record_type
25548 is non-constant, unshare the size expression and replace
25549 decls in it with possible outer var refs.
25551 PR middle-end/80809
25552 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
25553 GOVD_SHARED rather than GOVD_PRIVATE with it.
25554 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
25555 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
25557 PR middle-end/80853
25558 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
25559 as last argument to build_outer_var_ref for pointer bases of array
25560 section reductions.
25562 2017-05-19 Martin Sebor <msebor@redhat.com>
25564 * print-tree.c (print_node): Print DECL_READ_P flag.
25566 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25568 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
25569 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
25570 * cgraph.c: Likewise.
25571 * cgraphunit.c: Likewise.
25572 * gengtype.c: Likewise.
25573 * ipa-cp.c: Likewise.
25574 * ipa-devirt.c: Likewise.
25575 * ipa-icf.c: Likewise.
25576 * ipa-predicate.c: Likewise.
25577 * ipa-profile.c: Likewise.
25578 * ipa-prop.c: Likewise.
25579 * ipa-split.c: Likewise.
25581 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
25582 edge_predicate_pool, dump_inline_hints,
25583 inline_summary::account_size_time, redirect_to_unreachable,
25584 edge_set_predicate, set_hint_predicate,
25585 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
25586 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
25587 inline_summary_t::remove, remap_hint_predicate_after_duplication,
25588 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
25589 ipa_call_summary_t::remove, initialize_growth_caches,
25590 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
25591 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
25592 mark_modified, unmodified_parm_1, unmodified_parm,
25593 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
25594 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
25595 compute_bb_predicates, will_be_nonconstant_expr_predicate,
25596 will_be_nonconstant_predicate, record_modified_bb_info,
25597 get_minimal_bb, record_modified, param_change_prob,
25598 phi_result_unknown_predicate, predicate_for_phi_result,
25599 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
25600 estimate_function_body_sizes, compute_inline_parameters,
25601 compute_inline_parameters_for_curren, pass_data_inline_parameters,
25602 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
25603 inline_update_callee_summaries, remap_edge_change_prob,
25604 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
25605 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
25606 inline_analyze_function, inline_summary_t::insert,
25607 inline_generate_summary, read_ipa_call_summary, inline_read_section,
25608 inline_read_summary, write_ipa_call_summary, inline_write_summary,
25609 inline_free_summary): Move to ipa-fnsummary.h
25610 (predicate_t): Remove.
25611 * ipa-fnsummary.c: New file.
25612 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
25613 (enum inline_hints_vals, inline_hints, agg_position_info,
25614 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
25615 inline_summaries, ipa_call_summary, ipa_call_summary_t,
25616 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
25617 dump_inline_summary, dump_inline_hints, inline_generate_summary,
25618 inline_read_summary, inline_write_summary, inline_free_summary,
25619 inline_analyze_function, initialize_inline_failed,
25620 inline_merge_summary, inline_update_overall_summary,
25621 compute_inline_parameters): Move to ipa-fnsummary.h
25622 * ipa-fnsummary.h: New file.
25623 * ipa-inline-transform.h: Include ipa-inline.h.
25624 * ipa-inline.c: LIkewise.
25626 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25628 * ipa-inline.c (edge_badness): Use inlined_time instead of
25629 inline_summaries->get.
25631 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25633 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
25635 2017-05-22 Nathan Sidwell <nathan@acm.org>
25637 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
25638 (fdump-lang): Document 'raw' option.
25639 * dumpfile.h (TDI_tu): Delete.
25640 * dumpfile.c (dump_files): Remove translation-unit.
25641 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
25643 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
25645 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
25646 command option from $(AWK) call.
25647 * config/avr/genmultilib.awk: Simplify and rewrite so that it
25648 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
25649 [FORMAT]: Remove handling of variable.
25650 * config/avr/t-multilib: Regenerate.
25652 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25654 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
25656 (dump_inline_summary): Do not print self_time.
25657 (estimate_function_body_sizes): Do not set self_time.
25658 (compute_inline_parameters): Likewise.
25659 (inline_read_section, inline_write_summary): Do not stream self_time.
25660 * ipa-inline.h (inline_summary): Drop self_time.
25662 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25664 * ipa-inline-analysis.c (account_size_time): Rename to ...
25665 (inline_summary::account_size_time): ... this one.
25666 (reset_ipa_call_summary): Turn to ...
25667 (ipa_call_summary::reset): ... this one.
25668 (reset_inline_summary): Turn to ...
25669 (inline_summary::reset): ... this one.
25670 (inline_summary_t::remove): Update.
25671 (inline_summary_t::duplicate): Update.
25672 (ipa_call_summary_t::remove): Update.
25673 (dump_inline_summary): Update.
25674 (estimate_function_body_sizes): Update.
25675 (compute_inline_parameters): Update.
25676 (estimate_node_size_and_time): Update.
25677 (inline_merge_summary): Update.
25678 (inline_update_overall_summary): Update.
25679 (inline_read_section): Update.
25680 (inline_write_summary): Update.
25681 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
25682 add account_size_time and reset member functions.
25683 (ipa_call_summary): Add reset function.
25684 * ipa-predicate.h (predicate::operator &): Constify.
25686 2017-05-22 Richard Biener <rguenther@suse.de>
25688 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
25690 2017-05-19 Jason Merrill <jason@redhat.com>
25692 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
25694 2017-05-19 Marek Polacek <polacek@redhat.com>
25697 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
25698 TYPE_OVERFLOW_WRAPS checks.
25700 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
25702 * tree-core.h (enum omp_clause_default_kind): Add
25703 "OMP_CLAUSE_DEFAULT_PRESENT".
25704 * tree-pretty-print.c (dump_omp_clause): Handle it.
25705 * gimplify.c (enum gimplify_omp_var_data): Add
25706 "GOVD_MAP_FORCE_PRESENT".
25707 (gimplify_adjust_omp_clauses_1): Map it to
25708 "GOMP_MAP_FORCE_PRESENT".
25709 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
25711 * gimplify.c (oacc_default_clause): Clarify.
25713 2017-05-19 Nathan Sidwell <nathan@acm.org>
25715 LANG_HOOK_REGISTER_DUMPS
25716 * toplev.c (general_init): Call register dump lang hook.
25717 * doc/invoke.texi: Document -fdump-lang option family.
25718 * dumpfile.c (dump_files): Remove class dump here.
25719 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
25720 * dumpfile.h (tree_dump_index): Remove TDI_class.
25721 * langhooks-def.h (lhd_register_dumps): Declare.
25722 (LANG_HOOKS_REGISTER_DUMPS): Define.
25723 (LANG_HOOKS_INITIALIZER): Add it.
25724 * langhooks.c (lhd_register_dumps): Define.
25725 * langhooks.h (struct lang_hooks): Add register_dumps.
25727 2017-05-19 Nathan Sidwell <nathan@acm.org>
25729 * context.h (context::set_passes): New.
25730 * context.c (context::context): Do not create pass manager.
25731 * toplev.c (general_init): Create pass manager here.
25733 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
25735 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
25736 use this splitter if two add or or instructions would also work for
25737 the constant we want to generate.
25739 2017-05-19 Richard Biener <rguenther@suse.de>
25742 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
25743 predicate evaluation.
25745 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
25747 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
25749 * ipa-inline.c (want_inline_small_function_p): Do not cast to
25752 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
25754 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
25755 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
25756 (inline_edge_summary_vec): Turn into ...
25757 (ipa_call_summaries): ... this one.
25758 (redirect_to_unreachable, edge_set_predicate,
25759 evaluate_properties_for_edge, inline_summary_alloc,
25760 reset_ipa_call_summary, reset_inline_summary,
25761 inline_summary_t::duplicate): Update.
25762 (inline_edge_duplication_hook): Turn to ...
25763 (ipa_call_summary_t::duplicate): ... this one.
25764 (inline_edge_removal_hook): Turn to ...
25765 (ipa_call_summary_t::remove): ... this one.
25766 (dump_inline_edge_summary): Turn to ...
25767 (dump_ipa_call_summary): ... this one.
25768 (estimate_function_body_sizes): Update.
25769 (inline_update_callee_summaries): Update.
25770 (remap_edge_change_prob): Update.
25771 (remap_edge_summaries): Update.
25772 (inline_merge_summary): Update.
25773 (do_estimate_edge_time): Update.
25774 (inline_generate_summary): Update.
25775 (inline_read_section): Update.
25776 (inline_read_summary): Update.
25777 (inline_free_summary): Update.
25778 * ipa-inline.c (can_inline_edge_p): Update.
25779 (compute_inlined_call_time): Update.
25780 (want_inline_small_function_p): Update.
25781 (edge_badness): Update.
25782 (early_inliner): Update.
25783 * ipa-inline.h (inline_edge_summary): Turn to ...
25784 (ipa_call_summary): ... this one.
25785 (ipa_call_summary_t): New class.
25786 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
25787 (ipa_call_summaries): New.
25788 (inline_edge_summary): Remove.
25789 (estimate_edge_growth): Update.
25790 * ipa-profile.c (ipa_propagate_frequency_1): Update.
25791 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
25792 * ipa-split.c (execute_split_functions): Update.
25793 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
25795 2017-05-19 Richard Biener <rguenther@suse.de>
25797 PR middle-end/80764
25798 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
25800 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
25802 * config/rs6000/rs6000.c (struct machine_function): Add field
25803 fpr_is_wrapped_separately.
25804 (rs6000_get_separate_components): Use 64 components. Handle the
25805 new FPR components.
25806 (rs6000_components_for_bb): Handle the FPR components.
25807 (rs6000_emit_prologue_components): Handle the FPR components.
25808 (rs6000_emit_epilogue_components): Handle the FPR components.
25809 (rs6000_set_handled_components): Handle the FPR components.
25810 (rs6000_emit_prologue): Don't output prologue code for those FPRs
25811 that are already separately shrink-wrapped.
25812 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
25813 that are already separately shrink-wrapped.
25815 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25818 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
25821 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
25822 (define_peephole2 for Altivec d-form load): Add peepholes to catch
25823 cases where the register allocator uses a move and an offsettable
25824 memory operation to/from a FPR register on ISA 2.06/2.07.
25825 (define_peephole2 for Altivec d-form store): Likewise.
25827 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
25830 * config/i386/mmx.md (*mov<mode>_internal): Enable
25831 alternatives 11, 12, 13 and 14 also for 32bit targets.
25832 Remove alternatives 15, 16, 17 and 18.
25833 * config/i386/sse.md (vec_concatv2di): Change
25834 alternative (!x, *y) to (x, ?!*Yn).
25836 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
25838 * dumpfile.h (enum dump_kind): Remove stray comma.
25840 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
25842 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
25843 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
25844 predicate::num_conditions
25845 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
25846 (CHANGED): turn into predicate::changed.
25847 (agg_position_info): Move to ipa-predicate.h
25848 (add_condition, predicate::add_clause, predicate::operator &=,
25849 predicate::or_with, predicate::evaluate, predicate::probability,
25850 dump_condition, dump_clause, predicate::dump,
25851 predicate::remap_after_duplication, predicate::remap_after_inlining,
25852 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
25853 (evaluate_conditions_for_known_args): Update.
25854 (set_cond_stmt_execution_predicate): Update.
25855 * ipa-inline.h: Include ipa-predicate.h
25856 (condition, inline_param_summary, conditions, agg_position_info,
25857 predicate): Move to ipa-predicate.h
25858 * ipa-predicate.c: New file.
25859 * ipa-predicate.h: New file.
25861 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
25863 * final.c (leaf_function_p): Check we are not in a sequence.
25865 2017-05-18 Martin Liska <mliska@suse.cz>
25867 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
25868 * dumpfile.c (dump_register): Use new enum dump_kind.
25869 (get_dump_file_name): Likewise.
25870 (dump_enable_all): Likewise.
25871 (dump_switch_p_1): Likewise.
25872 (enable_rtl_dump_file): Remove usage of TDF_RTL.
25873 * dumpfile.h (enum dump_kind): New enum type.
25874 (struct dump_file_info): Create constructor and
25875 format fields and comments.
25876 * passes.c (pass_manager::register_one_dump_file):
25878 * statistics.c (statistics_early_init): Likewise.
25879 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
25880 TDF_TREE with TDF_SLIM.
25881 (gather_memory_references_ref): Likewise.
25883 2017-05-18 Martin Liska <mliska@suse.cz>
25885 * vec.h (struct vnull): Use it.
25887 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
25889 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
25890 (true_predicate, false_predicate, true_predicate_p,
25891 false_predicate_p): Remove.
25892 (single_cond_predicate, not_inlined_predicate): Turn to member function
25894 (add_condition): Update.
25895 (add_clause): Turn to...
25896 (predicate::add_clause): ... this one; update; allow passing NULL
25898 (and_predicates): Turn to ...
25899 (predicate::operator &=): ... this one.
25900 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
25901 (or_predicates): Turn to ...
25902 (predicate::or_with): ... this one.
25903 (evaluate_predicate): Turn to ...
25904 (predicate::evaluate): ... this one.
25905 (predicate_probability): Turn to ...
25906 (predicate::probability): ... this one.
25907 (dump_condition): Update.
25908 (dump_predicate): Turn to ...
25909 (predicate::dump): ... this one.
25910 (account_size_time): Update.
25911 (edge_set_predicate): Update.
25912 (set_hint_predicate): UPdate.
25913 (evaluate_conditions_for_known_args): Update.
25914 (evaluate_properties_for_edge): Update.
25915 (remap_predicate_after_duplication): Turn to...
25916 (predicate::remap_after_duplication): ... this one.
25917 (remap_hint_predicate_after_duplication): Update.
25918 (inline_summary_t::duplicate): UPdate.
25919 (dump_inline_edge_summary): Update.
25920 (dump_inline_summary): Update.
25921 (set_cond_stmt_execution_predicate): Update.
25922 (set_switch_stmt_execution_predicate): Update.
25923 (compute_bb_predicates): Update.
25924 (will_be_nonconstant_expr_predicate): Update.
25925 (will_be_nonconstant_predicate): Update.
25926 (phi_result_unknown_predicate): Update.
25927 (predicate_for_phi_result): Update.
25928 (array_index_predicate): Update.
25929 (estimate_function_body_sizes): Update.
25930 (estimate_node_size_and_time): Update.
25931 (estimate_ipcp_clone_size_and_time): Update.
25932 (remap_predicate): Rename to ...
25933 (predicate::remap_after_inlining): ... this one.
25934 (remap_hint_predicate): Update.
25935 (inline_merge_summary): Update.
25936 (inline_update_overall_summary): Update.
25937 (estimate_size_after_inlining): Update.
25938 (read_predicate): Rename to ...
25939 (predicate::stream_in): ... this one.
25940 (read_inline_edge_summary): Update.
25941 (write_predicate): Rename to ...
25942 (predicate::stream_out): ... this one.
25943 (write_inline_edge_summary): Update.
25944 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
25945 (clause_t): Turn to uint32_t
25946 (predicate): Turn to class; implement constructor and operators
25948 (size_time_entry): Update.
25949 (inline_summary): Update.
25950 (inline_edge_summary): Update.
25952 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
25954 * fold-const.c (fold_binary_loc): Move transformation...
25955 * match.pd (C - X CMP X): ... here.
25957 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
25959 * config/sparc/sparc.c (sparc_option_override): Set function
25960 alignment for -mcpu=niagara7 to 64 to match the I$ line.
25961 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
25963 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
25965 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
25967 2017-05-18 Marek Polacek <polacek@redhat.com>
25970 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
25971 (pass_ubsan::execute): Call gimple_assign_single_p instead of
25972 gimple_assign_load_p.
25974 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
25976 PR middle-end/80692
25977 * real.c (do_compare): Give decimal_do_compare preference over
25978 comparing just the signs.
25980 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
25982 * doc/md.texi (Canonicalization of Instructions): Describe the
25983 canonical form of instructions that inherently set a condition
25986 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
25988 PR middle-end/80775
25989 * tree-cfg.c: Move deletion of unreachable case statements to after
25990 the merging of consecutive case labels.
25992 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
25994 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
25995 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
25996 restoring of callee-saved registers.
25998 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
26000 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
26001 * config/visium/visium.c (single_set_and_flags): Likewise.
26002 * config/visium/visium.md (Substitutions): Likewise.
26004 2017-05-17 Martin Liska <mliska@suse.cz>
26006 * cfg.c: Introduce dump_flags_t type and
26007 use it instead of int type.
26009 * cfghooks.c: Likewise.
26010 * cfghooks.h (struct cfg_hooks): Likewise.
26011 * cfgrtl.c: Likewise.
26012 * cfgrtl.h: Likewise.
26013 * cgraph.c (cgraph_node::get_body): Likewise.
26014 * coretypes.h: Likewise.
26015 * domwalk.c: Likewise.
26016 * domwalk.h: Likewise.
26017 * dumpfile.c (struct dump_option_value_info): Likewise.
26018 (dump_enable_all): Likewise.
26019 (dump_switch_p_1): Likewise.
26020 (opt_info_switch_p): Likewise.
26021 * dumpfile.h (enum tree_dump_index): Likewise.
26022 (struct dump_file_info): Likewise.
26023 * genemit.c: Likewise.
26024 * generic-match-head.c: Likewise.
26025 * gengtype.c (open_base_files): Likewise.
26026 * gimple-pretty-print.c: Likewise.
26027 * gimple-pretty-print.h: Likewise.
26028 * graph.c (print_graph_cfg): Likewise.
26029 * graphite-scop-detection.c (dot_all_sese): Likewise.
26030 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
26031 * loop-unroll.c (report_unroll): Likewise.
26032 * passes.c (pass_manager::register_one_dump_file): Likewise.
26033 * print-tree.c: Likewise.
26034 * statistics.c: Likewise.
26035 * tree-cfg.c: Likewise.
26036 * tree-cfg.h: Likewise.
26037 * tree-dfa.c: Likewise.
26038 * tree-dfa.h: Likewise.
26039 * tree-dump.c (dump_function): Likewise.
26040 * tree-dump.h (struct dump_info): Likewise.
26041 * tree-pretty-print.c: Likewise.
26042 * tree-pretty-print.h: Likewise.
26043 * tree-ssa-live.c: Likewise.
26044 * tree-ssa-live.h: Likewise.
26045 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
26046 * tree-vect-loop.c: Likewise.
26047 * tree-vect-slp.c: Likewise.
26049 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
26050 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26052 PR tree-optimization/80457
26053 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
26054 of all arguments to a statement as scalar_to_vec operations.
26055 (vectorizable_call): Adjust call to vect_model_simple_cost for
26057 (vectorizable_conversion): Likewise.
26058 (vectorizable_assignment): Likewise.
26059 (vectorizable_shift): Likewise.
26060 (vectorizable_operation): Likewise.
26061 (vectorizable_comparison): Likewise.
26062 (vect_is_simple_cond): Record the def types for operands.
26063 (vectorizable_condition): Likewise, call vect_model_simple_cost.
26064 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
26065 for statement argument count.
26067 2017-05-16 Carl Love <cel@us.ibm.com>
26069 * config/rs6000/rs6000-c: Add support for built-in functions
26070 vector unsigned long long vec_bperm (vector unsigned long long,
26071 vector unsigned char)
26072 vector signed long long vec_mule (vector signed int,
26074 vector unsigned long long vec_mule (vector unsigned int,
26075 vector unsigned int)
26076 vector signed long long vec_mulo (vector signed int,
26078 vector unsigned long long vec_mulo (vector unsigned int,
26079 vector unsigned int)
26080 vector signed char vec_sldw (vector signed char,
26081 vector signed char,
26083 vector unsigned char vec_sldw (vector unsigned char,
26084 vector unsigned char,
26086 vector signed short vec_sldw (vector signed short,
26087 vector signed short,
26089 vector unsigned short vec_sldw (vector unsigned short,
26090 vector unsigned short,
26092 vector signed int vec_sldw (vector signed int,
26095 vector unsigned int vec_sldw (vector unsigned int,
26096 vector unsigned int,
26098 vector signed long long vec_sldw (vector signed long long,
26099 vector signed long long,
26101 vector unsigned long long vec_sldw (vector unsigned long long,
26102 vector unsigned long long,
26104 * config/rs6000/rs6000-c: Add support for built-in functions
26105 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
26106 * config/rs6000/altivec.h: Add defintion for vec_sldw.
26107 * doc/extend.texi: Update the built-in documentation for the
26108 new built-in functions.
26110 2017-05-16 Marek Polacek <polacek@redhat.com>
26114 * tree.c (save_expr): Don't fold the expression.
26116 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
26118 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
26119 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
26120 and (?*y,m). Update insn attributes.
26122 2017-05-16 Martin Liska <mliska@suse.cz>
26124 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
26125 flags argument of print_gimple_stmt, print_gimple_expr,
26126 print_generic_stmt and print_generic_expr.
26127 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
26128 * coretypes.h: Likewise.
26129 * except.c (dump_eh_tree): Likewise.
26130 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
26131 * gimple-pretty-print.h: Likewise.
26132 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
26133 (backprop::push_to_worklist): Likewise.
26134 (backprop::pop_from_worklist): Likewise.
26135 (backprop::process_use): Likewise.
26136 (backprop::intersect_uses): Likewise.
26137 (note_replacement): Likewise.
26138 * gimple-ssa-store-merging.c
26139 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
26140 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
26141 (pass_store_merging::execute): Likewise.
26142 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
26143 (ssa_base_cand_dump_callback): Likewise.
26144 (dump_incr_vec): Likewise.
26145 (replace_refs): Likewise.
26146 (replace_mult_candidate): Likewise.
26147 (create_add_on_incoming_edge): Likewise.
26148 (create_phi_basis): Likewise.
26149 (insert_initializers): Likewise.
26150 (all_phi_incrs_profitable): Likewise.
26151 (introduce_cast_before_cand): Likewise.
26152 (replace_one_candidate): Likewise.
26153 * gimplify.c (gimplify_expr): Likewise.
26154 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
26155 (set_rename): Likewise.
26156 (rename_uses): Likewise.
26157 (copy_loop_phi_nodes): Likewise.
26158 (add_close_phis_to_merge_points): Likewise.
26159 (copy_loop_close_phi_args): Likewise.
26160 (copy_cond_phi_args): Likewise.
26161 (graphite_copy_stmts_from_block): Likewise.
26162 (translate_pending_phi_nodes): Likewise.
26163 * graphite-poly.c (print_pdr): Likewise.
26164 (dump_gbb_cases): Likewise.
26165 (dump_gbb_conditions): Likewise.
26166 (print_scop_params): Likewise.
26167 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
26168 (build_cross_bb_scalars_use): Likewise.
26169 (gather_bbs::before_dom_children): Likewise.
26170 * hsa-dump.c (dump_hsa_immed): Likewise.
26171 * ipa-cp.c (print_ipcp_constant_value): Likewise.
26172 (get_replacement_map): Likewise.
26173 * ipa-inline-analysis.c (dump_condition): Likewise.
26174 (estimate_function_body_sizes): Likewise.
26175 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
26176 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
26177 * ipa-prop.c (ipa_dump_param): Likewise.
26178 (ipa_print_node_jump_functions_for_edge): Likewise.
26179 (ipa_modify_call_arguments): Likewise.
26180 (ipa_modify_expr): Likewise.
26181 (ipa_dump_param_adjustments): Likewise.
26182 (ipa_dump_agg_replacement_values): Likewise.
26183 (ipcp_modif_dom_walker::before_dom_children): Likewise.
26184 * ipa-pure-const.c (check_stmt): Likewise.
26185 (pass_nothrow::execute): Likewise.
26186 * ipa-split.c (execute_split_functions): Likewise.
26187 * omp-offload.c (dump_oacc_loop_part): Likewise.
26188 (dump_oacc_loop): Likewise.
26189 * trans-mem.c (tm_log_emit): Likewise.
26190 (tm_memopt_accumulate_memops): Likewise.
26191 (dump_tm_memopt_set): Likewise.
26192 (dump_tm_memopt_transform): Likewise.
26193 * tree-cfg.c (gimple_verify_flow_info): Likewise.
26194 (print_loop): Likewise.
26195 * tree-chkp-opt.c (chkp_print_addr): Likewise.
26196 (chkp_gather_checks_info): Likewise.
26197 (chkp_get_check_result): Likewise.
26198 (chkp_remove_check_if_pass): Likewise.
26199 (chkp_use_outer_bounds_if_possible): Likewise.
26200 (chkp_reduce_bounds_lifetime): Likewise.
26201 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
26202 (chkp_mark_completed_bounds): Likewise.
26203 (chkp_register_incomplete_bounds): Likewise.
26204 (chkp_mark_invalid_bounds): Likewise.
26205 (chkp_maybe_copy_and_register_bounds): Likewise.
26206 (chkp_build_returned_bound): Likewise.
26207 (chkp_get_bound_for_parm): Likewise.
26208 (chkp_build_bndldx): Likewise.
26209 (chkp_get_bounds_by_definition): Likewise.
26210 (chkp_generate_extern_var_bounds): Likewise.
26211 (chkp_get_bounds_for_decl_addr): Likewise.
26212 * tree-chrec.c (chrec_apply): Likewise.
26213 * tree-data-ref.c (dump_data_reference): Likewise.
26214 (dump_subscript): Likewise.
26215 (dump_data_dependence_relation): Likewise.
26216 (analyze_overlapping_iterations): Likewise.
26217 * tree-inline.c (expand_call_inline): Likewise.
26218 (tree_function_versioning): Likewise.
26219 * tree-into-ssa.c (dump_defs_stack): Likewise.
26220 (dump_currdefs): Likewise.
26221 (dump_names_replaced_by): Likewise.
26222 (dump_update_ssa): Likewise.
26223 (update_ssa): Likewise.
26224 * tree-object-size.c (pass_object_sizes::execute): Likewise.
26225 * tree-parloops.c (build_new_reduction): Likewise.
26226 (try_create_reduction_list): Likewise.
26227 (ref_conflicts_with_region): Likewise.
26228 (oacc_entry_exit_ok_1): Likewise.
26229 (oacc_entry_exit_single_gang): Likewise.
26230 * tree-pretty-print.h: Likewise.
26231 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
26232 (get_scalar_evolution): Likewise.
26233 (add_to_evolution): Likewise.
26234 (get_loop_exit_condition): Likewise.
26235 (analyze_evolution_in_loop): Likewise.
26236 (analyze_initial_condition): Likewise.
26237 (analyze_scalar_evolution): Likewise.
26238 (instantiate_scev): Likewise.
26239 (number_of_latch_executions): Likewise.
26240 (gather_chrec_stats): Likewise.
26241 (final_value_replacement_loop): Likewise.
26242 (scev_const_prop): Likewise.
26243 * tree-sra.c (dump_access): Likewise.
26244 (disqualify_candidate): Likewise.
26245 (create_access): Likewise.
26246 (reject): Likewise.
26247 (maybe_add_sra_candidate): Likewise.
26248 (create_access_replacement): Likewise.
26249 (analyze_access_subtree): Likewise.
26250 (analyze_all_variable_accesses): Likewise.
26251 (sra_modify_assign): Likewise.
26252 (initialize_constant_pool_replacements): Likewise.
26253 (find_param_candidates): Likewise.
26254 (decide_one_param_reduction): Likewise.
26255 (replace_removed_params_ssa_names): Likewise.
26256 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
26257 * tree-ssa-copy.c (dump_copy_of): Likewise.
26258 (copy_prop_visit_cond_stmt): Likewise.
26259 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
26260 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
26261 (record_equivalences_from_stmt): Likewise.
26262 * tree-ssa-dse.c (compute_trims): Likewise.
26263 (delete_dead_call): Likewise.
26264 (delete_dead_assignment): Likewise.
26265 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
26266 (forward_propagate_into_cond): Likewise.
26267 (pass_forwprop::execute): Likewise.
26268 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
26269 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
26271 (move_computations_worker): Likewise.
26272 (execute_sm): Likewise.
26273 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
26274 (remove_exits_and_undefined_stmts): Likewise.
26275 (remove_redundant_iv_tests): Likewise.
26276 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
26277 (adjust_iv_update_pos): Likewise.
26278 * tree-ssa-math-opts.c (bswap_replace): Likewise.
26279 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
26280 (value_replacement): Likewise.
26281 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
26282 * tree-ssa-pre.c (print_pre_expr): Likewise.
26283 (get_representative_for): Likewise.
26284 (create_expression_by_pieces): Likewise.
26285 (insert_into_preds_of_block): Likewise.
26286 (eliminate_insert): Likewise.
26287 (eliminate_dom_walker::before_dom_children): Likewise.
26288 (eliminate): Likewise.
26289 (remove_dead_inserted_code): Likewise.
26290 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
26291 * tree-ssa-reassoc.c (get_rank): Likewise.
26292 (eliminate_duplicate_pair): Likewise.
26293 (eliminate_plus_minus_pair): Likewise.
26294 (eliminate_not_pairs): Likewise.
26295 (undistribute_ops_list): Likewise.
26296 (eliminate_redundant_comparison): Likewise.
26297 (update_range_test): Likewise.
26298 (optimize_range_tests_var_bound): Likewise.
26299 (optimize_vec_cond_expr): Likewise.
26300 (rewrite_expr_tree): Likewise.
26301 (rewrite_expr_tree_parallel): Likewise.
26302 (linearize_expr): Likewise.
26303 (break_up_subtract): Likewise.
26304 (linearize_expr_tree): Likewise.
26305 (attempt_builtin_powi): Likewise.
26306 (attempt_builtin_copysign): Likewise.
26307 (transform_stmt_to_copy): Likewise.
26308 (transform_stmt_to_multiply): Likewise.
26309 (dump_ops_vector): Likewise.
26310 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
26311 (print_scc): Likewise.
26312 (set_ssa_val_to): Likewise.
26313 (visit_reference_op_store): Likewise.
26314 (visit_use): Likewise.
26315 (sccvn_dom_walker::before_dom_children): Likewise.
26316 (run_scc_vn): Likewise.
26317 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
26319 (expr_hash_elt::print): Likewise.
26320 (const_and_copies::pop_to_marker): Likewise.
26321 (const_and_copies::record_const_or_copy_raw): Likewise.
26322 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
26323 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
26324 (dump_predicates): Likewise.
26325 (find_uninit_use): Likewise.
26326 (warn_uninitialized_phi): Likewise.
26327 (pass_late_warn_uninitialized::execute): Likewise.
26328 * tree-ssa.c (verify_vssa): Likewise.
26329 (verify_ssa): Likewise.
26330 (maybe_optimize_var): Likewise.
26331 * tree-vrp.c (dump_value_range): Likewise.
26332 (dump_all_value_ranges): Likewise.
26333 (dump_asserts_for): Likewise.
26334 (register_edge_assert_for_2): Likewise.
26335 (vrp_visit_cond_stmt): Likewise.
26336 (vrp_visit_switch_stmt): Likewise.
26337 (vrp_visit_stmt): Likewise.
26338 (vrp_visit_phi_node): Likewise.
26339 (simplify_cond_using_ranges_1): Likewise.
26340 (fold_predicate_in): Likewise.
26341 (evrp_dom_walker::before_dom_children): Likewise.
26342 (evrp_dom_walker::push_value_range): Likewise.
26343 (evrp_dom_walker::pop_value_range): Likewise.
26344 (execute_early_vrp): Likewise.
26346 2017-05-16 Richard Biener <rguenther@suse.de>
26348 * dwarf2out.c (loc_list_from_tree_1): Do not create
26349 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
26351 2017-05-16 Richard Biener <rguenther@suse.de>
26353 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
26355 (note_variable_value_in_expr): If we resolved the decl ref
26356 do not push to the stack.
26358 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
26360 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
26361 operations in fast-math mode.
26362 (vaddq_f16): Likewise.
26363 (vmul_f16): Likewise.
26364 (vmulq_f16): Likewise.
26365 (vsub_f16): Likewise.
26366 (vsubq_f16): Likewise.
26367 * config/arm/neon.md (add<mode>3): New.
26369 (fma:<VH:mode>3): New. Also remove outdated comment.
26372 2017-05-16 Martin Liska <mliska@suse.cz>
26376 * ipa-devirt.c (warn_types_mismatch): Fix typo.
26377 (odr_types_equivalent_p): Likewise.
26379 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
26381 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
26383 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
26386 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
26387 non-interunit SSE move alternatives with '?'.
26388 (zero-extendsidi peephole2): New peephole to skip intermediate
26389 general register in SSE zero-extend sequence.
26391 2017-05-15 Jeff Law <law@redhat.com>
26393 * reorg.c (relax_delay_slots): Create a new variable to hold
26394 the temporary target rather than clobbering TARGET_LABEL.
26396 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
26397 missing argument to extract_bit_field call.
26398 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
26400 2017-05-15 Martin Liska <mliska@suse.cz>
26403 * gcc.c (process_command): Do not allow empty argument of -o option.
26405 2017-05-15 Renlin Li <renlin.li@arm.com>
26407 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
26408 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
26409 * config/aarch64/constraints.md (Usf): Add long call check.
26410 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
26411 (call_value): Likewise.
26412 (sibcall): Likewise.
26413 (sibcall_value): Likewise.
26415 (call_value_insn): New.
26416 (sibcall_insn): Update rtx pattern.
26417 (sibcall_value_insn): Likewise.
26418 (call_internal): Remove.
26419 (call_value_internal): Likewise.
26420 (sibcall_internal): Likewise.
26421 (sibcall_value_internal): Likewise.
26422 (call_reg): Likewise.
26423 (call_symbol): Likewise.
26424 (call_value_reg): Likewise.
26425 (call_value_symbol): Likewise.
26427 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
26430 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
26432 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
26434 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
26435 compatible with CCGOCmode and with CCZmode.
26437 2017-05-14 Martin Sebor <msebor@redhat.com>
26439 PR middle-end/77671
26440 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
26441 (gimple_fold_builtin_snprintf): Same.
26442 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
26443 (gimple_fold_builtin_snprintf): Same.
26444 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
26445 of character types.
26446 (is_call_safe): New function.
26447 (try_substitute_return_value): Call it.
26448 (try_simplify_call): New function.
26449 (pass_sprintf_length::handle_gimple_call): Call it.
26451 2017-05-14 Martin Sebor <msebor@redhat.com>
26453 PR middle-end/80669
26454 * builtins.c (expand_builtin_stpncpy): Simplify.
26456 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
26458 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
26459 * config/i386/i386.h
26460 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
26461 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
26462 (struct machine_function): Add new members call_ms2sysv,
26463 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
26464 (struct machine_frame_state): New fields sp_realigned and
26465 sp_realigned_offset.
26466 * config/i386/i386.c
26467 (enum xlogue_stub): New enum.
26468 (enum xlogue_stub_sets): New enum.
26469 (class xlogue_layout): New class.
26470 (struct ix86_frame): New fields stack_realign_allocate_offset,
26471 stack_realign_offset and outlined_save_offset. Modify comments to
26472 detail stack layout when using out-of-line stubs.
26473 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
26474 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
26475 -mcall-ms2sysv-xlogues.
26476 (stub_managed_regs): New static variable.
26477 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
26478 registers managed by out-of-line stub.
26479 (disable_call_ms2sysv_xlogues): New function.
26480 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
26481 m->call_ms2sysv when appropriate and compute frame layout for
26483 (sp_valid_at, fp_valid_at): New inline functions.
26484 (choose_basereg): New function.
26485 (choose_baseaddr): Add align parameter, use choose_basereg and modify
26487 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
26488 Use align parameter of choose_baseaddr to generated aligned SSE movs
26490 (pro_epilogue_adjust_stack): Modify to track
26491 machine_frame_state::sp_realigned.
26492 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
26493 (ix86_nsaved_sseregs): Likewise.
26494 (ix86_emit_save_regs): Likewise.
26495 (ix86_emit_save_regs_using_mov): Likewise.
26496 (ix86_emit_save_sse_regs_using_mov): Likewise.
26497 (get_scratch_register_on_entry): Likewise.
26498 (gen_frame_set): New function.
26499 (gen_frame_load): Likewise.
26500 (gen_frame_store): Likewise.
26501 (emit_outlined_ms2sysv_save): Likewise.
26502 (emit_outlined_ms2sysv_restore): Likewise.
26503 (ix86_expand_prologue): Modify stack re-alignment code and call
26504 emit_outlined_ms2sysv_save when appropriate.
26505 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
26506 parameter rtx_insn *insn, which allows the function to be used to only
26507 generate the notes.
26508 (ix86_expand_epilogue): Modify validity checks of frame and stack
26509 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
26510 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
26511 * config/i386/predicates.md
26512 (save_multiple): New predicate.
26513 (restore_multiple): Likewise.
26514 * config/i386/sse.md
26515 (save_multiple<mode>): New pattern.
26516 (save_multiple_realign<mode>): Likewise.
26517 (restore_multiple<mode>): Likewise.
26518 (restore_multiple_and_return<mode>): Likewise.
26519 (restore_multiple_leave_return<mode>): Likewise.
26520 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
26522 2017-05-14 Julia Koval <julia.koval@intel.com>
26524 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
26525 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
26526 (__builtin_ia32_xsetbv): New builtins.
26527 * config/i386/i386.c (ix86_expand_special_args_builtin):
26529 (ix86_expand_builtin): Special expand for new intrinsics.
26530 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
26531 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
26532 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
26534 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26536 * cfganal.c (inverted_post_order_compute): Change argument type
26538 * cfganal.h (inverted_post_order_compute): Adjust prototype.
26539 * df-core.c (rest_of_handle_df_initialize): Adjust.
26540 (rest_of_handle_df_finish): Likewise.
26541 (df_analyze_1): Likewise.
26542 (df_analyze): Likewise.
26543 (loop_inverted_post_order_compute): Change argument to be a vec *.
26544 (df_analyze_loop): Adjust.
26545 (df_get_n_blocks): Likewise.
26546 (df_get_postorder): Likewise.
26547 * df.h (struct df_d): Change field to be a vec.
26548 * lcm.c (compute_laterin): Adjust.
26549 (compute_available): Likewise.
26550 * lra-lives.c (lra_create_live_ranges_1): Likewise.
26551 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
26552 * tree-ssa-pre.c (compute_antic): Likewise.
26554 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26556 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
26557 (depth_first_search::depth_first_search): Change structure init
26558 function to this constructor.
26559 (depth_first_search::add_bb): Rename function to this member.
26560 (depth_first_search::execute): Likewise.
26561 (flow_dfs_compute_reverse_finish): Adjust.
26563 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26565 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
26566 (longest_simple_path): Likewise.
26567 * shrink-wrap.c (spread_components): Likewise.
26568 (disqualify_problematic_components): Likewise.
26569 (emit_common_heads_for_components): Likewise.
26570 (emit_common_tails_for_components): Likewise.
26571 (insert_prologue_epilogue_for_components): Likewise.
26573 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26575 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
26578 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26580 * df-core.c (df_set_blocks): Start using auto_bitmap.
26581 (df_compact_blocks): Likewise.
26582 * df-problems.c (df_rd_confluence_n): Likewise.
26583 * df-scan.c (df_insn_rescan_all): Likewise.
26584 (df_process_deferred_rescans): Likewise.
26585 (df_update_entry_block_defs): Likewise.
26586 (df_update_exit_block_uses): Likewise.
26587 (df_entry_block_bitmap_verify): Likewise.
26588 (df_exit_block_bitmap_verify): Likewise.
26589 (df_scan_verify): Likewise.
26590 * lra-constraints.c (lra_constraints): Likewise.
26591 (undo_optional_reloads): Likewise.
26592 (lra_undo_inheritance): Likewise.
26593 * lra-remat.c (calculate_gen_cands): Likewise.
26594 (do_remat): Likewise.
26595 * lra-spills.c (assign_spill_hard_regs): Likewise.
26596 (spill_pseudos): Likewise.
26597 * tree-ssa-pre.c (bitmap_set_and): Likewise.
26598 (bitmap_set_subtract_values): Likewise.
26600 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26602 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
26603 management with auto_bitmap.
26604 (fix_inter_tick): Likewise.
26605 (fix_recovery_deps): Likewise.
26606 * ira.c (add_store_equivs): Likewise.
26607 (find_moveable_pseudos): Likewise.
26608 (split_live_ranges_for_shrink_wrap): Likewise.
26609 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
26610 (rtx_reuse_manager::seen_def_p): Likewise.
26611 (rtx_reuse_manager::set_seen_def): Likewise.
26612 * print-rtl.h (class rtx_reuse_manager): Likewise.
26614 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26616 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
26618 (migrate_btr_def): Likewise.
26619 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
26620 * df-core.c (loop_post_order_compute): Likewise.
26621 (loop_inverted_post_order_compute): Likewise.
26622 * hsa-common.h: Likewise.
26623 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
26624 * init-regs.c (initialize_uninitialized_regs): Likewise.
26625 * ipa-inline.c (resolve_noninline_speculation): Likewise.
26626 (inline_small_functions): Likewise.
26627 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
26628 * ira.c (combine_and_move_insns): Likewise.
26629 (build_insn_chain): Likewise.
26630 * loop-invariant.c (find_invariants): Likewise.
26631 * lower-subreg.c (propagate_pseudo_copies): Likewise.
26632 * predict.c (tree_predict_by_opcode): Likewise.
26633 (predict_paths_leading_to): Likewise.
26634 (predict_paths_leading_to_edge): Likewise.
26635 (estimate_loops_at_level): Likewise.
26636 (estimate_loops): Likewise.
26637 * shrink-wrap.c (try_shrink_wrapping): Likewise.
26638 (spread_components): Likewise.
26639 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
26640 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
26641 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
26642 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
26643 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
26644 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
26645 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
26646 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
26647 (mark_threaded_blocks): Likewise.
26648 (thread_through_all_blocks): Likewise.
26649 * tree-ssa.c (verify_ssa): Likewise.
26650 (execute_update_addresses_taken): Likewise.
26651 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
26653 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26655 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
26657 (post_order_compute): Likewise.
26658 (inverted_post_order_compute): Likewise.
26659 (pre_and_rev_post_order_compute_fn): Likewise.
26661 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26663 * genrecog.c (int_set::int_set): Explicitly construct our
26664 auto_vec base class.
26665 * vec.h (auto_vec::auto_vec): New constructor.
26667 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26669 * bitmap.h (class auto_bitmap): New constructor taking
26670 bitmap_obstack * argument.
26672 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26674 * bitmap.h (class auto_bitmap): Change type of m_bits to
26675 bitmap_head, and adjust ctor / dtor and member operators.
26677 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
26679 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
26680 when returned register mode doesn't match original mode.
26682 2017-05-12 Jeff Law <law@redhat.com>
26683 Jakub Jelinek <jakub@redhat.com>
26685 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
26686 we look for cc setter after the compare-elim changes.
26687 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
26688 within the vector to match what compare-elim now expects.
26689 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
26690 (xorsi3_flags, one_cmplsi2_flags): Likewise.
26692 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
26693 after the compare-elim changes.
26694 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
26695 the vector to match what compare-elim now expects.
26696 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
26697 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
26698 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
26699 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
26700 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
26702 * config/visium/visium.c (single_set_and_flags): Fix where
26703 we look for cc setter after the compare-elim changes.
26704 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
26705 with the vector to match what compare-elim now expects.
26706 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
26707 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
26708 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
26709 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
26710 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
26711 (neg<mode>2_insn_set_overflow): Likewise.
26713 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
26715 PR middle-end/79794
26716 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
26717 maybe_expand_insn call, set ops[0].target. If still set after call,
26718 set alt_rtl. Add extra arg to recursive calls.
26719 (extract_bit_field): Add alt_rtl argument. Pass to
26721 * expmed.h (extract_bit_field): Fix prototype.
26722 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
26723 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
26724 to extract_bit_field_calls.
26725 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
26726 Pass alt_rtl to extract_bit_field calls.
26727 * calls.c (store_unaligned_arguments_into_psuedos)
26728 load_register_parameters): Pass extra NULL to extract_bit_field calls.
26729 * optabs.c (maybe_legitimize_operand): Clear op->target when call
26731 * optabs.h (struct expand_operand): Add target bitfield.
26733 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
26735 * compare-elim.c (try_eliminate_compare): Canonicalize
26736 operation with embedded compare to
26737 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
26738 (set (reg) (operation)].
26740 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
26742 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
26745 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
26746 cost of adding a carry flag for ADC instruction.
26747 [case MINUS]: Ignore the cost of subtracting a carry flag
26748 for SBB instruction.
26750 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
26752 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
26754 * config/rs6000/bmiintrin.h: New file.
26755 * config/rs6000/bmi2intrin.h: New file.
26756 * config/rs6000/x86intrin.h: New file.
26758 2017-05-12 Jeff Law <law@redhat.com>
26760 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
26763 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
26765 PR middle-end/80707
26766 * tree-cfg.c: Remove cfg edges of unreachable case statements.
26768 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26770 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
26771 early expansion of vector divide builtins.
26772 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
26773 builtins identified as having unsigned arguments.
26775 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26777 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
26778 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
26779 expansion of vector logical operations (and, andc, or, xor,
26782 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26784 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
26785 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
26787 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26789 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
26790 early GIMPLE expansion of vector multiplies.
26792 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26794 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
26795 TARGET_HAVE_MOVT conditional.
26796 (movt splitter): Likewise.
26798 2017-05-12 Richard Biener <rguenther@suse.de>
26800 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
26801 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
26802 Fold all stmts not inplace.
26804 2017-05-12 Richard Biener <rguenther@suse.de>
26806 PR tree-optimization/80713
26807 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
26808 inserted_exprs bit for not removed stmts.
26810 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
26812 PR middle-end/69921
26813 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
26814 parallelized" attribute for parallelized OpenACC kernels.
26815 * omp-offload.c (execute_oacc_device_lower): Use it.
26817 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
26818 Set "oacc kernels" attribute.
26819 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
26820 parameter. Adjust all users.
26821 (oacc_fn_attrib_kernels_p): Remove function.
26822 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
26823 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
26824 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
26825 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
26826 assert "oacc kernels" attribute is set.
26828 2017-05-11 Carl Love <cel@us.ibm.com>
26830 * config/rs6000/rs6000-c: Add support for built-in functions
26831 vector unsigned char vec_popcnt (vector signed char)
26832 vector unsigned char vec_popcnt (vector unsigned char)
26833 vector unsigned short vec_popcnt (vector signed short)
26834 vector unsigned short vec_popcnt (vector unsigned short)
26835 vector unsigned int vec_popcnt (vector signed int)
26836 vector unsigned int vec_popcnt (vector unsigned int)
26837 vector unsigned long long vec_popcnt (vector signed long long)
26838 vector unsigned long long vec_popcnt (vector unsigned long long)
26839 vector signed long long vec_slo (vector signed long long,
26840 vector signed char)
26841 vector signed long long vec_slo (vector signed long long,
26842 vector unsigned char)
26843 vector unsigned long long vec_slo (vector unsigned long long,
26844 vector signed char)
26845 vector unsigned long long vec_slo (vector unsigned long long,
26846 vector unsigned char)
26847 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
26848 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
26849 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
26850 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
26851 * doc/extend.texi: Update the built-in documentation file for the
26852 new built-in functions.
26854 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
26856 * attribs.h (sorted_attr_string): Move machine independent
26857 functions for target clone support from the i386 port to common
26858 code. Rename ix86_function_versions to common_function_versions.
26859 Rename make_name to make_unique_name.
26860 (common_function_versions): Likewise.
26861 (make_unique_name): Likewise.
26862 (make_dispatcher_decl): Likewise.
26863 (is_function_default_version): Likewise.
26864 * attribs.c (attr_strcmp): Likewise.
26865 (sorted_attr_string): Likewise.
26866 (common_function_versions): Likewise.
26867 (make_unique_name): Likewise.
26868 (make_dispatcher_decl): Likewise.
26869 (is_function_default_version): Likewise.
26870 * config/i386/i386.c (attr_strcmp): Likewise.
26871 (sorted_attr_string): Likewise.
26872 (ix86_function_versions): Likewise.
26873 (make_name): Likewise.
26874 (make_dispatcher_decl): Likewise.
26875 (is_function_default_version): Likewise.
26876 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
26878 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26881 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
26882 Account for direct move costs for vec_construct of integer
26885 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
26888 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
26889 (UNSPEC_STX_ATOMIC): Ditto.
26890 (loaddi_via_sse): New insn.
26891 (storedi_via_sse): Ditto.
26892 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
26893 Update corresponding peephole2 patterns.
26894 (atomic_storedi_fpu): Ditto.
26896 2017-05-11 Julia Koval <julia.koval@intel.com>
26898 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
26899 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
26901 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
26902 (__builtin_ia32_rsqrt14ss_mask): New builtins.
26903 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
26905 2017-05-11 Nathan Sidwell <nathan@acm.org>
26907 * graphite-poly.c: Include dumpfile.h.
26909 * dumpfle.h (dump_function): Declare here ...
26910 * tree-dump.h (dump_function): ... not here.
26911 * dumpfile.c: #include tree-cfg.h.
26912 (dump_function): Move here from ...
26913 * tree-dump.c (dump_function): ... here.
26914 * gimplify.c: #include splay-tree.h, not tree-dump.h.
26915 * graphite-poly.c: Don't include tree-dump.h.
26916 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
26917 * print-tree.c: Likewise.
26918 * stor-layout.c: Likewise.
26919 * tree-nested.c: Likewise.
26921 * dumpfile.c (dump_start): Use TDF_FLAGS.
26922 (dump_enable_all): Fix TDF_KIND check thinko.
26924 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
26926 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26927 array entries to represent two legal parameterizations of the
26928 overloaded __builtin_cmpb function, as represented by the
26929 P6_OV_BUILTIN_CMPB constant.
26930 (altivec_resolve_overloaded_builtin): Add special case handling
26931 for the __builtin_cmpb function, as represented by the
26932 P6_OV_BUILTIN_CMPB constant.
26933 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
26934 (BU_P6_64BIT_2): New macro.
26935 (BU_P6_OVERLOAD_2): New macro
26936 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
26937 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
26938 (CMPB): Add overload support to represent both 32-bit and 64-bit
26939 compare-bytes function.
26940 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
26941 support for TARGET_CMPB.
26942 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
26943 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26944 documentation of the __builtin_cmpb overloaded built-in function.
26946 2017-05-11 Richard Biener <rguenther@suse.de>
26948 PR tree-optimization/80705
26949 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
26950 bases are not vectorizable.
26952 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26954 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
26955 when counting register pressure.
26957 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26959 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
26960 (iv_ca_more_deps): Renamed to ...
26961 (iv_ca_compare_deps): ... this.
26962 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
26964 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26966 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
26968 (determine_group_iv_costs): ... here.
26969 (find_inv_vars_cb): Record inv var if it's not recorded before.
26971 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26973 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
26974 (get_shiftadd_cost): Ditto.
26976 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26978 * tree-ssa-address.c: Include header file.
26979 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
26981 (add_to_parts): Refactor.
26982 (addr_to_parts): New parameter. Update use of move_hint_to_base.
26983 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
26986 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26988 PR tree-optimization/53090
26989 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
26991 (extract_cond_operands): Detect condition with IV on both sides
26992 and return COMP_IV_EXPR_2.
26993 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
26994 (rewrite_use_compare): Simplify by removing call to function
26995 extract_cond_operands.
26997 2017-05-11 Bin Cheng <bin.cheng@arm.com>
26999 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
27000 (extract_cond_operands): Detect condition comparing against non-
27001 invariant bound and return appropriate enum value.
27002 (find_interesting_uses_cond): Update use of extract_cond_operands.
27003 Handle its return value accordingly.
27004 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
27006 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27008 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
27009 nonlinear iv_use computation in loop invariant sensitive way.
27011 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27013 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
27014 (find_iv_candidates): Call relate_compare_use_with_all_cands.
27016 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27018 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
27019 (dump_cand): Support iv_cand.inv_exprs.
27020 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
27022 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
27025 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27027 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
27029 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
27030 as local function. Include necessary header files.
27031 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
27033 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27035 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
27037 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27039 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
27040 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
27041 RSHIFT_EXPR and BIT_NOT_EXPR.
27043 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27045 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
27046 (adjust_setup_cost): New parameter supporting round up adjustment.
27047 (struct address_cost_data): Delete.
27048 (force_expr_to_var_cost): Don't bound cost with spill_cost.
27049 (split_address_cost, ptr_difference_cost): Delete.
27050 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
27051 (struct ainc_cost_data): New struct.
27052 (get_address_cost_ainc): New function.
27053 (get_address_cost, get_computation_cost): Reimplement.
27054 (determine_group_iv_cost_address): Record inv_expr for all uses of
27056 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
27057 (iv_ca_has_deps): Reimplemented to ...
27058 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
27060 (iv_ca_extend): Call iv_ca_more_deps.
27062 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27064 * tree-ssa-address.c (struct mem_address): Move to header file.
27065 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
27066 * tree-ssa-address.h (struct mem_address): Move from C file.
27067 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
27069 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27071 * tree-affine.h (aff_combination_type): New interface.
27072 (aff_combination_zero_p): Remove static.
27073 (aff_combination_const_p): New interface.
27074 (aff_combination_singleton_var_p): New interfaces.
27076 2017-05-11 Richard Biener <rguenther@suse.de>
27078 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
27079 Skip unreachable blocks and destinations.
27080 (eliminate): Move stmt removal and fixup ...
27081 (fini_eliminate): ... here. Skip inserted exprs.
27082 (pass_pre::execute): Move fini_pre after fini_eliminate.
27083 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
27084 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
27085 PRE to get rid of dead code that has invalid SSA form and
27086 split critical edges again.
27088 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27090 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
27092 2017-05-11 Richard Biener <rguenther@suse.de>
27094 * passes.c (execute_function_todo): Verify loops if they are
27095 said to be up-to-date.
27096 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
27097 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
27099 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
27102 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
27103 handle calling assemble_external ourself.
27106 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
27107 modes with zero size. Enhance comment.
27109 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27111 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
27112 built-ins for vec_xl and vec_xst with short and char pointer
27115 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
27117 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
27118 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
27119 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
27120 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
27121 (_mm_maskz_min_round_ss): New intrinsics.
27122 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
27123 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
27124 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
27125 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
27126 (__builtin_ia32_minss_mask_round): New builtins.
27127 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27128 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27129 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
27131 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
27132 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
27134 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
27137 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
27139 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
27140 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
27141 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
27142 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
27143 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
27144 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
27145 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
27146 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27147 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
27148 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
27149 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
27150 (__builtin_ia32_mulss_mask_round): New builtins.
27151 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27152 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27153 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
27155 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
27156 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
27158 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
27161 2017-05-10 Julia Koval <julia.koval@intel.com>
27163 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
27164 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
27165 (_mm256_setr_m128i): New intrinsics.
27167 2017-05-10 Julia Koval <julia.koval@intel.com>
27169 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
27170 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
27171 (_mm_maskz_rcp14_ss): New intrinsics.
27172 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
27173 (__builtin_ia32_rcp14ss_mask): New builtins.
27174 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
27176 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
27178 PR tree-optimization/51513
27179 * tree-cfg.c (gimple_seq_unreachable_p): New function.
27180 (assert_unreachable_fallthru_edge_p): Use it.
27181 (group_case_labels_stmt): Likewise.
27182 * tree-cfg.h: Prototype it.
27183 * stmt.c: Include cfghooks.h and tree-cfg.h.
27184 (emit_case_dispatch_table) <gap_label>: New local variable.
27185 Use it to fill dispatch table gaps.
27186 Test for default_label before updating probabilities.
27187 (expand_case) <default_label>: Remove unneeded initialization.
27188 Test for unreachable default case statement and remove its edge.
27189 Set default_label accordingly.
27190 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
27192 2017-05-10 Carl Love <cel@us.ibm.com>
27194 * config/rs6000/rs6000-c: Add support for built-in functions
27195 vector signed char vec_neg (vector signed char)
27196 vector signed short int vec_neg (vector short int)
27197 vector signed int vec_neg (vector signed int)
27198 vector signed long long vec_neg (vector signed long long)
27199 vector float vec_neg (vector float)
27200 vector double vec_neg (vector double)
27201 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
27203 * config/rs6000/altivec.h: Add define for vec_neg
27204 * doc/extend.texi: Update the built-in documentation for the
27205 new built-in functions.
27207 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27209 PR tree-optimization/77644
27210 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
27212 2017-05-10 Nathan Sidwell <nathan@acm.org>
27214 * dumpfile.h (TDI_lang_all): New.
27215 (TDF_KIND): New. Renumber others
27216 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
27218 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
27220 (get_dump_file_name): Adjust suffix generation.
27221 (dump_enable_all): Use TDF_KIND.
27222 * doc/invoke.texi (-fdump-lang-all): Document.
27224 * dumpfile.h: Tabify.
27226 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
27229 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
27230 Move member access before delete.
27232 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
27234 * tree-inline.c (expand_call_inline): Split block at stmt
27237 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
27240 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
27241 are now unused after splitting mov{sf,sd}_hardfloat.
27242 (f32_lr2): Likewise.
27243 (f32_lm): Likewise.
27244 (f32_lm2): Likewise.
27245 (f32_li): Likewise.
27246 (f32_li2): Likewise.
27247 (f32_lv): Likewise.
27248 (f32_sr): Likewise.
27249 (f32_sr2): Likewise.
27250 (f32_sm): Likewise.
27251 (f32_sm2): Likewise.
27252 (f32_si): Likewise.
27253 (f32_si2): Likewise.
27254 (f32_sv): Likewise.
27255 (f32_dm): Likewise.
27256 (f32_vsx): Likewise.
27257 (f32_av): Likewise.
27258 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
27259 For movsf, order stores so the VSX stores occur before the GPR
27260 store which encourages the register allocator to use a traditional
27261 FPR instead of a GPR. For movsd, order the stores so that the GPR
27262 store comes before the VSX stores to allow the power6 to work.
27263 This is due to the power6 not having a 32-bit integer store
27264 instruction from a FPR.
27265 (movsf_hardfloat): Likewise.
27266 (movsd_hardfloat): Likewise.
27268 2017-05-09 Martin Sebor <msebor@redhat.com>
27270 PR translation/80280
27271 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
27274 PR translation/80280
27275 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
27276 data member added in r247778.
27277 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
27279 2017-05-09 Nathan Sidwell <nathan@acm.org>
27281 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
27283 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
27286 2017-05-09 Marek Polacek <polacek@redhat.com>
27288 * doc/invoke.texi: Fix typo.
27290 2017-05-09 Richard Biener <rguenther@suse.de>
27292 * tree-vrp.c (vrp_val_is_max): Adjust comment.
27293 (vrp_val_is_min): Likewise.
27294 (set_value_range_to_value): Likewise.
27295 (set_value_range_to_nonnegative): Likewise.
27296 (gimple_assign_nonzero_p): Likewise.
27297 (gimple_stmt_nonzero_p): Likewise.
27298 (vrp_int_const_binop): Likewise. Remove unreachable case.
27299 (adjust_range_with_scev): Adjust comments.
27300 (compare_range_with_value): Likewise.
27301 (extract_range_from_phi_node): Likewise.
27302 (test_for_singularity): Likewise.
27304 2017-05-09 Richard Biener <rguenther@suse.de>
27306 * tree-vrp.c (get_single_symbol): Add assert that we don't
27307 get overflowed constants as invariant part.
27308 (compare_values_warnv): Add comment before the TREE_NO_WARNING
27309 checks. Use wi::cmp instead of recursing for integer constants.
27310 (compare_values): Just ignore whether we assumed undefined
27311 overflow instead of failing the compare.
27312 (extract_range_for_var_from_comparison_expr): Add comment before the
27313 TREE_NO_WARNING sets.
27314 (test_for_singularity): Likewise.
27315 (extract_range_from_comparison): Do not disable optimization
27316 when we assumed undefined overflow.
27317 (extract_range_basic): Remove init of unused var.
27319 2017-05-09 Richard Biener <rguenther@suse.de>
27321 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
27322 (extract_range_from_multiplicative_op_1): Adjust.
27323 (extract_range_from_binary_expr_1): Use int_const_binop.
27325 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
27328 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
27329 rs6000_store_data_bypass_p in seven define_bypass directives and
27330 in several comments.
27331 * config/rs6000/rs6000-protos.h: Add prototype for
27332 rs6000_store_data_bypass_p function.
27333 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
27334 function implements slightly different (rs6000-specific) semantics
27335 than store_data_bypass_p, returning false rather than aborting
27336 with assertion error when arguments do not satisfy the
27337 requirements of store data bypass.
27338 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
27339 rs6000_store_data_bypass_p.
27341 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
27343 * config/xtensa/xtensa-protos.h
27344 (xtensa_initial_elimination_offset): New declaration.
27345 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
27346 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
27347 macro definition, add case for FRAME_POINTER_REGNUM when
27348 FRAME_GROWS_DOWNWARD.
27349 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
27350 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
27351 xtensa_initial_elimination_offset.
27353 2017-05-08 Nathan Sidwell <nathan@acm.org>
27355 * doc/invoke.texi: Alphabetize -fdump options.
27357 2017-05-08 Martin Sebor <msebor@redhat.com>
27359 PR translation/80280
27360 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
27362 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
27364 * target.def (compute_frame_layout): New optional target hook.
27365 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
27366 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
27367 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
27369 * reload1.c (verify_initial_elim_offsets): Likewise.
27370 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
27371 (use_simple_return_p): Call arm_compute_frame_layout if needed.
27372 (arm_get_frame_offsets): Split up into this ...
27373 (arm_compute_frame_layout): ... and this function.
27375 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
27377 * config/aarch64/constraints.md (Usa): New constraint.
27378 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
27380 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
27382 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
27383 with_multilib_list after it has been checked.
27385 2017-05-08 Richard Biener <rguenther@suse.de>
27387 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
27388 (bitmap_set_subtract_values): Likewise.
27390 2017-05-08 Richard Biener <rguenther@suse.de>
27392 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
27393 (gimple_assign_nonzero): ... this and remove strict_overflow_p
27395 (gimple_stmt_nonzero_warnv_p): Rename to ...
27396 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
27398 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
27399 (extract_range_basic): Adjust, do not disable propagation on
27400 strict overflow sensitive simplification.
27401 (vrp_visit_cond_stmt): Likewise.
27403 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
27405 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
27406 body size unconditionally.
27408 2017-05-07 Jeff Law <law@redhat.com>
27411 2017-05-06 Jeff Law <law@redhat.com>
27412 PR tree-optimization/78496
27413 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
27416 PR tree-optimization/78496
27417 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
27418 (simplify_stmt_using_ranges): Call it.
27419 (vrp_dom_walker::before_dom_children): Extract equivalences
27420 from an ASSERT_EXPR with an equality comparison against a
27423 2017-05-06 Jeff Law <law@redhat.com>
27425 PR tree-optimization/78496
27426 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
27429 PR tree-optimization/78496
27430 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
27431 (simplify_stmt_using_ranges): Call it.
27432 (vrp_dom_walker::before_dom_children): Extract equivalences
27433 from an ASSERT_EXPR with an equality comparison against a
27436 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
27438 * lra-constraints.c (lra_copy_reg_equiv): New function.
27439 (split_reg): Use it to copy equivalence information from the
27440 original register to the spill register.
27442 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
27444 PR rtl-optimization/75964
27445 * simplify-rtx.c (simplify_const_relational_operation): Remove
27446 invalid handling of comparisons of integer ABS.
27448 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
27450 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
27451 initialize to zero.
27452 (init_regs): Remove declaration.
27453 (function_arg_advance_32): Initialize error_p as boolean variable.
27455 2017-05-05 Nathan Sidwell <nathan@acm.org>
27457 * store-motion.c (remove_reachable_equiv_notes): Reformat long
27458 lines. Use for (;;).
27460 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27462 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
27463 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
27464 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
27465 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
27466 VF=2 that require versioning.
27468 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27470 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
27473 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27475 * diagnostic.h (diagnostic_override_option_index): Convert from
27476 macro to inline function.
27478 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27480 * diagnostic.c (last_module_changed_p): New function.
27481 (set_last_module): New function.
27482 (diagnostic_report_current_module): Convert macro usage to
27483 the above functions.
27484 * diagnostic.h (diagnostic_context::last_module): Strengthen
27485 from const line_map * to const line_map_ordinary *.
27486 (diagnostic_last_module_changed): Delete macro.
27487 (diagnostic_set_last_module): Delete macro.
27489 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27491 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
27492 with diagnostic_report_diagnostic.
27493 (diagnostic_n_impl_richloc): Likewise.
27494 * diagnostic.h (report_diagnostic): Delete macro.
27495 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
27496 with diagnostic_report_diagnostic.
27497 * substring-locations.c (format_warning_va): Likewise.
27499 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27501 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
27502 save/restor of format_spec. Move option-printing code to...
27503 (print_option_information): ...this new function, and
27504 reimplement by simply printing to the pretty_printer,
27505 rather than appending to the format string.
27507 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27509 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
27510 handling logic into...
27511 (update_effective_level_from_pragmas): ...this new function.
27513 2017-05-04 Andrew Waterman <andrew@sifive.com>
27515 * config/riscv/riscv.opt (mstrict-align): New option.
27516 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
27517 (SLOW_UNALIGNED_ACCESS): Define.
27518 (riscv_slow_unaligned_access): Declare.
27519 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
27521 (riscv_slow_unaligned_access): New variable.
27522 (rocket_tune_info): Set slow_unaligned_access to true.
27523 (optimize_size_tune_info): Set slow_unaligned_access to false.
27524 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
27525 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
27526 (riscv_option_override): Set riscv_slow_unaligned_access.
27527 * doc/invoke.texi: Add -mstrict-align to RISC-V.
27529 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
27531 * config/riscv/riscv.md: Unify indentation.
27533 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27538 * config/rs6000/rs6000.md (u code attribute): Add FIX and
27540 (extendsi<mode>2): Add support for doing sign extension via
27541 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
27542 don't have ISA 3.0 instructions.
27543 (extendsi<mode>2 splitter): Likewise.
27544 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
27545 generate the normal insns since SImode can now go in vector
27546 registers. Disallow the special UNSPECs needed for previous
27547 machines to hide SImode being used. Add new insns
27548 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
27549 (fix_trunc<mode>si2_stfiwx): Likewise.
27550 (fix_trunc<mode>si2_internal): Likewise.
27551 (fixuns_trunc<mode>si2): Likewise.
27552 (fixuns_trunc<mode>si2_stfiwx): Likewise.
27553 (fctiw<u>z_<mode>_smallint): Likewise.
27554 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
27555 of floating point to 32-bit integer from doing a direct move to
27556 the GPR registers to do a store.
27557 (fctiwz_<mode>): Break long line.
27559 2017-05-05 Bin Cheng <bin.cheng@arm.com>
27561 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
27562 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
27563 (addr_list, addr_offset_valid_p): New.
27564 (split_address_groups): Check offset validity with above function.
27565 (gt-tree-ssa-loop-ivopts.h): Include header file.
27567 2017-05-05 Nathan Sidwell <nathan@acm.org>
27569 * config.gcc (arm*-*-*): Add missing 'fi'.
27571 2017-05-05 Steve Ellcey <sellcey@cavium.com>
27573 * doc/invoke.texi (-fopt-info): Explicitly say order of options
27574 included in -fopt-info does not matter.
27575 * doc/optinfo.texi (-fopt-info): Fix description of default
27576 behavour. Explicitly say order of options included in -fopt-info
27579 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
27581 * config.gcc: Allow combinations of aprofile and rmprofile values for
27582 --with-multilib-list.
27583 * config/arm/t-multilib: New file.
27584 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
27585 variables. Remove setting of ISA and floating-point ABI in
27586 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
27587 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
27588 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
27589 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
27591 * config/arm/t-rmprofile: Likewise except for the matches changes.
27592 * doc/install.texi (--with-multilib-list): Document the combination of
27593 aprofile and rmprofile values and warn about pitfalls in doing that.
27595 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
27597 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
27598 (movdi_aarch64): Likewise.
27600 2017-05-05 Jakub Jelinek <jakub@redhat.com>
27602 PR tree-optimization/80632
27603 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
27605 (build_arrays): Initialize it for virtual phis.
27606 (fix_phi_nodes): Use it for virtual phis.
27608 PR tree-optimization/80558
27609 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
27610 [x, y] op z into [x op, y op z] for op & or | if conditions
27613 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
27614 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27617 * config/arm/arm.md (use_literal_pool): Remove.
27618 (64-bit immediate split): No longer takes cost into consideration
27619 if arm_disable_literal_pool is enabled.
27620 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
27621 used when arm_disable_literal_pool is enabled.
27622 (arm_max_const_double_inline_cost): Remove use of
27623 arm_disable_literal_pool.
27624 (push_minipool_fix): Add assert.
27625 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
27626 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
27627 (no_literal_pool_sf_immediate): New.
27629 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27631 PR tree-optimization/80613
27632 * tree-ssa-dce.c (propagate_necessity): Remove cases for
27633 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
27635 2017-05-05 Richard Biener <rguenther@suse.de>
27637 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
27639 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
27641 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
27642 of this flag from insn conditions due to removal from r247495.
27644 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
27646 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
27648 (arm_early_store_addr_dep_ptr): Likewise.
27649 * config/arm/aarch-common-protos.h
27650 (arm_early_load_addr_dep_ptr): Add prototype.
27651 (arm_early_store_addr_dep_ptr): Likewise.
27652 * config/arm/cortex-a53.md: Add new bypasses.
27654 2017-05-05 Jakub Jelinek <jakub@redhat.com>
27656 * tree.c (next_type_uid): Change type to unsigned.
27657 (type_hash_canon): Decrement back next_type_uid if
27658 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
27659 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
27662 2017-05-04 Martin Sebor <msebor@redhat.com>
27664 * builtins.c: Fix a trivial typo in a comment.
27666 PR middle-end/79234
27667 * builtins.c (check_sizes): Adjust to handle reading past the end.
27668 Avoid printing excessive upper bound of ranges. Use %E to print
27669 tree nodes instead of converting them to %wu.
27670 (expand_builtin_memchr): New function.
27671 (compute_dest_size): Rename...
27672 (compute_objsize): ...to this.
27673 (expand_builtin_memcpy): Adjust.
27674 (expand_builtin_mempcpy): Adjust.
27675 (expand_builtin_strcat): Adjust.
27676 (expand_builtin_strcpy): Adjust.
27677 (check_strncat_sizes): Adjust.
27678 (expand_builtin_strncat): Adjust.
27679 (expand_builtin_strncpy): Adjust and simplify.
27680 (expand_builtin_memset): Adjust.
27681 (expand_builtin_bzero): Adjust.
27682 (expand_builtin_memcmp): Adjust.
27683 (expand_builtin): Handle memcmp.
27684 (maybe_emit_chk_warning): Check strncat just once.
27686 2017-05-04 Martin Sebor <msebor@redhat.com>
27688 PR preprocessor/79214
27689 PR middle-end/79222
27690 PR middle-end/79223
27691 * builtins.c (check_sizes): Add inlining context and issue
27692 warnings even when -Wno-system-headers is set.
27693 (check_strncat_sizes): Same.
27694 (expand_builtin_strncat): Same.
27695 (expand_builtin_memmove): New function.
27696 (expand_builtin_stpncpy): Same.
27697 (expand_builtin): Handle memmove and stpncpy.
27699 2017-05-04 Bin Cheng <bin.cheng@arm.com>
27701 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
27702 which is not used any more.
27704 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27706 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
27708 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27710 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
27711 (cortexa53_tunings): Likewise.
27712 (cortexa57_tunings): Likewise.
27713 (cortexa72_tunings): Likewise.
27714 (cortexa73_tunings): Likewise.
27716 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27718 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
27719 Set loop alignment to 8.
27721 2017-05-04 Martin Sebor <msebor@redhat.com>
27723 PR translation/80280
27724 * builtins.c (expand_builtin_object_size): Add missing quoting to
27725 %D and like directives.
27726 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
27727 (hsa_type_for_tree_type): Same.
27728 (verify_function_arguments): Same.
27729 * symtab.c (symbol_table::change_decl_assembler_name): Same.
27730 * varasm.c (get_section): Same.
27733 2017-05-04 Martin Sebor <msebor@redhat.com>
27735 PR translation/80280
27736 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
27738 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27740 * config/aarch64/aarch64.c (generic_addrcost_table):
27741 Change HI/TI mode setting.
27743 2017-05-04 Martin Jambor <mjambor@suse.cz>
27745 PR tree-optimization/80622
27746 * tree-sra.c (comes_initialized_p): New function.
27747 (build_accesses_from_assign): Only set write lazily when
27748 comes_initialized_p is false.
27749 (analyze_access_subtree): Use comes_initialized_p.
27750 (propagate_subaccesses_across_link): Assert !comes_initialized_p
27751 instead of testing for PARM_DECL.
27753 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27755 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
27756 constraint on operand 0 to allow more general addressing modes.
27757 Adjust output template.
27758 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
27760 * config/aarch64/aarch64-protos.h
27761 (aarch64_address_valid_for_prefetch_p): Declare prototype.
27762 * config/aarch64/constraints.md (Dp): New address constraint.
27763 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
27766 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
27768 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
27769 update use of estimate_ipcp_clone_size_and_time.
27770 (estimate_local_effects): Update use of
27771 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
27772 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
27773 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
27774 Return nonspecialized time.
27776 2017-05-04 Richard Biener <rguenther@suse.de>
27778 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
27779 for the last VUSE which def dominates the PHI. Directly call
27781 (get_continuation_for_phi_1): Remove.
27783 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
27785 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
27786 to explain the use of truncating division. Cap the number of
27787 iterations to the maximum given by nb_iterations_upper_bound,
27790 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
27792 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
27793 * configure: Regenerate.
27794 * config.in: Regenerate.
27795 * config/i386/driver-mingw32.c: new file.
27796 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
27797 * config.host: Link driver-mingw32.o on MinGW host.
27798 * doc/install.texi: Document new --enable-mingw-wildcard configure
27801 2017-05-04 Marek Polacek <polacek@redhat.com>
27803 PR tree-optimization/80612
27804 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
27806 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27807 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
27809 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
27810 (movt splitter): Likewise.
27811 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
27812 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
27813 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
27814 block for Thumb-1 with MOVT.
27815 (thumb2_legitimate_address_p): Move code block ...
27816 (can_avoid_literal_pool_for_label_p): ... into this new function.
27817 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
27819 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
27820 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
27821 "M-profile targets with the MOVT instruction".
27823 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27825 * config/arm/arm-builtins.c (arm_init_builtins): Rename
27826 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
27827 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
27829 2017-05-04 Martin Liska <mliska@suse.cz>
27831 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
27832 variable cond_code.
27834 2017-05-04 Richard Biener <rguenther@suse.de>
27836 * tree.c (array_at_struct_end_p): Handle arrays at struct
27837 end with flexarrays more conservatively. Refactor and treat
27838 arrays of arrays or aggregates more strict. Fix
27839 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
27840 * tree.c (array_at_struct_end_p): Adjust prototype.
27841 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
27842 * gimple-fold.c (get_range_strlen): Likewise.
27843 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
27845 2017-05-04 Richard Biener <rguenther@suse.de>
27847 PR tree-optimization/31130
27848 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
27850 (supports_overflow_infinity): Likewise.
27851 (is_negative_overflow_infinity): Likewise.
27852 (is_positive_overflow_infinity): Likewise.
27853 (is_overflow_infinity): Likewise.
27854 (stmt_overflow_infinity): Likewise.
27855 (overflow_infinity_range_p): Likewise.
27856 (usable_range_p): Remove as always returning true.
27857 (make_overflow_infinity): Remove.
27858 (negative_overflow_infinity): Likewise.
27859 (positive_overflow_infinity): Likewise.
27860 (avoid_overflow_infinity): Likewise.
27861 (set_value_range): Adjust accordingly.
27862 (set_value_range_to_nonnegative): Likewise, remove now unused
27863 overflow_infinity arg.
27864 (vrp_operand_equal_p): Adjust.
27865 (update_value_range): Likewise.
27866 (range_int_cst_singleton_p): Likewise.
27867 (operand_less_p): Likewise.
27868 (compare_values_warnv): Likewise.
27869 (extract_range_for_var_from_comparison_expr): Likewise.
27870 (vrp_int_const_binop): Likewise.
27871 (zero_nonzero_bits_from_vr): Likewise.
27872 (extract_range_from_multiplicative_op_1): Likewise.
27873 (extract_range_from_binary_expr_1): Likewise.
27874 (extract_range_from_unary_expr): Likewise.
27875 (extract_range_from_comparison): Likewise.
27876 (extract_range_basic): Likewise.
27877 (adjust_range_with_scev): Likewise.
27878 (compare_ranges): Likewise.
27879 (compare_range_with_value): Likewise.
27880 (dump_value_range): Likewise.
27881 (test_for_singularity): Likewise, remove strict_overflow_p parameter
27883 (simplify_cond_using_ranges): Adjust.
27885 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
27887 * brig-builtins.def: Added a builtin for class_f64.
27888 * builtin-types.def: Added a builtin type needed by class_f64.
27890 2017-05-03 Jason Merrill <jason@redhat.com>
27892 * timevar.def: Add TV_CONSTEXPR.
27894 2017-05-03 David Malcolm <dmalcolm@redhat.com>
27896 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
27898 2017-05-03 Martin Jambor <mjambor@suse.cz>
27900 * ipa-prop.c (ipa_update_after_lto_read): Removed.
27901 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
27902 * ipa-cp.c (ipcp_propagate_stage): Do not call
27903 ipa_update_after_lto_read.
27904 * ipa-inline.c (ipa_inline): Likewise.
27906 2017-05-03 Martin Jambor <mjambor@suse.cz>
27908 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
27909 tag. Added a default constructor and a destructor.
27910 (ipa_edge_args_sum_t): New class;
27911 (ipa_edge_args_sum): Declare.
27912 (ipa_edge_args_vector): Remove declaration.
27913 (IPA_EDGE_REF): Use ipa_edge_args_sum.
27914 (ipa_free_edge_args_substructures): Remove declaration.
27915 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
27916 (ipa_edge_args_info_available_for_edge_p): Likewise.
27917 * ipa-prop.c (ipa_edge_args_vector): Removed.
27918 (edge_removal_hook_holder): Likewise.
27919 (edge_duplication_hook_holder): Likewise.
27920 (ipa_edge_args_sum): New variable.
27921 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
27922 ipa_edge_args_vector.
27923 (ipa_free_edge_args_substructures): Likewise.
27924 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
27925 ipa_edge_args_vector.
27926 (ipa_edge_removal_hook): Turned into method
27927 ipa_edge_args_sum_t::remove.
27928 (ipa_edge_duplication_hook): Turned into method
27929 ipa_edge_args_sum_t::duplicate.
27930 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
27931 registering edge hooks.
27932 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
27933 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
27934 ipa_edge_args_sum instead of ipa_edge_args_vector.
27935 * ipa-profile.c (ipa_profile): Likewise.
27937 2017-05-03 Martin Jambor <mjambor@suse.cz>
27939 * symbol-summary.h (function_summary): New method exists.
27940 (function_summary::symtab_removal): Deallocate through release.
27941 (call_summary): New class.
27942 (gt_ggc_mx): New overload.
27943 (gt_pch_nx): Likewise.
27944 (gt_pch_nx): Likewise.
27946 2017-05-03 Jeff Law <law@redhat.com>
27948 PR tree-optimization/78496
27949 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
27950 from simplify_cond_using_ranges. Split off code to walk
27951 backwards through casts into ...
27952 (simplify_cond_using_ranges_2): New function.
27953 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
27954 (execute_vrp): After identifying jump threads, call
27955 simplify_cond_using_ranges_2.
27957 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
27960 * ipa-inline.h (inline_summary): Add ctor.
27961 (create_ggc): Do not use ggc_cleared_alloc.
27963 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
27964 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27966 * gcc.c (handle_braces): Support escaping in switch matching
27968 * doc/invoke.texi (Spec Files): Document it.
27969 Remove superfluous @code markup in items.
27971 2017-05-03 David Malcolm <dmalcolm@redhat.com>
27973 * diagnostic-show-locus.c (struct column_range): New struct.
27974 (get_affected_columns): New function.
27975 (get_printed_columns): New function.
27976 (struct correction): New struct.
27977 (correction::ensure_capacity): New function.
27978 (correction::ensure_terminated): New function.
27979 (struct line_corrections): New struct.
27980 (line_corrections::~line_corrections): New dtor.
27981 (line_corrections::add_hint): New function.
27982 (layout::print_trailing_fixits): Reimplement in terms of the new
27984 (selftest::test_overlapped_fixit_printing): New function.
27985 (selftest::diagnostic_show_locus_c_tests): Call it.
27987 2017-05-03 Nathan Sidwell <nathan@acm.org>
27989 Canonicalize canonical type hashing
27990 * tree.h (type_hash_canon_hash): Declare.
27991 * tree.c (type_hash_list, attribute_hash_list): Move into
27992 type_hash_canon_hash.
27993 (build_type_attribute_qual_variant): Break out hash code calc into
27994 type_hash_canon_hash.
27995 (type_hash_canon_hash): New. Generic type hash computation.
27996 (build_range_type_1, build_array_type_1, build_function_type,
27997 build_method_type_directly, build_offset_type, build_complex_type,
27998 make_vector_type): Call it.
28000 2017-05-03 Richard Biener <rguenther@suse.de>
28002 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
28003 When all DRs have unknown misaligned do not always peel
28004 when there is a store but apply the same costing model as if
28005 there were only loads.
28007 2017-05-03 Richard Biener <rguenther@suse.de>
28010 PR tree-optimization/80492
28011 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
28012 compare_base_decls returning dont-know properly.
28014 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
28016 * config/arm/iterators.md (CCSI): New mode iterator.
28017 (arch): New mode attribute.
28018 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
28019 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
28020 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
28021 code iterator for success result mode.
28022 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
28023 the corresponding new insn generators.
28025 2017-05-03 Bin Cheng <bin.cheng@arm.com>
28028 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28029 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28031 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
28033 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
28034 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
28035 (DDR_A): Wrap DDR argument in brackets.
28036 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
28037 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
28038 (DDR_REVERSED_P): Likewise.
28040 2017-05-03 Jakub Jelinek <jakub@redhat.com>
28042 PR tree-optimization/79472
28043 * tree-switch-conversion.c (struct switch_conv_info): Add
28044 contiguous_range and default_case_nonstandard fields.
28045 (collect_switch_conv_info): Compute contiguous_range and
28046 default_case_nonstandard fields, don't clear final_bb if
28047 contiguous_range and only the default case doesn't have the required
28049 (check_all_empty_except_final): Set default_case_nonstandard instead
28050 of failing if contiguous_range and the default case doesn't have empty
28052 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
28053 and only the default case doesn't have the required constants. Skip
28055 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
28056 if default_case_nonstandard.
28057 (build_constructors): Build constant 1 just once. Assert that default
28058 values aren't inserted in between cases if contiguous_range. Skip
28060 (build_arrays): Skip virtual phis.
28061 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
28062 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
28063 Handle virtual phis.
28064 (gen_inbound_check): Handle default_case_nonstandard case.
28065 (process_switch): Adjust check_final_bb caller. Call
28066 gather_default_values with the first non-default case instead of
28067 default case if default_case_nonstandard.
28069 2017-05-02 Nathan Sidwell <nathan@acm.org>
28071 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
28072 check. Fix formatting.
28074 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
28076 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
28077 errors when comparing specialized and unspecialized times.
28079 2017-05-02 David Malcolm <dmalcolm@redhat.com>
28081 * diagnostic-show-locus.c
28082 (layout::should_print_annotation_line_p): Make private.
28083 (layout::print_annotation_line): Make private.
28084 (layout::annotation_line_showed_range_p): Make private.
28085 (layout::show_ruler): Make private.
28086 (layout::print_source_line): Make private. Pass in line and
28087 line_width, rather than calling location_get_source_line. Drop
28089 (layout::print_leading_fixits): New method.
28090 (layout::print_any_fixits): Rename to...
28091 (layout::print_trailing_fixits): ...this, and make private.
28092 Don't print newline fixits.
28093 (diagnostic_show_locus): Move logic for printing one row into...
28094 (layout::print_line): ...this new function. Move the
28095 location_get_source_line call and error-handling from
28096 print_source_line to here. Call print_leading_fixits, and rename
28097 print_any_fixits to print_trailing_fixits.
28098 (selftest::test_fixit_insert_containing_newline): Update now that
28099 newlines are partially supported.
28100 (selftest::test_fixit_insert_containing_newline_2): New test.
28101 (selftest::test_fixit_replace_containing_newline): Update comments.
28102 (selftest::diagnostic_show_locus_c_tests): Call the new test.
28103 * edit-context.c (class added_line): New class.
28104 (class edited_line): Describe newline handling in comment.
28105 (edited_line::actually_edited_p): New method.
28106 (edited_line::print_content): Delete redundant decl.
28107 (edited_line::m_predecessors): New field.
28108 (edited_file::print_content): Call edited_line::print_content.
28109 (edited_file::print_diff): Update to support newlines.
28110 (edited_file::print_diff_hunk): Likewise.
28111 (edited_file::print_run_of_changed_lines): New function.
28112 (edited_file::print_diff_line): Convert to...
28113 (print_diff_line): ...this.
28114 (edited_file::get_effective_line_count): New function.
28115 (edited_line::edited_line): Initialize new field m_predecessors.
28116 (edited_line::~edited_line): Clean up m_predecessors.
28117 (edited_line::apply_fixit): Handle newlines.
28118 (edited_line::get_effective_line_count): New function.
28119 (edited_line::print_content): New function.
28120 (edited_line::print_diff_lines): New function.
28121 (selftest::test_applying_fixits_insert_containing_newline): New
28123 (selftest::test_applying_fixits_replace_containing_newline): New
28125 (selftest::insert_line): New function.
28126 (selftest::test_applying_fixits_multiple_lines): Add example of
28128 (selftest::edit_context_c_tests): Call the new tests.
28130 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28132 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
28133 parameter cand. Update dump information.
28134 (get_computation_cost): Update uses.
28136 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28138 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
28139 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
28140 (get_computation_at, rewrite_use_address): Update use of
28141 get_computation_aff.
28143 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28145 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
28146 (get_computation): Delete.
28147 (get_computation_cost): Implement like get_computation_cost_at.
28148 Use get_computation_at.
28149 (get_computation_cost_at): Delete.
28150 (rewrite_use_nonlinear_expr): Use get_computation_at.
28151 (rewrite_use_compare, remove_unused_ivs): Ditto.
28153 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28155 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
28157 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28159 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
28160 (ivopts_global_cost_for_size): Rename parameter and update uses.
28161 (iv_ca_recount_cost): Update uses.
28162 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
28163 candidates seperately in n_invs and n_cands.
28164 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
28166 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28168 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
28169 (find_inv_vars_cb): New.
28170 (find_depends): Renamed to ...
28171 (find_inv_vars): ... this.
28172 (add_candidate_1, force_var_cost): Call find_inv_vars.
28173 (split_address_cost, determine_group_iv_cost_cond): Ditto.
28175 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28177 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
28178 inv_vars. Add inv_exprs.
28179 (struct iv_cand): Rename depends_on to inv_vars.
28180 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
28181 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
28182 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
28183 (dump_cand): Dump inv_vars.
28184 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
28185 (record_invariant, find_depends, add_candidate_1): Ditto.
28186 (set_group_iv_cost, force_var_cost): Ditto.
28187 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
28188 (get_computation_cost_at, get_computation_cost): Ditto.
28189 (determine_group_iv_cost_generic): Ditto.
28190 (determine_group_iv_cost_address): Ditto.
28191 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
28192 (determine_group_iv_costs): Ditto.
28193 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
28194 (iv_ca_set_remove_invariants): Renamed to ...
28195 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
28196 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
28197 (iv_ca_set_add_invariants): Renamed to ...
28198 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
28199 (iv_ca_set_cp): Use iv_ca_set_add_invs.
28200 (iv_ca_has_deps): Support inv_vars and inv_exprs.
28201 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
28202 (create_new_ivs): Remove useless dump.
28204 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28206 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
28208 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
28209 (iv_ca_set_no_cp, create_new_iv): Ditto.
28211 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28213 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28215 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
28217 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
28218 function tree_check2.
28220 2017-05-02 Martin Liska <mliska@suse.cz>
28222 * doc/gcov.texi: Add missing preposition.
28223 * gcov.c (function_info::function_info): Properly fill up
28224 all member variables.
28226 2017-05-02 Tamar Christina <tamar.christina@arm.com>
28228 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
28230 2017-05-02 Tamar Christina <tamar.christina@arm.com>
28232 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
28234 2017-05-02 Martin Liska <mliska@suse.cz>
28237 * lto-streamer-in.c (lto_read_tree_1): Remove
28238 LTO_STREAMER_DEBUG.
28239 * lto-streamer.c (struct tree_hash_entry): Likewise.
28240 (struct tree_entry_hasher): Likewise.
28241 (tree_entry_hasher::hash): Likewise.
28242 (tree_entry_hasher::equal): Likewise.
28243 (lto_streamer_init): Likewise.
28244 (lto_orig_address_map): Likewise.
28245 (lto_orig_address_get): Likewise.
28246 (lto_orig_address_remove): Likewise.
28247 * lto-streamer.h: Likewise.
28248 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
28249 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
28251 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
28253 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
28254 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
28255 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
28256 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
28257 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
28258 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
28259 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
28260 (mm_maskz_sub_ss): New intrinsics.
28261 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28262 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
28263 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
28264 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
28265 (__builtin_ia32_subss_mask_round): New builtins.
28266 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28267 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28268 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
28270 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
28271 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
28273 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
28276 2017-05-02 Martin Jambor <mjambor@suse.cz>
28278 PR tree-optimization/78687
28279 * tree-sra.c (access): New field parent.
28280 (process_subtree_disqualification): New function.
28281 (disqualify_candidate): Call it.
28282 (build_accesses_from_assign): Reset write flag if creating an
28284 (build_access_subtree): Fill in parent field and also prpagate
28285 down grp_write flag.
28286 (create_artificial_child_access): New parameter set_grp_write, set
28287 grp_write to its value.
28288 (propagate_subaccesses_across_link): Also propagate grp_write flag
28290 (propagate_all_subaccesses): Push the closest parent back to work
28291 queue if add_access_to_work_queue returned true.
28293 2017-05-02 Richard Biener <rguenther@suse.de>
28295 * common.opt (fstrict-overflow): Alias negative to fwrapv.
28296 * doc/invoke.texi (fstrict-overflow): Remove all traces of
28297 -fstrict-overflow documentation.
28298 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
28299 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
28300 flag_strict_overflow.
28301 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
28302 * lto-opts.c (lto_write_options): Do not stream it.
28303 * lto-wrapper.c (merge_and_complain): Do not handle it.
28304 * opts.c (default_options_table): Do not set -fstrict-overflow.
28305 (finish_options): Likewise do not clear it when sanitizing.
28306 * simplify-rtx.c (simplify_const_relational_operation): Do not
28307 test flag_strict_overflow.
28309 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
28311 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
28312 using enabled attribute.
28313 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
28314 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
28315 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
28316 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
28317 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
28318 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
28319 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
28320 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
28321 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
28322 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
28324 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
28326 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
28328 2017-05-02 Richard Biener <rguenther@suse.de>
28330 PR tree-optimization/80591
28332 2017-04-10 Richard Biener <rguenther@suse.de>
28334 * tree-ssa-structalias.c (find_func_aliases): Properly handle
28337 2017-05-02 Richard Biener <rguenther@suse.de>
28339 PR tree-optimization/80549
28340 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
28341 (cleanup_tree_cfg_noloop): Create forwarders to known loop
28342 headers if they do not have a preheader.
28344 2017-05-02 Martin Liska <mliska@suse.cz>
28347 * common.opt: Fix typo.
28348 * doc/invoke.texi: Likewise.
28350 2017-05-01 Jan Beulich <jbeulich@suse.com>
28352 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
28353 swapping, add (x,x,m,x,n) alternative.
28355 2017-05-01 Nathan Sidwell <nathan@acm.org>
28357 * calls.c (combine_pending_stack_adjustment_and_call): Remove
28358 unnecessary unadjusted_alignment check.
28360 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
28363 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
28365 * gimplify.c (gimplify_cilk_detach): New function.
28366 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
28367 * tree-core.h: Document EXPR_CILK_SPAWN.
28368 * tree.h (EXPR_CILK_SPAWN): Define.
28370 2017-05-01 David Malcolm <dmalcolm@redhat.com>
28372 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
28373 to use new fixit_hint representation, using the "replace" logic.
28374 (get_line_span_for_fixit_hint): Likewise.
28375 (layout::print_any_fixits): Likewise.
28376 (selftest::test_one_liner_many_fixits): Rename to...
28377 (selftest::test_one_liner_many_fixits_1): ...this, and update
28378 comment and expected output to reflect that the multiple fix-it
28379 hints are now consolidated into one insertion.
28380 (selftest::test_one_liner_many_fixits_2): New test.
28381 (selftest::test_diagnostic_show_locus_one_liner): Update for
28383 (selftest::test_fixit_consolidation): Update for fix-it API
28385 * diagnostic.c (print_parseable_fixits): Likewise.
28386 * edit-context.c (edited_line::m_line_events): Convert from
28387 auto_vec <line_event *> to auto_vec <line_event>.
28388 (class line_event): Convert from abstract base class to a concrete
28389 class, taking over the role of replace_event.
28390 (class insert_event): Delete.
28391 (class replace_event): Rename to class line_event. Convert to
28393 (edit_context::add_fixits): Reimplement.
28394 (edit_context::apply_insert): Delete.
28395 (edit_context::apply_replace): Rename to...
28396 (edit_context::apply_fixit): ...this. Convert to half-open range.
28397 (edited_file::apply_insert): Delete.
28398 (edited_file::apply_replace): Rename to...
28399 (edited_file::apply_fixit): ...this.
28400 (edited_line::~edited_line): Drop deletion of events.
28401 (edited_line::apply_insert): Delete.
28402 (edited_line::apply_replace): Rename to...
28403 (edited_line::apply_fixit): ...this. Convert to half-open range.
28404 Update for change to type of m_line_events.
28405 * edit-context.h (edit_context::apply_insert): Delete.
28406 (edit_context::apply_replace): Rename to...
28407 (edit_context::apply_fixit): ...this.
28409 2017-05-01 Martin Sebor <msebor@redhat.com>
28411 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
28414 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
28417 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
28418 __get_cpuid_max returns 0.
28419 (__get_cpuid_count): Ditto.
28421 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
28423 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
28424 replacement expression is another instance of one of its arguments.
28426 2017-05-01 Jakub Jelinek <jakub@redhat.com>
28429 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
28430 check for stack push/pop autoinc.
28431 * config/i386/i386.c (ix86_agi_dependent): Return false
28432 if the only reason why modified_in_p returned true is that
28433 addr is SP based and set_insn is a push or pop.
28435 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
28437 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
28440 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
28443 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
28444 (account_size_time): Use two predicates - exec_pred and
28446 (evaluate_conditions_for_known_args): Compute both clause and
28448 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
28449 (inline_summary_t::duplicate): Update.
28450 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
28452 (compute_inline_parameters): Likewise.
28453 (estimate_edge_size_and_time): Update caluclation of time.
28454 (estimate_node_size_and_time): Compute both time and nonspecialized
28456 (estimate_ipcp_clone_size_and_time): Update.
28457 (inline_merge_summary): Update.
28458 (do_estimate_edge_time): Update.
28459 (do_estimate_edge_size): Update.
28460 (do_estimate_edge_hints): Update.
28461 (inline_read_section, inline_write_summary): Stream both new predicates.
28462 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
28464 (compute_inlined_call_time): Cleanup.
28465 (big_speedup_p): Update.
28466 (edge_badness): Update.
28467 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
28468 (size_time_entry): Replace predicate by exec_predicate and
28469 nonconst_predicate.
28470 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
28471 (estimate_edge_time): Return also nonspec_time.
28472 (reset_edge_growth_cache): Update.
28474 2017-04-29 Jakub Jelinek <jakub@redhat.com>
28476 PR rtl-optimization/80491
28477 * ifcvt.c (noce_process_if_block): When looking for x setter
28478 with missing else_bb, don't check only the insn right before
28479 cond_earliest, but look for the last insn that x is modified in
28480 within the same bb.
28482 PR rtl-optimization/80491
28483 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
28485 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
28487 PR tree-optimization/80487
28488 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
28490 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28492 PR tree-optimization/79697
28493 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
28494 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
28495 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
28497 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
28498 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
28500 2017-04-28 Martin Sebor <msebor@redhat.com>
28502 PR tree-optimization/80523
28503 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
28504 (init_target_to_host_charmap, target_to_host, target_strtol10): New
28506 (maybe_warn, format_directive, parse_directive): Use new functions.
28507 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
28509 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
28511 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
28513 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
28515 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
28516 target_header_dir): Set correctly.
28517 * configure: Regenerated.
28518 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
28519 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
28520 instead of SYSTEM_HEADER_DIR.
28522 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
28524 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
28525 (estimate_local_effects): Likewise.
28526 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
28527 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
28528 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
28529 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
28530 do_estimate_edge_time, estimate_edge_time): Likewise.
28531 * ipa-inline-analysis.c (estimate_node_size_and_time,
28532 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
28533 (estimate_time_after_inlining): Remove.
28535 2017-04-28 Martin Liska <mliska@suse.cz>
28537 * doc/gcov.texi: Enhance documentation of gcov.
28539 2017-04-28 Martin Liska <mliska@suse.cz>
28541 * doc/gcov.texi: Sort options in alphabetic order.
28542 * doc/gcov-dump.texi: Likewise.
28543 * doc/gcov-tool.texi: Likewise.
28544 * gcov.c (print_usage): Likewise.
28545 * gcov-dump.c (print_usage): Likewise.
28546 * gcov-tool.c (print_merge_usage_message): Likewise.
28547 (print_rewrite_usage_message): Likewise.
28548 (print_overlap_usage_message): Likewise.
28550 2017-04-28 Martin Liska <mliska@suse.cz>
28552 PR gcov-profile/53915
28553 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
28555 2017-04-28 Martin Liska <mliska@suse.cz>
28557 PR gcov-profile/79891
28558 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
28559 is marked by compiler as living on a line.
28560 (get_cycles_count): Remove usage of the union.
28561 (output_intermediate_file): Likewise.
28562 (find_source): Fix GNU coding style.
28563 (accumulate_line_counts): Remove old non-all block mode.
28564 (output_lines): Remove usage of the union.
28565 * profile.c (output_location): Include all BBs, even if
28566 belonging to a same line (and file) as a previous BB.
28568 2017-04-28 Martin Liska <mliska@suse.cz>
28570 * gcov.c (process_args): Handle new argument 'w'.
28571 (read_graph_file): Assign ID to BBs.
28572 (output_branch_count): Display BB # if verbose flag is set.
28573 (output_lines): Likewise for arcs.
28574 (print_usage): Add '--verbose' option help.
28575 * doc/gcov.texi: Document --verbose (-w) option.
28577 2017-04-28 Martin Liska <mliska@suse.cz>
28579 * gcov.c (struct block_location_info): New struct.
28580 (process_file): Fill up the new structure.
28581 (read_graph_file): Replace usage of encoding by the newly added
28583 (add_line_counts): Likewise.
28584 (accumulate_line_counts): Remove usage of the union.
28585 (function_info::function_info): New function.
28586 (function_info::~function_info): Likewise.
28587 (process_file): Call delete instead of release_function.
28588 (release_function): Release the function.
28589 (release_structures): Call delete instead of release_function.
28590 (solve_flow_graph): Replace usage of num_blocks.
28591 (find_exception_blocks): Likewise.
28592 (output_lines): Fix GNU coding style.
28594 2017-04-28 Martin Liska <mliska@suse.cz>
28597 * coverage.c (coverage_remove_note_file): New function.
28598 * coverage.h: Declare the function.
28599 * toplev.c (finalize): Clean if an error has been seen.
28601 2017-04-28 Martin Liska <mliska@suse.cz>
28603 PR gcov-profile/80031
28604 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
28605 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
28606 * gcov.c (read_graph_file): Read just number of blocks.
28607 * profile.c (branch_prob): Do not stream 0 flags per a basic
28610 2017-04-28 Martin Liska <mliska@suse.cz>
28612 * gcov-dump.c (tag_*): Add new argument to declarations.
28613 (dump_gcov_file): Likewise.
28614 (tag_blocks): Add and use new argument depth.
28615 (tag_arcs): Likewise.
28616 (tag_lines): Likewise.
28617 (tag_counters): Likewise.
28618 (tag_summary): Likewise.
28619 (dump_working_sets): Use depth to do a proper indentation.
28621 2017-04-28 Jakub Jelinek <jakub@redhat.com>
28624 * cgraph.h (symtab_node::debug_symtab): No longer inline.
28625 * symtab.c (symtab_node::debug_symtab): Move definition here.
28627 2017-04-28 Richard Biener <rguenther@suse.de>
28629 * lto-streamer.h (LTO_major_version): Bump to 7.
28631 2017-04-28 Richard Biener <rguenther@suse.de>
28633 * tree-vrp.c (assert_info): New struct.
28634 (add_assert_info): New helper.
28635 (register_edge_assert_for_2): Refactor to add asserts to a vector
28637 (register_edge_assert_for_1): Likewise.
28638 (register_edge_assert_for): Likewise.
28639 (finish_register_edge_assert_for): New helper actually registering
28640 asserts where live on edge.
28641 (find_conditional_asserts): Adjust.
28642 (find_switch_asserts): Likewise.
28643 (evrp_dom_walker::try_find_new_range): Generalize.
28644 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
28646 2017-04-27 Marek Polacek <polacek@redhat.com>
28649 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
28650 arg10 and arg11 to itype.
28652 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
28654 * doc/extend.texi (Object Size Checking): Improve grammar.
28656 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
28659 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
28660 that the logic for permitting reciprocal estimates matches that
28663 2017-04-27 Jakub Jelinek <jakub@redhat.com>
28666 * tree.c (type_cache_hasher::equal): Only compare
28667 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
28668 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
28669 non-aggregate element types.
28670 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
28671 about the flag on ARRAY_TYPEs in the comment, formatting fix.
28673 2017-04-27 Richard Biener <rguenther@suse.de>
28675 PR middle-end/80533
28676 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
28677 stripping ARRAY_REFs from MEM_EXPR make sure we're not
28678 keeping a reference to a trailing array.
28680 2017-04-27 Richard Biener <rguenther@suse.de>
28682 PR middle-end/80539
28683 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
28684 being in loop-closed SSA form conservatively.
28685 (chrec_fold_multiply_poly_poly): Likewise.
28687 2017-04-27 Tamar Christina <tamar.christina@arm.com>
28689 PR middle-end/79665
28690 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
28691 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
28693 2017-04-27 Jakub Jelinek <jakub@redhat.com>
28696 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
28697 (aarch64_function_arg_alignment): Return unsigned int again, but still
28698 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
28699 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
28700 Don't emit -Wpsabi note.
28701 (aarch64_function_arg_boundary): Likewise.
28702 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
28705 2017-04-26 Nathan Sidwell <nathan@acm.org>
28707 * tree.h (crc32_unsigned_n): Declare.
28708 (crc32_unsigned, crc32_unsigned): Make inline.
28709 * tree.c (crc32_unsigned_bits): Replace with ...
28710 (crc32_unsigned_n): ... this.
28711 (crc32_unsigned, crc32_byte): Remove.
28712 (crc32_string): Remove unnecessary braces.
28714 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
28716 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
28717 * ipa-inline-analysis.c (MAX_TIME): Remove.
28718 (account_size_time): Use sreal for time.
28719 (dump_inline_summary): Update.
28720 (estimate_function_body_sizes): Update.
28721 (estimate_edge_size_and_time): Update.
28722 (estimate_calls_size_and_time): Update.
28723 (estimate_node_size_and_time): Update.
28724 (inline_merge_summary): Update.
28725 (inline_update_overall_summary): Update.
28726 (estimate_time_after_inlining): Update.
28727 (inline_read_section): Update.
28728 (inline_write_summary): Update.
28729 * ipa-inline.c (compute_uninlined_call_time): Update.
28730 (compute_inlined_call_time): Update.
28731 (recursive_inlining): Update.
28732 (inline_small_functions): Update.
28733 (dump_overall_stats): Update.
28734 * ipa-inline.h: Include sreal.h.
28735 (size_time_entry): Turn time to sreal.
28736 (inline_summary): Turn self_time nad time to sreal.
28738 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
28740 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
28742 (sreal::stream_out, sreal::stream_in): New.
28743 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
28745 2017-04-25 Jakub Jelinek <jakub@redhat.com>
28747 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
28750 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
28753 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
28755 (dimode_scalar_chain::compute_convert_gain): Ditto.
28756 (dimode_scalar_chain::make_vector_copies): Ditto.
28757 (dimode_scalar_chain::convert_reg): Ditto.
28758 (dimode_scalar_chain::convert_insn): Ditto.
28759 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
28760 (VI248_AVX512BW_1): New mode iterator.
28761 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
28762 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
28765 2017-04-25 Martin Sebor <msebor@redhat.com>
28767 PR tree-optimization/80497
28768 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
28769 constants are representable in HOST_WIDE_INT.
28770 (parse_directive): Ditto.
28772 2017-04-25 Martin Sebor <msebor@redhat.com>
28775 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
28776 (new_zero_array): Adjust signature.
28777 (dom_info::dom_init): Used unsigned rather that size_t.
28778 (dom_info::dom_info): Same.
28780 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28781 Jakub Jelinek <jakub@redhat.com>
28784 * config/arm/arm.c: Include gimple.h.
28785 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
28786 returns negative, increment ncrn only if it returned positive.
28787 (arm_needs_doubleword_align): Return int instead of bool,
28788 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
28789 members, but if there is any such non-FIELD_DECL
28790 > PARM_BOUNDARY aligned decl, return -1 instead of false.
28791 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
28792 returns negative, increment nregs only if it returned positive.
28793 (arm_setup_incoming_varargs): Likewise.
28794 (arm_function_arg_boundary): Emit -Wpsabi note if
28795 arm_needs_doubleword_align returns negative, return
28796 DOUBLEWORD_ALIGNMENT only if it returned positive.
28798 2017-04-25 Marek Polacek <polacek@redhat.com>
28801 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
28802 first argument to type.
28804 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
28807 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
28808 type checks to test for compatibility instead of equality.
28810 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28811 Jakub Jelinek <jakub@redhat.com>
28814 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
28816 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
28817 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
28818 the alignment computation, but return their maximum in warn_alignment.
28819 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
28820 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
28822 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
28823 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
28826 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28828 * config/arc/simdext.md (dmpyh): Fix typo.
28830 2017-04-25 Richard Biener <rguenther@suse.de>
28832 PR tree-optimization/80492
28833 * alias.c (compare_base_decls): Handle registers with asm
28834 specification conservatively.
28835 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
28836 compare_base_decls returning dont-know properly.
28838 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28840 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
28841 (legitimate_offset_address_p): New function.
28842 (arc_legitimate_address_p): Use above function.
28844 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28846 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
28848 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28850 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
28851 ACCH registers whenever they are available.
28853 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28855 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
28856 double regs fix when not used.
28858 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28860 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
28862 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
28863 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
28865 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28867 * config/arc/arc.c (arc_output_addsi): Check for h-register class
28868 when emitting short ADD instructions.
28870 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28872 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
28874 (cmpsi_cc_c_insn): Likewise.
28875 (cbranchsi4_scratch): Compute proper instruction length using
28876 compact_hreg_operand.
28877 * config/arc/predicates.md (compact_hreg_operand): New predicate.
28879 2017-04-25 Richard Biener <rguenther@suse.de>
28881 PR middle-end/80509
28882 * passes.c (pass_manager::pass_manager): Initialize
28883 m_name_to_pass_map.
28885 2017-04-25 Richard Biener <rguenther@suse.de>
28887 PR tree-optimization/79201
28888 * tree-ssa-sink.c (statement_sink_location): Handle calls.
28890 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28893 * config/s390/vector.md: Split MEM->GPR vector moves for
28894 non-s_operand addresses.
28896 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28899 * config/s390/predicates.md (reload_const_wide_int_operand): New
28901 * config/s390/s390.md ("movti"): Remove d/P alternative.
28902 ("movti_bigconst"): New pattern definition.
28904 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
28907 * s390-protos.h (s390_expand_cs_hqi): Removed.
28908 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
28909 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
28910 modes as well as CCZ1mode and CCZmode.
28911 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
28912 signature of s390_emit_compare_and_swap.
28913 (s390_expand_cs_hqi): Likewise, make static.
28914 (s390_expand_cs_tdsi): Generate an explicit compare before trying
28915 compare-and-swap, in some cases.
28916 (s390_expand_cs): Wrapper function.
28917 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
28919 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
28920 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
28921 patterns for small and large integers. Forbid symref memory operands.
28922 Move expander to s390.c. Require cc register.
28923 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
28924 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
28925 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
28926 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
28927 symref memory operands. Remove CC mode and call s390_match_ccmode
28929 ("atomic_exchange<mode>"): Allow and implement all integer modes.
28931 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
28933 * config/s390/s390.md (define_peephole2): New peephole to help
28934 combining the load-and-test pattern with volatile memory.
28936 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
28938 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
28939 with CCZmode for TARGET_Z196.
28941 2017-04-25 Jakub Jelinek <jakub@redhat.com>
28943 PR rtl-optimization/80501
28944 * combine.c (make_compound_operation_int): Set subreg_code to SET
28945 even for AND with mask of the sign bit of mode.
28947 PR rtl-optimization/80500
28948 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
28949 sum's initial value.
28951 2017-04-25 Julian Brown <julian@codesourcery.com>
28952 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28954 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
28956 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
28958 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
28960 2017-04-25 Julian Brown <julian@codesourcery.com>
28961 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28963 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
28964 (thunderx2t99_sha): New Reservation.
28966 2017-04-25 Julian Brown <julian@codesourcery.com>
28967 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28969 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
28970 type for 1-element load.
28972 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
28974 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
28976 2017-04-24 Martin Jambor <mjambor@suse.cz>
28978 PR tree-optimization/80293
28979 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
28980 char arrays not totally scalarizable if it is false.
28981 (analyze_all_variable_accesses): Pass correct value in the new
28982 parameter. Add a statistics counter.
28984 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
28986 PR middle-end/79931
28987 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
28989 2017-04-24 Richard Biener <rguenther@suse.de>
28991 PR tree-optimization/80494
28992 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
28993 out for complex types.
28995 2017-04-24 Richard Biener <rguenther@suse.de>
28997 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
28998 * tree-ssa-sccvn.c (print_scc): Print SCC size.
28999 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
29000 (DFS): Adjust and never fail.
29001 (sccvn_dom_walker::fail): Remove.
29002 (sccvn_dom_walker::before_dom_children): Adjust.
29003 (run_scc_vn): Likewise and never fail.
29004 * tree-ssa-pre.c (pass_pre::execute): Adjust.
29005 (pass_fre::execute): Likewise.
29007 2017-04-24 Richard Biener <rguenther@suse.de>
29009 PR tree-optimization/79725
29010 * tree-ssa-sink.c (statement_sink_location): Return whether
29011 failure reason was zero uses. Move that check later.
29012 (sink_code_in_bb): Deal with zero uses by removing the stmt
29015 2017-04-24 Richard Biener <rguenther@suse.de>
29018 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
29019 pointer-based references.
29021 2017-04-24 Richard Biener <rguenther@suse.de>
29024 * pass_manager.h (pass_manager::operator new): Remove.
29025 (pass_manager::operator delete): Likewise.
29026 * passes.c (pass_manager::operator new): Remove.
29027 (pass_manager::operator delete): Likewise.
29028 (pass_manager::pass_manager): Zero individual pass members.
29030 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
29033 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
29034 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
29035 Check "XEXP (src, 1)" operand here.
29036 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
29037 Check "XEXP (src, 1)" operand here.
29038 (dimode_scalar_chain::make_vector_copies): Detect count register
29039 of a shift instruction. Zero extend count register from QImode
29040 to DImode to satisfy vector shift pattern count operand predicate.
29041 Substitute vector shift count operand with a DImode copy.
29042 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
29045 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
29047 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
29048 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
29049 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
29050 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
29051 (UNSPEC_NOREX_MEM): Remove definition.
29053 2017-04-21 Richard Biener <rguenther@suse.de>
29055 PR tree-optimization/79547
29056 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
29057 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
29058 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
29059 without any constraints.
29061 2017-04-21 Richard Biener <rguenther@suse.de>
29063 PR tree-optimization/78847
29064 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
29066 2017-04-21 Richard Biener <rguenther@suse.de>
29068 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
29069 (build_distinct_type_copy): Likewise.
29070 (build_variant_type_copy): Likewise.
29071 * tree.c (build_qualified_type): Pass down mem-stat info.
29072 (build_distinct_type_copy): Likewise.
29073 (build_variant_type_copy): Likewise.
29075 2017-04-21 Richard Biener <rguenther@suse.de>
29077 PR tree-optimization/80237
29078 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
29080 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
29081 for a simplified result.
29083 2016-04-21 Richard Biener <rguenther@suse.de>
29085 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
29086 sth as strict as a simple_iv but a chrec without symbols and an
29087 operand defined in the loop we are peeling (and not some subloop).
29088 (propagate_constants_for_unrolling): Propagate all constants.
29090 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
29093 * config/i386/i386.c (print_reg): Remove assert for disalowed
29094 regno values, call output_operand_lossage instead.
29096 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
29099 * config/i386/constraints.md (Yc): New register constraint.
29100 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
29101 Use Yc constraint for alternative 2 of operand 0. Remove
29102 preferred_for_speed attribute.
29104 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
29106 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
29107 lastprivate clauses in SIMT case.
29109 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
29111 * doc/invoke.texi (-Wextra-semi): Document new warning option.
29113 2017-04-20 Richard Biener <rguenther@suse.de>
29115 PR tree-optimization/57796
29116 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
29117 as N scalar stores.
29118 (vect_model_load_cost): Cost gathers as N scalar loads.
29120 2017-04-20 Richard Biener <rguenther@suse.de>
29122 * ggc-page.c (ggc_allocated_p): Rename to ...
29123 (safe_lookup_page_table_entry): ... this and return the lookup
29125 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
29127 2017-04-20 Richard Biener <rguenther@suse.de>
29129 PR tree-optimization/80453
29130 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
29131 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
29132 from the conditions.
29133 (vn_phi_eq): Pass them down.
29134 (vn_phi_lookup): Record them.
29135 (vn_phi_insert): Likewise.
29137 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
29139 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
29140 uninitialized variable warning to avoid buffer overrun.
29142 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
29145 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
29146 is suppressed for '{ 0 }' in C.
29148 2017-04-20 Jakub Jelinek <jakub@redhat.com>
29150 * BASE-VER: Set to 8.0.0.
29152 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
29154 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
29155 priority .init_array and .fini_array section with SECTION_NOTYPE
29158 2017-04-20 Jakub Jelinek <jakub@redhat.com>
29160 PR middle-end/80423
29161 * tree.h (build_array_type): Add typeless_storage default argument.
29162 * tree.c (type_cache_hasher::equal): Also compare
29163 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
29164 (build_array_type): Add typeless_storage argument, set
29165 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
29167 (build_nonshared_array_type): Adjust build_array_type_1 caller.
29168 (build_array_type): Likewise. Add typeless_storage argument.
29170 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
29171 Jakub Jelinek <jakub@redhat.com>
29173 PR tree-optimization/80426
29174 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
29175 operation on symbolic operands, also compute the overflow for the
29176 invariant part when the operation degenerates into a negation.
29178 2017-04-19 Jakub Jelinek <jakub@redhat.com>
29181 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
29182 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
29185 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
29187 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
29190 * config/avr/avr.c (tree.h): Include it.
29191 (cgraph.h): Include it.
29192 (avr_encode_section_info): Don't warn for uninitialized progmem
29193 variable if it's just an alias.
29195 2017-04-19 Richard Biener <rguenther@suse.de>
29198 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
29199 when needed by AutoPGO.
29201 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
29204 * doc/lto.texi: Remove an extra 'that'.
29206 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
29208 PR rtl-optimization/80429
29209 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
29210 are only used in debug insns.
29212 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
29213 Vladimir Makarov <vmakarov@redhat.com>
29215 * config/sparc/predicates.md (input_operand): Add comment. Return
29216 true for any memory operand when LRA is in progress.
29217 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
29219 2017-04-18 Jeff Law <law@redhat.com>
29222 * mips.md ({return,simple_return}_internal): Do not overwrite
29225 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29227 PR tree-optimization/80443
29228 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
29229 instead of adding 1, subtract -1 and similarly instead of subtracting
29232 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
29234 PR rtl-optimization/80357
29235 * haifa-sched.c (tmp_bitmap): New variable.
29236 (model_recompute): Handle duplicate use records.
29237 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
29238 (free_global_sched_pressure_data): Free it.
29240 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
29243 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29244 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29245 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29246 instead of SYSTEM_HEADER_DIR.
29248 2017-04-18 Jeff Law <law@redhat.com>
29250 PR middle-end/80422
29251 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
29252 predecessors after walking up the insn chain.
29254 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29257 * dwarf2out.c (modified_type_die): Try harder not to emit internal
29258 sizetype type into debug info.
29260 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29263 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
29264 unneeded test for TARGET_UPPER_REGS_SF.
29265 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
29267 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29270 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
29271 instead of gsi_after_labels.
29273 2017-04-18 Jeff Law <law@redhat.com>
29275 * regcprop.c (maybe_mode_change): Avoid creating copies of the
29279 2017-04-13 Jeff Law <law@redhat.com>
29280 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
29281 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
29283 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
29286 * config/avr/avr.c (intl.h): Include it.
29287 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
29289 2017-04-18 Martin Liska <mliska@suse.cz>
29291 PR gcov-profile/78783
29292 * gcov-tool.c (gcov_output_files): Validate that destination
29293 file is either removed by the tool or by a user.
29295 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
29296 Guy Benyei <guybe@mellanox.com>
29298 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
29299 block, and do not negate it, the stored id is already negative.
29301 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
29303 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
29305 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29308 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
29309 masks of options that should be turned off if the VSX vector
29310 options are turned off.
29311 (OTHER_P8_VECTOR_MASKS): Likewise.
29312 (OTHER_VSX_VECTOR_MASKS): Likewise.
29313 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
29314 rs6000_disable_incompatible_switches to validate no type switches
29316 (rs6000_incompatible_switch): New function to disallow turning on
29317 other vector options if -mno-vsx, -mno-power8-vector, or
29318 -mno-power9-vector are specified.
29320 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
29322 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
29324 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
29326 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
29327 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
29328 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
29329 (ARG_POINTER_CFA_OFFSET): Likewise.
29331 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
29333 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
29334 conditions to take advantage of various optimizations.
29336 2017-04-13 Jeff Law <law@redhat.com>
29338 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
29339 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
29340 (zero_extendsidi2_dext): Likewise.
29342 2017-04-13 Jakub Jelinek <jakub@redhat.com>
29345 * fold-const.c (fold_ternary_loc): Revert
29346 use op0 instead of fold_convert_loc (loc, type, arg0) part of
29349 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
29351 PR rtl-optimization/80343
29352 * lra-remat.c (update_scratch_ops): Assign original hard reg to
29353 new scratch pseudo.
29355 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
29358 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
29359 to ubsan_encode_value.
29361 2017-04-13 Jeff Law <law@redhat.com>
29363 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
29364 appearing in DEBUG_INSNs.
29366 2017-04-13 Martin Liska <mliska@suse.cz>
29368 PR gcov-profile/80413
29369 * gcov-io.c (gcov_write_string): Copy to buffer just when
29370 allocated size is greater than zero.
29372 2017-04-13 Jakub Jelinek <jakub@redhat.com>
29375 * dwarf2out.c (decls_for_scope): Ignore declarations of
29376 current_function_decl in BLOCK_NONLOCALIZED_VARS.
29378 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
29381 * ipa-visibility.c (non_local_p): Fix typos.
29382 (localize_node): When localizing symbol in same comdat group,
29383 dissolve the group only when we know external symbols are going
29385 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
29387 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29389 PR tree-optimization/79390
29390 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
29391 order does not result in usable sequence, retry with reversed operand
29397 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
29398 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
29399 op0 instead of fold_convert_loc (loc, type, arg0).
29401 2017-04-12 Jeff Law <law@redhat.com>
29403 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
29404 has a delay slot in the generated code.
29406 * config/cris/cris.md (cris_preferred_reload_class): Return
29407 GENNONACR_REGS rather than GENERAL_REGS.
29409 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29412 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
29413 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
29414 signedness of the result type.
29416 2017-04-12 Richard Biener <rguenther@suse.de>
29417 Jeff Law <law@redhat.com>
29419 PR tree-optimization/80359
29420 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
29421 trim stores to TARGET_MEM_REFs.
29423 2017-04-12 Richard Biener <rguenther@suse.de>
29425 PR tree-optimization/79390
29426 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
29427 threading case even more.
29429 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
29432 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
29433 for quad_address_p for TImode, instead of just not indexed_address.
29435 2017-04-12 Richard Biener <rguenther@suse.de>
29436 Bernd Edlinger <bernd.edlinger@hotmail.de>
29438 PR middle-end/79671
29439 * alias.c (component_uses_parent_alias_set_from): Handle
29440 TYPE_TYPELESS_STORAGE.
29441 (get_alias_set): Likewise.
29442 * tree-core.h (tree_type_common): Add typeless_storage flag.
29443 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
29444 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
29445 for types containing members with TYPE_TYPELESS_STORAGE.
29446 (place_field): Likewise.
29447 (layout_type): Likewise for ARRAY_TYPE.
29448 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
29449 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
29450 TYPE_TYPELESS_STORAGE.
29451 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
29453 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29456 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
29457 first argument to type.
29459 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29463 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
29464 CONST0_RTX (mode) rather than const0_rtx where appropriate.
29465 (rs6000_expand_binop_builtin): Likewise.
29466 (rs6000_expand_ternop_builtin): Likewise; also add missing
29467 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
29468 vshasigma built-ins.
29469 * doc/extend.texi: Document that vec_xxpermdi's third argument
29470 must be a constant.
29472 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
29474 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
29475 Use shift_const cost parameter when calculating gain of STV shifts.
29477 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
29479 PR rtl-optimization/70478
29480 * lra-constraints.c (process_alt_operands): Check memory for
29481 disfavoring memory insn operand.
29483 2017-04-11 Jakub Jelinek <jakub@redhat.com>
29485 PR middle-end/80100
29486 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
29487 left shift in unsigned HOST_WIDE_INT type.
29489 PR rtl-optimization/80385
29490 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
29491 (not (neg X)) into (plus X -1) for complex or non-integral modes.
29494 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
29495 if they have any depend clauses.
29497 2017-04-11 Martin Liska <mliska@suse.cz>
29500 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
29501 * ipa-split.c (split_function): Create a local comdat symbol
29502 if caller is in a comdat group.
29504 2017-04-11 Martin Liska <mliska@suse.cz>
29507 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
29510 2017-04-11 Martin Sebor <msebor@redhat.com>
29512 PR middle-end/80364
29513 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
29514 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
29516 (directive::set_width, directive::set_precision, format_character):
29518 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
29521 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
29524 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
29525 conflict, set target->arch_name instead of target->cpu_name.
29527 2017-04-11 Richard Biener <rguenther@suse.de>
29529 PR tree-optimization/80374
29530 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
29531 build_zero_cst, remove fold_convertible_p check again.
29533 2017-04-11 Martin Liska <mliska@suse.cz>
29536 * ubsan.c (instrument_object_size): Do not instrument register
29539 2017-04-11 Jakub Jelinek <jakub@redhat.com>
29542 * config/i386/i386-builtin-types.def
29543 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
29544 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
29545 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
29546 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
29547 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
29548 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
29549 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
29550 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
29551 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
29552 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
29553 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
29554 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
29555 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
29556 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
29557 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
29558 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
29559 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
29560 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
29561 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
29562 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
29563 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
29564 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
29565 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
29566 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
29567 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
29568 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
29569 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
29570 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
29571 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
29572 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
29573 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
29574 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
29575 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
29576 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
29577 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
29578 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
29579 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
29580 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
29581 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
29582 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
29583 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
29584 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
29585 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
29586 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
29587 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
29588 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
29590 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
29591 flag to second_arg_count, handle 4 argument function type _COUNT
29592 aliases, handle second_arg_count on second argument rather than last.
29594 2017-04-10 Jeff Law <law@redhat.com>
29596 PR tree-optimization/80374
29597 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
29598 record anything if we can not convert integer_zero_node to the
29601 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29604 * config/rs6000/rs6000.c (rs6000_option_override_internal):
29605 Enhance special handling given to the TARGET_P9_MINMAX option in
29606 relation to certain other options.
29608 2017-04-10 Bin Cheng <bin.cheng@arm.com>
29610 PR tree-optimization/80153
29611 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
29612 remove POINTER_PLUS_EXPR's base part directly, rather than through
29615 2017-04-10 Richard Biener <rguenther@suse.de>
29616 Bin Cheng <bin.cheng@arm.com>
29618 PR tree-optimization/80153
29619 * tree-affine.c (aff_combination_to_tree): Get base pointer from
29620 the first element of pointer type aff_tree. Build result expr in
29622 (add_elt_to_tree): Convert to type unconditionally. Remove other
29623 fold_convert calls.
29624 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
29625 (rewrite_use_nonlinear_expr): Check invariant using iv information.
29627 2017-04-10 Richard Biener <rguenther@suse.de>
29629 * tree-ssa-structalias.c (find_func_aliases): Properly handle
29632 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
29634 PR rtl-optimization/70478
29635 * lra-constraints.c (curr_small_class_check): New.
29636 (update_and_check_small_class_inputs): New.
29637 (process_alt_operands): Update curr_small_class_check. Disfavor
29638 alternative insn memory operands. Check available regs for small
29641 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
29644 * config/mips/mips.opt (-mvirt): Update description.
29645 * doc/invoke.texi (-mvirt): Likewise.
29647 2017-04-10 Richard Biener <rguenther@suse.de>
29649 PR middle-end/80362
29650 * fold-const.c (fold_binary_loc): Look at unstripped ops when
29651 looking for NEGATE_EXPR in -A / -B to A / B folding.
29653 2017-04-10 Martin Liska <mliska@suse.cz>
29655 PR gcov-profile/80224
29656 * gcov.c (print_usage): Fix usage string.
29657 (get_gcov_intermediate_filename): Remove.
29658 (output_gcov_file): Use both for normal and intermediate format.
29659 (generate_results): Do not initialize special file for
29660 intermediate format.
29662 2017-04-10 Richard Biener <rguenther@suse.de>
29664 PR tree-optimization/80304
29665 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
29668 2017-04-10 Nathan Sidwell <nathan@acm.org>
29671 * config/rs6000/rs6000.c (rs6000_vector_type): New.
29672 (rs6000_init_builtins): Use it.
29674 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29676 * config/arm/arm.md (<mrc>): Add mode to SET source.
29677 (<mrrc>): Likewise.
29679 2017-04-10 Richard Biener <rguenther@suse.de>
29681 PR middle-end/80344
29682 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
29684 2017-04-10 Jakub Jelinek <jakub@redhat.com>
29687 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
29688 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
29689 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
29690 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
29691 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
29692 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
29693 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
29694 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
29695 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
29696 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
29697 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
29698 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
29699 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
29700 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
29701 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
29702 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
29703 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
29704 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
29705 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
29706 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
29707 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
29708 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
29709 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
29711 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
29713 PR rtl-optimization/70478
29714 * lra-constraints.c: Reverse the last patch.
29716 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
29718 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
29719 Add comment for WCHAR_T.
29721 2017-04-08 Martin Liska <mliska@suse.cz>
29724 2017-04-07 Martin Liska <mliska@suse.cz>
29727 * ipa-split.c (split_function): Add function part to a same comdat
29730 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29733 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
29735 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
29737 * rs6000/rs6000.c (vec_load_pendulum): Rename...
29738 (vec_pairing): ...to this.
29739 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
29740 (rs6000_sched_init): Adjust for name change.
29741 (struct rs6000_sched_context): Likewise.
29742 (rs6000_init_sched_context): Likewise.
29743 (rs6000_set_sched_context): Likewise.
29745 2017-04-07 Jakub Jelinek <jakub@redhat.com>
29751 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
29753 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
29754 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
29755 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
29757 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
29759 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
29761 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
29763 PR rtl-optimization/70703
29764 * ira-color.c (update_conflict_hard_regno_costs): Use
29765 int64_t instead of HOST_WIDE_INT.
29767 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
29769 PR rtl-optimization/70478
29770 * lra-constraints.c (process_alt_operands): Disfavor alternative
29771 insn memory operands.
29773 2017-04-07 Jeff Law <law@redhat.com>
29775 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
29776 CALL and NOTE_INSN_CALL_ARG_LOCATION.
29778 2017-04-07 Martin Liska <mliska@suse.cz>
29781 * config/aarch64/aarch64.c (aarch64_process_target_attr):
29782 Show error message instead of an ICE.
29784 2017-04-07 Martin Liska <mliska@suse.cz>
29787 * ipa-split.c (split_function): Add function part to a same comdat
29790 2017-04-07 Richard Biener <rguenther@suse.de>
29792 PR middle-end/80341
29793 * tree.c (get_unwidened): Also handle ! for_type case for
29795 * convert.c (do_narrow): Split out from ...
29796 (convert_to_integer_1): ... here. Do not pass final truncation
29797 type to get_unwidened for TRUNC_DIV_EXPR.
29799 2017-04-07 Richard Biener <rguenther@suse.de>
29801 * tree-affine.c (wide_int_ext_for_comb): Take type rather
29803 (aff_combination_const): Adjust.
29804 (aff_combination_scale): Likewise.
29805 (aff_combination_add_elt): Likewise.
29806 (aff_combination_add_cst): Likewise.
29807 (aff_combination_convert): Likewise.
29808 (add_elt_to_tree): Likewise. Remove unused argument.
29809 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
29811 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
29813 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
29815 * config/arm/arm.c (arm_default_short_enums): Use
29816 ARM_DEFAULT_SHORT_ENUMS.
29817 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
29819 2017-04-06 Jakub Jelinek <jakub@redhat.com>
29822 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
29823 members with redundant out-of-class redeclaration.
29825 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
29828 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
29829 * config/i386/i386.md (*zero_extendsidi2):
29830 Add (?*x,*x) and (?*v,*v) alternatives.
29832 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
29835 * config/i386/i386.c (ix86_expand_builtin)
29836 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
29837 mode from insn data. Convert operands to insn operand mode.
29838 Copy operands that don't satisfy insn predicate to a register.
29840 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
29842 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
29845 2017-04-06 Richard Biener <rguenther@suse.de>
29847 PR tree-optimization/80334
29848 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
29849 preserve alignment of accesses.
29851 2017-04-06 Richard Biener <rguenther@suse.de>
29853 PR tree-optimization/80262
29854 * tree-sra.c (build_ref_for_offset): Preserve address-space
29856 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
29857 Drop useless address-space information on MEM_REF offsets.
29859 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
29861 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
29863 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
29865 PR rtl-optimization/70703
29866 * ira-color.c (update_conflict_hard_regno_costs): Use
29867 HOST_WIDE_INT instead of long.
29869 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
29872 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
29873 not defined for x86_64 target. Add -mmmx target option when __SSE2__
29875 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
29876 for x86_64 target. Handle -m3dnowa option.
29878 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
29880 PR rtl-optimization/70703
29881 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
29882 (update_conflict_hard_regno_costs): Use long instead of unsigned
29883 arithmetic for cost calculation.
29885 2017-04-05 Jakub Jelinek <jakub@redhat.com>
29886 Bernd Edlinger <bernd.edlinger@hotmail.de>
29889 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
29892 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
29895 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
29896 ptr_mode with Pmode throughout.
29897 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
29898 into probe_stack_range and use DImode.
29900 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
29903 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
29904 call_eh_return is true.
29906 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29908 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
29909 Initialize last_match_fntype_index.
29911 2017-04-05 Jakub Jelinek <jakub@redhat.com>
29914 * tree-nvr.c: Include internal-fn.h.
29915 (pass_return_slot::execute): Ignore internal calls without
29918 2017-04-04 Jakub Jelinek <jakub@redhat.com>
29919 Richard Biener <rguenther@suse.de>
29922 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
29923 captures used multiple times, except for the last use.
29924 * generic-match-head.c: Include gimplify.h.
29926 2017-04-04 Jakub Jelinek <jakub@redhat.com>
29928 PR tree-optimization/79390
29929 * target.h (struct noce_if_info): Declare.
29930 * targhooks.h (default_noce_conversion_profitable_p): Declare.
29931 * target.def (noce_conversion_profitable_p): New target hook.
29932 * ifcvt.h (struct noce_if_info): New type, moved from ...
29933 * ifcvt.c (struct noce_if_info): ... here.
29934 (noce_conversion_profitable_p): Renamed to ...
29935 (default_noce_conversion_profitable_p): ... this. No longer
29937 (noce_try_store_flag_constants, noce_try_addcc,
29938 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
29939 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
29940 instead of noce_conversion_profitable_p.
29941 * config/i386/i386.c: Include ifcvt.h.
29942 (ix86_option_override_internal): Don't override
29943 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
29944 (ix86_noce_conversion_profitable_p): New function.
29945 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
29946 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
29947 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
29948 * doc/tm.texi: Regenerated.
29950 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29952 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
29955 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
29958 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
29959 instructions for small multiply cores.
29961 2017-04-04 Jeff Law <law@redhat.com>
29963 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
29965 (mips_expand_vec_perm_const): Initialize elements in orig_perm
29966 that are not set by the loop over the elements.
29968 2017-04-04 Jakub Jelinek <jakub@redhat.com>
29971 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
29972 int mode, convert_modes it to mode as unsigned, otherwise use
29973 lowpart_subreg to mode rather than SImode.
29974 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
29975 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
29976 Use DImode instead of SImode for the shift count operand.
29977 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
29980 2017-04-04 Richard Biener <rguenther@suse.de>
29982 PR middle-end/80281
29983 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
29984 arithmetic done for the negate or the plus. Simplify.
29985 (A - (-B) -> A + B): Likewise.
29986 * fold-const.c (split_tree): Make sure to not negate pointers.
29988 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
29990 PR rtl-optimization/60818
29991 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
29992 a compare of comparisons with the thing compared if this results
29993 in a different machine mode.
29995 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
29997 * alias.c (base_alias_check): Fix typo in comment.
29998 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
29999 * cgraphunit.c (symbol_table::compile): Likewise.
30000 * collect2.c (maybe_run_lto_and_relink): Likewise.
30001 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
30002 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
30003 * config/avr/avr.c (avr_map_op_t): Likewise.
30004 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
30005 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
30006 * config/epiphany/epiphany.md (movcc): Likewise.
30007 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
30008 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
30010 * config/mips/mips.c (mips_save_restore_reg): Likewise.
30011 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
30012 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
30013 * config/sh/sh.c (sh_rtx_costs): Likewise.
30014 * fold-const.c (fold_truth_andor): Likewise.
30015 * genautomata.c (collapse_flag): Likewise.
30016 * gengtype.h (struct type::u::s): Likewise.
30017 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
30018 * input.c (FORMAT_AMOUNT): Likewise.
30019 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
30020 (known_aggs_to_agg_replacement_list): Likewise.
30021 * ipa-inline-analysis.c: Likewise.
30022 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
30023 * ipa-polymorphic-call.c
30024 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
30025 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
30026 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
30028 * modulo-sched.c (apply_reg_moves): Likewise.
30029 * omp-expand.c (build_omp_regions_1): Likewise.
30030 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
30031 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
30032 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
30033 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
30034 * value-prof.c: Likewise.
30035 * var-tracking.c (val_reset): Likewise.
30037 2017-04-03 Richard Biener <rguenther@suse.de>
30039 PR tree-optimization/80275
30040 * fold-const.c (split_address_to_core_and_offset): Handle
30043 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
30045 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
30046 descriptors is at least equal to that of functions.
30048 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
30050 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
30052 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
30055 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
30056 (mov<IMOD4:mode>): New expander.
30057 (*mov<IMOD4:mode>_internal): New insn and split pattern.
30059 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
30061 PR rtl-optimization/79405
30062 * fwprop.c (propagations_left): New variable.
30063 (forward_propagate_into): Decrement it.
30064 (fwprop_init): Initialize it.
30065 (fw_prop): If the variable has reached zero, stop propagating.
30066 (fwprop_addr): Ditto.
30068 2017-03-31 Jakub Jelinek <jakub@redhat.com>
30071 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
30072 a FUNCTION_DECL, pass it as decl instead of origin to
30075 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
30077 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
30080 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
30083 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
30084 TARGET_VSX_SMALL_INTEGER.
30086 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30088 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30089 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
30091 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
30093 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
30094 extraction from odd-numbered MSA register.
30096 2017-03-31 Jakub Jelinek <jakub@redhat.com>
30098 PR middle-end/80173
30099 * expmed.c (store_bit_field_1): Don't attempt to create
30100 a word subreg out of hard registers wider than word if they
30101 have HARD_REGNO_NREGS of 1 for their mode.
30103 PR middle-end/80163
30104 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
30105 conversions to integer types wider than word and pointer.
30108 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
30109 (rtx_equal_for_cselib_p): Pass 0 to it.
30110 * cselib.c (cselib_hasher::equal): Likewise.
30111 (rtx_equal_for_cselib_1): Add depth argument. If depth
30112 is 128, don't look up VALUE locs and punt. Increment
30113 depth in recursive calls when walking VALUE locs.
30115 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
30117 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
30118 (make_gcov_file_name): Use the canonical path name for generating
30120 (read_line): Fix handling of files with ascii null bytes.
30122 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
30124 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
30125 to initialise a vector register instead
30126 of using a const_int.
30128 2017-03-30 Jakub Jelinek <jakub@redhat.com>
30130 PR translation/80189
30131 * gimplify.c (omp_default_clause): Use %qs instead of %s in
30132 diagnostic messages.
30134 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
30137 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
30138 (dfp_diex_<mode>): Update mode of operand 1.
30139 * doc/extend.texi (dxex, dxexq): Document change to return type.
30140 (diex, diexq): Document change to argument type.
30142 2017-03-30 Martin Jambor <mjambor@suse.cz>
30145 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
30146 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
30147 it reflects the signature changes performed at the callee side.
30148 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
30149 to cgraph_build_function_type_skip_args.
30150 (build_function_decl_skip_args): Adjust call to the above function.
30152 2017-03-30 Jakub Jelinek <jakub@redhat.com>
30155 * config/i386/sse.md
30156 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
30157 register as dest whenever it is a MEM not rtx_equal_p to the
30158 corresponding dup operand, and when forcing into reg move the
30159 reg into the memory afterwards.
30160 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
30161 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
30162 for the force_reg mode.
30163 (avx512vl_vextractf128<mode>): Use register as dest either
30164 always when a MEM, or when it is a MEM not rtx_equal_p to the
30165 corresponding dup operand, or even not when it is a CONST_VECTOR
30166 depending on the mode and lo vs. hi.
30167 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
30169 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
30170 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
30171 Likewise. Require that operands[2] is even.
30172 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
30173 Remove extraneous parens. Require that operands[2] is a multiple
30175 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
30176 operands[0] is a MEM if <mask_applied>, the predicates/constraints
30177 disallow memory then.
30179 2017-03-30 Richard Biener <rguenther@suse.de>
30181 PR tree-optimization/77498
30182 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
30183 to non-constants over backedges.
30185 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
30187 PR rtl-optimization/80233
30188 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
30189 as last_combined_insn. Do not test for BARRIER_P separately.
30191 2017-03-29 Andreas Schwab <schwab@suse.de>
30194 * calls.c (prepare_call_address): Convert funexp to Pmode before
30195 copying to temp reg.
30197 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30199 PR tree-optimization/80158
30200 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
30201 Handle possible future case of more than one alternate
30203 (replace_rhs_if_not_dup): Likewise.
30204 (replace_one_candidate): Likewise.
30206 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
30208 PR rtl-optimization/80193
30209 * ira.c (ira): Do not check allocation for LRA.
30211 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
30213 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
30214 (nvptx_output_simt_exit): Declare.
30215 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
30216 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
30217 (init_softstack_frame): Move initialization of crtl->is_leaf to...
30218 (nvptx_declare_function_name): ...here. Emit declaration of local
30219 memory space buffer for omp_simt_enter insn.
30220 (nvptx_output_unisimt_switch): New.
30221 (nvptx_output_softstack_switch): New.
30222 (nvptx_output_simt_enter): New.
30223 (nvptx_output_simt_exit): New.
30224 * config/nvptx/nvptx.h (struct machine_function): New fields
30225 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
30226 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
30227 (UNSPECV_SIMT_EXIT): Ditto.
30228 (omp_simt_enter_insn): New insn.
30229 (omp_simt_enter): New expansion.
30230 (omp_simt_exit): New insn.
30231 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
30233 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
30234 (expand_GOMP_SIMT_ENTER_ALLOC): New.
30235 (expand_GOMP_SIMT_EXIT): New.
30236 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
30237 (GOMP_SIMT_ENTER_ALLOC): Ditto.
30238 (GOMP_SIMT_EXIT): Ditto.
30239 * target-insns.def (omp_simt_enter): New insn.
30240 (omp_simt_exit): Ditto.
30241 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
30243 (lower_rec_simd_input_clauses): Implement SIMT privatization.
30244 (lower_rec_input_clauses): Likewise.
30245 (lower_lastprivate_clauses): Handle SIMT privatization.
30247 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
30248 (ompdevlow_adjust_simt_enter): New.
30249 (find_simtpriv_var_op): New.
30250 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
30251 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
30253 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
30254 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
30255 (copy_decl_for_dup_finish): Ditto.
30257 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
30259 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
30262 * config/i386/i386.c (ix86_option_override_internal): Always
30263 allow -mpreferred-stack-boundary=3 for 64-bit targets.
30265 2017-03-28 Bin Cheng <bin.cheng@arm.com>
30267 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
30269 2017-03-28 Bin Cheng <bin.cheng@arm.com>
30271 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
30272 mark new edge's irreducible flag accordign to it.
30273 (vect_do_peeling): Check loop preheader edge's irreducible flag
30274 and pass it to function slpeel_add_loop_guard.
30276 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
30278 PR tree-optimization/80218
30279 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
30280 Update block frequencies and counts.
30282 2017-03-28 Richard Biener <rguenther@suse.de>
30284 PR tree-optimization/78644
30285 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
30286 of a simplification result we may not use it at all.
30288 2017-03-28 Richard Biener <rguenther@suse.de>
30291 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
30292 without arguments, generate default definition of a SSA name.
30294 2017-03-28 Richard Biener <rguenther@suse.de>
30296 PR middle-end/80222
30297 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
30298 TYPE_REF_CAN_ALIAS_ALL references.
30299 * fold-const.c (fold_indirect_ref_1): Likewise.
30301 2017-03-28 Martin Liska <mliska@suse.cz>
30304 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
30305 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
30307 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
30308 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
30310 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
30311 (EXTRA_SPECS): Define.
30312 (SUBTARGET_EXTRA_SPECS): Likewise.
30313 (SUBTARGET_CPP_SPEC): Likewise.
30314 * config/arc/elf.h (EXTRA_SPECS): Renamed to
30315 SUBTARGET_EXTRA_SPECS.
30316 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
30318 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
30320 * config/arc/simdext.md (vst64_insn): Update pattern.
30321 (vld32wh_insn): Likewise.
30322 (vld32wl_insn): Likewise.
30323 (vld64_insn): Likewise.
30324 (vld32_insn): Likewise.
30326 2017-03-28 Marek Polacek <polacek@redhat.com>
30329 * fold-const.c (fold_comparison): Use protected_set_expr_location
30330 instead of SET_EXPR_LOCATION.
30332 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
30334 * tree.c (add_expr): Avoid name lookup warning.
30336 2017-03-27 Jeff Law <law@redhat.com>
30338 PR tree-optimization/80216
30339 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
30340 function name. Limit recursion depth.
30341 (record_temporary_equivalences): Corresponding changes.
30343 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
30345 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
30348 2017-03-27 Jakub Jelinek <jakub@redhat.com>
30351 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
30353 * cfgcleanup.c (reg_note_cfa_p): New array.
30354 (insns_have_identical_cfa_notes): New function.
30355 (old_insns_match_p): Don't cross-jump in between /f
30356 and non-/f instructions. If both i1 and i2 are frame related,
30357 verify all CFA notes, their order and content.
30359 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
30362 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
30363 HImode and SImode with zero extend to DImode to one insn.
30364 (bswap<mode>2_extenddi): Likewise.
30365 (bswapsi2_extenddi): Likewise.
30366 (bswaphi2_extendsi): Likewise.
30367 (bswaphi2): Combine bswap HImode and SImode into one insn.
30368 Separate memory insns from swapping register.
30369 (bswapsi2): Likewise.
30370 (bswap<mode>2): Likewise.
30371 (bswaphi2_internal): Delete, no longer used.
30372 (bswapsi2_internal): Likewise.
30373 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
30374 store, and gpr<-gpr swap insns.
30375 (bswap<mode>2_store): Likewise.
30376 (bswaphi2_reg): Register only splitter, combine with the splitter.
30377 (bswaphi2 splitter): Likewise.
30378 (bswapsi2_reg): Likewise.
30379 (bswapsi2 splitter): Likewise.
30380 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
30381 the insns into load, store, and register/register insns.
30382 (bswapdi2_ldbrx): Likewise.
30383 (bswapdi2_load): Likewise.
30384 (bswapdi2_store): Likewise.
30385 (bswapdi2_reg): Likewise.
30387 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
30389 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
30390 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
30392 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
30395 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
30397 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30398 special handling for target option conflicts between dform
30399 options (-mpower9-dform, -mpower9-dform-vector,
30400 -mpower9-dform-scalar) and -mno-direct-move.
30402 2017-03-27 Richard Biener <rguenther@suse.de>
30404 PR tree-optimization/80181
30405 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
30407 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30409 * config/arc/predicates.md (move_double_src_operand): Replace the
30410 call to move_double_src_operand with a call to address_operand.
30412 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30414 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
30415 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
30416 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
30418 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30420 * config/arc/predicates.md (long_immediate_loadstore_operand):
30421 Consider scaled addresses cases.
30423 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30425 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
30426 restored when in interrupt.
30427 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
30428 doesn't have delay slot.
30430 2017-03-27 Richard Biener <rguenther@suse.de>
30433 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
30434 inlined thunk clones.
30436 2017-03-27 Jakub Jelinek <jakub@redhat.com>
30439 * asan.c (instrument_derefs): Copy over last operand from
30440 original COMPONENT_REF to the new COMPONENT_REF with
30441 DECL_BIT_FIELD_REPRESENTATIVE.
30442 * ubsan.c (instrument_object_size): Likewise.
30444 2017-03-27 Richard Biener <rguenther@suse.de>
30446 PR tree-optimization/80170
30447 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
30448 sure DR/SCEV didnt fold in constants we do not see when looking
30449 at the reference base alignment.
30451 2017-03-27 Richard Biener <rguenther@suse.de>
30453 PR middle-end/80171
30454 * gimple-fold.c (fold_ctor_reference): Properly guard against
30455 NULL return value from canonicalize_constructor_val.
30457 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
30460 * config/i386/i386.c (ix86_expand_builtin)
30461 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
30462 flags reg setting and flags reg using instructions.
30463 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
30464 clobbering instructions to zero extend op2.
30466 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
30468 * doc/install.texi (Configuration) <--with-aix-soname>:
30469 Update link to AIX ld.
30471 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
30473 PR rtl-optimization/80160
30474 PR rtl-optimization/80159
30475 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
30476 reg_alternate_class into account.
30478 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
30481 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
30482 to consider in curr_insn_transform.
30484 2017-03-24 Jakub Jelinek <jakub@redhat.com>
30486 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
30487 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
30488 and emit_mode_inner.
30490 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30492 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
30493 argument to the overloaded builtin variants. Use the new flag to
30494 deprecate certain builtin variants.
30495 * config/s390/s390-builtin-types.def: Add new builtin types.
30496 * config/s390/s390-builtins.h: Support new flags field for
30497 overloaded builtins.
30498 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
30499 (s390_macro_to_expand): Enable vector float data type.
30500 (s390_cpu_cpp_builtins_internal): Indicate support of the new
30501 builtins by incrementing the __VEC__ version number.
30502 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
30504 (s390_resolve_overloaded_builtin): Emit error messages depending
30505 on the builtin flags.
30506 * config/s390/s390.c (s390_expand_builtin): Support additional
30507 flags argument. Change error message to match the messages
30508 emitted in s390-c.c.
30509 * config/s390/s390.md: New UNSPEC_* constants.
30510 (op_type): Add new instruction types.
30511 * config/s390/vecintrin.h: Add new builtins and test data class
30513 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
30514 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
30515 (VEC_INEXACT, VEC_NOINEXACT): New constants.
30516 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
30517 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
30518 ("vec_mergel<mode>"): V_HW -> VEC_HW.
30520 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
30521 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
30522 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
30523 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
30525 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
30526 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
30527 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
30528 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
30530 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
30531 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
30532 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
30533 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
30534 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
30535 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
30536 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
30538 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
30539 ("vec_scatter_element<V_HW_4:mode>_DI")
30540 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
30541 ("vec_fpint<mode>", "vflls")
30542 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
30543 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
30544 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
30545 ("*vec_cmphe<mode>_cc"): ... these.
30547 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
30548 mode constant instead of magic value.
30550 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30552 * config/s390/s390.c (s390_expand_vec_compare): Support other
30553 vector floating point modes than just V2DF.
30554 (s390_expand_vcond): Likewise.
30555 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
30556 (s390_cannot_change_mode_class): Prevent mode changes between TF
30557 and V1TF in vector registers.
30558 * config/s390/s390.md (DF, SF): New mode attributes.
30559 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
30560 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
30561 SFmode support for VRs.
30562 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
30564 (VFT, VF_HW): New mode iterators.
30565 (vw, sdx): New mode attributes.
30566 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
30567 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
30568 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
30569 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
30570 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
30571 also the new vector floating point modes. Renaming to ...
30573 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
30574 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
30575 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
30576 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
30577 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
30578 ("vec_unordered<mode>"): ... these.
30580 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
30581 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
30582 ("*vec_extendv2df"): New insn definitions.
30584 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30586 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
30587 ("mulditi3_2", "*muldi3_sign"): New patterns.
30588 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
30589 rename the pattern definition.
30591 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30593 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
30595 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
30597 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30599 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
30600 instruction if possible.
30601 * config/s390/vector.md (vec_halfnumelts): New mode
30603 ("*vec_vllezlf<mode>"): New pattern.
30605 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30607 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
30608 ("popcountv4si2", "popcountv2di2"): Rename to ...
30609 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
30610 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
30612 ("popcount<mode>2_vxe"): New pattern.
30614 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30616 * common/config/s390/s390-common.c (processor_flags_table): Add
30618 * config.gcc: Add arch12.
30619 * config/s390/driver-native.c (s390_host_detect_local_cpu):
30620 Default to arch12 for unknown CPU model numbers.
30621 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
30622 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
30623 PROCESSOR_max sanity check.
30624 * config/s390/s390-opts.h (enum processor_type): Add
30626 * config/s390/s390.c (processor_table): Add arch12.
30627 (s390_expand_builtin): Add check for B_VXE flag.
30628 (s390_issue_rate): Add PROCESSOR_ARCH12.
30629 (s390_get_sched_attrmask): Likewise.
30630 (s390_get_unit_mask): Likewise.
30631 (s390_sched_score): Enable z13 scheduling for arch12.
30632 (s390_sched_reorder): Likewise.
30633 (s390_sched_variable_issue): Likewise.
30634 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
30636 (s390_tune_attr): Use z13 scheduling also for arch12.
30637 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
30638 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
30639 (TARGET_VXE_P): New macros.
30640 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
30641 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
30642 * config/s390/s390.opt: Add arch12 as processor_type.
30644 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30646 * config/s390/s390.md
30647 ("fixuns_truncdddi2", "fixuns_trunctddi2")
30648 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
30649 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
30651 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
30652 Rename expanders to ...
30654 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
30655 ("fixuns_truncdddi2_emu"): ... these.
30657 ("fixuns_trunc<mode>si2_emu"): New expander.
30659 ("*fixuns_truncdfdi2_z13"): Rename to ...
30660 ("*fixuns_truncdfdi2_vx"): ... this.
30662 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30664 * config/s390/2964.md: Remove the single element vector compare
30665 instructions which are no longer used.
30666 * config/s390/s390.c (s390_select_ccmode): Remove handling of
30668 (s390_canonicalize_comparison): Remove handling of DFmode
30670 (s390_expand_vec_compare_scalar): Remove function.
30671 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
30672 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
30674 ("*cmp<mode>_ccs"): Add wfcdb instruction.
30676 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30678 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
30680 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
30681 will anyway by matched by mov<mode>_64dfp.
30683 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30685 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
30686 vlef/vstef. Add missing operand to vleif.
30688 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30690 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
30691 pair for all vector types with 64 bit elements.
30692 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
30693 * config/s390/vector.md (V_HW_64): ... here.
30694 (V_128_NOSINGLE): New mode iterator.
30695 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
30696 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
30697 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
30698 ("*vec_load_pairv2di"): Change to ...
30699 ("*vec_load_pair<mode>"): ... this one.
30701 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30703 * config/s390/constraints.md: Add comments.
30704 (jKK): Reject element sizes > 8 bytes.
30705 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
30707 * config/s390/s390.md: Add the s_operand checks formerly in
30708 s390_split_ok_p to various splitters where they are still
30710 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
30711 for 128 bit vectors. Plus two splitters.
30713 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30715 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
30718 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30721 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
30722 error if the boundary argument is not constant.
30724 2017-03-24 Jakub Jelinek <jakub@redhat.com>
30726 PR rtl-optimization/80112
30727 * loop-doloop.c (doloop_condition_get): Don't check condition
30728 if cmp isn't SET with IF_THEN_ELSE src.
30730 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30732 PR tree-optimization/80158
30733 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
30734 replacing a candidate statement, also replace it for the
30735 candidate's alternate interpretation.
30736 (replace_rhs_if_not_dup): Likewise.
30737 (replace_one_candidate): Likewise.
30739 2017-03-24 Richard Biener <rguenther@suse.de>
30741 PR tree-optimization/80167
30742 * graphite-isl-ast-to-gimple.c
30743 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
30745 (translate_isl_ast_to_gimple::get_rename): Likewise.
30747 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
30749 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
30750 handling of certain combinations of target options, including the
30751 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
30752 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
30754 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30757 * config/arm/arm.md (*load_multiple): Add reload_completed to
30758 matching condition.
30760 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30761 Richard Biener <rguenth@suse.de>
30763 PR tree-optimization/79908
30764 PR tree-optimization/80136
30765 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
30766 been cast away, gimplify_and_add suffices.
30768 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
30770 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
30772 2017-03-23 Richard Biener <rguenther@suse.de>
30774 PR tree-optimization/80032
30775 * gimplify.c (gimple_push_cleanup): Forced unconditional
30776 cleanups still have to go to the conditional_cleanups
30779 2017-03-22 Jakub Jelinek <jakub@redhat.com>
30781 PR tree-optimization/80072
30782 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
30784 (next_operand_entry_id): Change type to unsigned int.
30785 (sort_by_operand_rank): Make sure to return the right return value
30786 even if unsigned fields are bigger than INT_MAX.
30787 (struct oecount): Change cnt and id type to unsigned int.
30788 (oecount_hasher::equal): Formatting fix.
30789 (oecount_cmp): Make sure to return the right return value
30790 even if unsigned fields are bigger than INT_MAX.
30791 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
30794 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
30795 TREE_READONLY on result if writing it more than once.
30798 * doc/invoke.texi (-fsanitize=thread): Document that with
30799 -fnon-call-exceptions atomics are not able to throw
30803 * tsan.c: Include tree-eh.h.
30804 (instrument_builtin_call): Call maybe_clean_eh_stmt or
30805 maybe_clean_or_replace_eh_stmt where needed.
30806 (instrument_memory_accesses): Add cfg_changed argument.
30807 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
30808 if it returned true.
30809 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
30811 PR rtl-optimization/63191
30812 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
30813 wrapper function, moved the whole old content into ...
30814 (ix86_delegitimize_address_1): ... this. New inline function.
30815 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
30816 true as last argument instead of ix86_delegitimize_address.
30818 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
30820 * config/aarch64/aarch64.c (generic_branch_cost): Copy
30821 cortexa57_branch_cost.
30823 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
30825 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
30827 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30830 * doc/md.texi (Constraints): Document wA constraint.
30831 * config/rs6000/constraints.md (wA): New.
30832 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
30833 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
30834 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
30835 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
30837 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
30840 * gimplify.c (is_oacc_declared): New function.
30841 (oacc_default_clause): Use it to set default flags for acc declared
30842 variables inside parallel regions.
30843 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
30844 declared variables.
30845 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
30846 declare attribute to any decl as necessary.
30848 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
30851 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
30852 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
30853 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
30854 (arm_arch_lpae): This.
30855 * config/arm/arm.c (arm_arch7ve): Rename into ...
30856 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
30857 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
30860 2017-03-22 Martin Liska <mliska@suse.cz>
30863 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
30864 error message instead of an ICE.
30866 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30868 * doc/extend.texi (6.11 Additional Floating Types): Revise.
30870 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
30872 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30874 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30877 2017-03-21 Martin Sebor <msebor@redhat.com>
30879 * doc/extend.texi: Use "cannot" instead of "can't."
30880 * doc/hostconfig.texi: Same.
30881 * doc/install.texi: Same.
30882 * doc/invoke.texi: Same.
30883 * doc/loop.texi: Same.
30884 * doc/md.texi: Same.
30885 * doc/objc.texi: Same.
30886 * doc/rtl.texi: Same.
30887 * doc/tm.texi: Same.
30888 * doc/tm.texi.in: Same.
30889 * doc/trouble.texi: Same.
30891 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
30894 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
30895 (collect_checksum_attributes): Set it.
30896 (die_checksum_ordered): Use it.
30898 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30900 PR tree-optimization/79908
30901 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
30902 change: For a VA_ARG whose LHS has been cast away, use
30903 force_gimple_operand to construct the side effects.
30905 2017-03-21 David Malcolm <dmalcolm@redhat.com>
30907 PR translation/80001
30908 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
30909 more amenable to translation.
30910 (oacc_loop_auto_partitions): Likewise.
30912 2017-03-21 Marek Polacek <polacek@redhat.com>
30913 Martin Sebor <msebor@redhat.com>
30915 PR tree-optimization/80109
30916 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
30917 on INTEGRAL_TYPE_P.
30919 2017-03-21 Jakub Jelinek <jakub@redhat.com>
30920 Segher Boessenkool <segher@kernel.crashing.org>
30923 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
30924 check reg_used_between_p between insn and one of succ or succ2
30925 depending on if succ is artificial insn not inserted into insn
30928 2017-03-21 Martin Liska <mliska@suse.cz>
30930 PR gcov-profile/80081
30931 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
30932 * doc/gcc.texi: Include gcov-dump stuff.
30933 * doc/gcov-dump.texi: New file.
30935 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
30937 PR rtl-optimization/79150
30938 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
30939 conditional jump, if the jump is the last insn of the loop.
30941 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30942 Richard Biener <rguenth@suse.de>
30944 PR tree-optimization/79908
30945 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
30946 been cast away, use force_gimple_operand to construct the side
30949 2017-03-21 Martin Liska <mliska@suse.cz>
30951 PR libfortran/79956
30952 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
30955 2017-03-21 Brad Spengler <spender@grsecurity.net>
30958 * plugin.c (htab_hash_plugin): New function.
30959 (add_new_plugin): Use it and adjust.
30960 (parse_plugin_arg_opt): Adjust.
30961 (init_one_plugin): Likewise.
30963 2017-03-21 Richard Biener <rguenther@suse.de>
30965 PR tree-optimization/80032
30966 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
30967 if set force the cleanup to happen unconditionally.
30968 (gimplify_target_expr): Push inserted clobbers with force_uncond
30969 to avoid them being removed by control-dependent DCE.
30971 2017-03-21 Richard Biener <rguenther@suse.de>
30973 PR tree-optimization/80122
30974 * tree-inline.c (copy_bb): Do not expans va-arg packs or
30975 va_arg_pack_len when the inlined call stmt requires pack
30977 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
30979 2017-03-21 Jakub Jelinek <jakub@redhat.com>
30982 * tsan.c (instrument_builtin_call): If the memory model argument
30983 is not a constant, assume it is valid.
30986 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
30989 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
30991 PR rtl-optimization/79910
30992 * combine.c (can_combine_p): Do not allow combining an I0 or I1
30993 if its dest is used by an insn before I2 (other than the combined
30994 insns themselves, which are properly handled already).
30996 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
30999 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
31001 * combine.c (record_used_regs): New static function.
31002 (try_combine): Handle situations where there is an additional
31003 instruction between I2 and I3 which needs to have a LOG_LINK
31007 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
31009 * combine.c (try_combine): Delete redundant i1 test. Call
31010 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
31012 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31015 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
31016 alternatives 13/14.
31018 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31020 PR tree-optimization/80054
31021 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
31022 the optimization if a PHI or any of its arguments is not dominated
31023 by the candidate's basis. Use gphi* rather than gimple* as
31025 (replace_profitable_candidates): Clean up a gimple* variable that
31026 should be a gphi* variable.
31028 2017-03-20 Martin Sebor <msebor@redhat.com>
31031 * doc/extend.texi (attribute constructor): Document present limitation.
31033 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
31036 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
31037 __POWER9_VECTOR__ #ifdef control, change template definition to
31038 use Power9-specific built-in function.
31039 (vec_any_eq): Likewise.
31040 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
31041 to control outcomes from this test.
31042 (vector_ae_<mode>p): For VEC_F modes, likewise.
31044 2017-03-20 Ian Lance Taylor <iant@google.com>
31046 * config/i386/i386.c (ix86_function_regparm): Save an extra
31047 register for -fsplit-stack with DECL_STATIC_CHAIN.
31049 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
31052 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
31053 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
31055 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
31057 * config/riscv/riscv.c (riscv_print_operand): Use "fence
31059 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
31062 2017-03-20 Marek Polacek <polacek@redhat.com>
31065 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
31067 2017-03-20 Richard Biener <rguenther@suse.de>
31069 PR tree-optimization/80113
31070 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
31071 allocate extra SSA name for PHI def.
31072 (add_close_phis_to_outer_loops): Likewise.
31073 (add_close_phis_to_merge_points): Likewise.
31074 (copy_loop_close_phi_args): Likewise.
31075 (copy_cond_phi_nodes): Likewise.
31077 2017-03-20 Martin Liska <mliska@suse.cz>
31079 PR middle-end/79753
31080 * tree-chkp.c (chkp_build_returned_bound): Do not build
31081 returned bounds for a LHS that's not a BOUNDED_P type.
31083 2017-03-20 Martin Liska <mliska@suse.cz>
31087 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
31088 COMPLEX_CST and VECTOR_CST.
31090 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31093 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
31094 target operand. A new splitter adds the clobber statement in case
31095 the target operand is dead anyway.
31097 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
31099 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
31100 to age-old versions of binutils and glibc.
31102 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
31104 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
31106 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
31108 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
31110 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
31112 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
31113 requirement for binutils 2.13.
31115 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
31117 * combine.c (try_combine): Delete redundant i1 test. Call
31118 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
31120 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
31122 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
31123 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
31125 <riscv64-*-elf>: Re-arrange section
31126 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
31127 <riscv32-*-linux>: Likewise.
31128 <riscv64-*-elf>: Likewise
31129 <riscv64-*-linux>: Likewise.
31131 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
31134 * aarch64.opt(verbose-cost-dump): Fix typo.
31136 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
31139 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
31140 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
31142 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
31144 * reload.c (find_reloads): When reloading a nonoffsettable address,
31145 use RELOAD_OTHER for it and its address reloads.
31147 PR rtl-optimization/79910
31148 * combine.c (record_used_regs): New static function.
31149 (try_combine): Handle situations where there is an additional
31150 instruction between I2 and I3 which needs to have a LOG_LINK
31153 2017-03-17 Jeff Law <law@redhat.com>
31155 PR tree-optimization/71437
31156 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
31157 conditional in the hash table first.
31158 (vrp_dom_walker::before_dom_children): Extract condition from
31159 ASSERT_EXPR. Record condition, its inverion and any implied
31160 conditions as well.
31162 2017-03-17 Marek Polacek <polacek@redhat.com>
31163 Markus Trippelsdorf <markus@trippelsdorf.de>
31165 PR tree-optimization/80079
31166 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
31169 2017-03-17 Richard Biener <rguenther@suse.de>
31171 PR middle-end/80075
31172 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
31173 Properly verify the LHS before the RHS possibly claims to be
31175 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
31178 2017-03-17 Martin Jambor <mjambor@suse.cz>
31180 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
31181 (List of -O2 options): Likewise.
31182 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
31185 2017-03-17 Tom de Vries <tom@codesourcery.com>
31187 * gcov-dump.c (print_usage): Print bug_report_url.
31189 2017-03-17 Richard Biener <rguenther@suse.de>
31191 PR middle-end/80050
31192 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
31193 (parser::peek): Likewise.
31195 2017-03-17 Richard Biener <rguenther@suse.de>
31197 PR tree-optimization/80048
31198 * sese.c (free_sese_info): Properly release rename_map and
31199 copied_bb_map elements.
31201 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
31203 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
31204 Add linked-list forward and backlinks. Insert on
31205 construction, remove on destruction.
31206 (class pass_store_merging): Add m_stores_head field.
31207 (pass_store_merging::terminate_and_process_all_chains):
31208 Iterate over m_stores_head list.
31209 (pass_store_merging::terminate_all_aliasing_chains):
31211 (pass_store_merging::execute): Check for debug stmts first.
31212 Push new chains onto the m_stores_head stack.
31214 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
31217 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
31218 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
31219 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
31221 2017-03-16 Jeff Law <law@redhat.com>
31223 PR tree-optimization/71437
31224 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
31225 member function. Implementation moved into after_dom_children
31226 member function and into the threader's thread_outgoing_edges
31228 (dom_opt_dom_walker::after_dom_children): Simplify by moving
31229 some code into new thread_outgoing_edges.
31230 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
31231 definition. Simplify marker handling (do it here). Assume we always
31232 have the available expression and the const/copies tables.
31233 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
31235 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
31236 * tree-vrp.c (equiv_stack): No longer file scoped.
31237 (vrp_dom_walker): New class.
31238 (vrp_dom_walker::before_dom_children): New member function.
31239 (vrp_dom_walker::after_dom_children): Likewise.
31240 (identify_jump_threads): Setup domwalker. Use it rather than
31241 walking edges in a random order by hand. Simplify setup/finalization.
31242 (finalize_jump_threads): Remove.
31243 (vrp_finalize): Do not call identify_jump_threads here.
31244 (execute_vrp): Do it here instead and call thread_through_all_blocks
31247 PR tree-optimization/71437
31248 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
31250 (simplify_stmt_for_jump_threading): Add basic_block argument. All
31252 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
31253 (dom_opt_dom_walker::thread_across_edge): Remove
31254 handle_dominating_asserts argument. All callers changed.
31255 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
31256 changes. Remove calls to lhs_of_dominating_assert. Other
31257 uses of handle_dominating_asserts turn into unconditional code
31258 (simplify_control_stmt_condition_1): Likewise.
31259 (simplify_control_stmt_condition): Likewise.
31260 (thread_through_normal_block, thread_across_edge): Likewise.
31261 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
31262 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
31263 object if it is not an SSA_NAME.
31264 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
31265 before calling into the VRP specific simplifiers.
31266 (identify_jump_threads): Remove handle_dominating_asserts
31269 2017-03-16 Jakub Jelinek <jakub@redhat.com>
31272 * tree-diagnostic.c (default_tree_printer): No longer static.
31273 * tree-diagnostic.h (default_tree_printer): New prototype.
31275 2017-03-16 Tamar Christina <tamar.christina@arm.com>
31277 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
31278 Change ins into fmov.
31280 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31282 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
31283 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
31284 Use h_con constraint for operand 1.
31285 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
31286 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
31288 2017-03-15 Jeff Law <law@redhat.com>
31290 PR tree-optimization/71437
31291 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
31292 (record_temporary_equivalences): Use it.
31294 PR tree-optimization/71437
31295 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
31296 tree-ssa-scopedtables.
31297 (lookup_avail_expr, build_and_record_new_cond): Likewise.
31298 (record_conditions, record_cond, vuse_eq): Likewise.
31299 (record_edge_info): Adjust to API tweak of record_conditions.
31300 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
31301 (record_temporary_equivalences, optimize_stmt): Likewise.
31302 (eliminate_redundant_computations): Likewise.
31303 (record_equivalences_from_stmt): Likewise.
31304 * tree-ssa-scopedtables.c: Include options.h and params.h.
31305 (vuse_eq): New function, moved from tree-ssa-dom.c
31306 (build_and_record_new_cond): Likewise.
31307 (record_conditions): Likewise. Accept vector of conditions rather
31308 than edge_equivalence structure for first argument.
31309 for the first argument.
31310 (avail_exprs_stack::lookup_avail_expr): New member function, moved
31311 from tree-ssa-dom.c.
31312 (avail_exprs_stack::record_cond): Likewise.
31313 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
31314 from tree-ssa-dom.c.
31315 (avail_exprs_stack): Add new member functions lookup_avail_expr
31317 (record_conditions): Declare.
31319 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
31322 * lra-constraints.c (process_alt_operands): Increase reject for
31323 reloading an input/output operand.
31325 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
31328 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
31329 insns to convert from signed/unsigned char/short to IEEE 128-bit
31331 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
31333 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
31336 * config/i386/i386.c (ix86_vector_duplicate_value): Create
31337 subreg of inner mode for values already in registers.
31339 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
31341 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
31342 iteration reg is used after the loop.
31344 2017-03-14 Martin Sebor <msebor@redhat.com>
31346 PR tree-optimization/79800
31347 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
31348 precision in negative-positive range.
31349 (format_floating): Call non-const overload with adjusted precision.
31351 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
31354 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
31357 2017-03-14 Martin Sebor <msebor@redhat.com>
31359 PR middle-end/80020
31360 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
31361 * builtins.def (aligned_alloc): Use it.
31364 * Makefile.in (GTFILES): Add calls.c.
31365 * calls.c: Include "gt-calls.h".
31367 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
31369 PR rtl-optimization/79728
31370 * regs.h (struct target_regs): New field
31371 x_contains_allocatable_regs_of_mode.
31372 (contains_allocatable_regs_of_mode): New macro.
31373 * reginfo.c (init_reg_sets_1): Initialize it, and change
31374 contains_reg_of_mode so it includes global regs as well.
31375 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
31376 rather than contains_regs_of_mode.
31378 2017-03-14 Martin Liska <mliska@suse.cz>
31380 * doc/invoke.texi: Document options that can't be combined with
31381 -fcheck-pointer-bounds.
31383 2017-03-14 Martin Liska <mliska@suse.cz>
31385 PR middle-end/79831
31386 * doc/invoke.texi (-Wchkp): Document the option.
31388 2017-03-14 Martin Liska <mliska@suse.cz>
31390 * Makefile.in: Install gcov-dump.
31392 2017-03-14 Martin Liska <mliska@suse.cz>
31394 * multiple_target.c (expand_target_clones): Bail out for
31395 an invalid attribute.
31397 2017-03-14 Richard Biener <rguenther@suse.de>
31399 * alias.c (struct alias_set_entry): Pack properly.
31400 * cfgloop.h (struct loop): Likewise.
31401 * cse.c (struct set): Likewise.
31402 * ipa-utils.c (struct searchc_env): Likewise.
31403 * loop-invariant.c (struct invariant): Likewise.
31404 * lra-remat.c (struct cand): Likewise.
31405 * recog.c (struct change_t): Likewise.
31406 * rtl.h (struct address_info): Likewise.
31407 * symbol-summary.h (function_summary): Likewise.
31408 * tree-loop-distribution.c (struct partition): Likewise.
31409 * tree-object-size.c (struct object_size_info): Likewise.
31410 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
31411 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
31412 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
31413 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
31414 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
31415 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
31416 (struct _stmt_vec_info): Likewise.
31418 2017-03-14 Martin Liska <mliska@suse.cz>
31421 * multiple_target.c (create_dispatcher_calls): Check that
31422 a target can create a function dispatcher.
31424 2017-03-14 Martin Liska <mliska@suse.cz>
31427 * multiple_target.c (expand_target_clones): Drop local.local
31428 flag for default implementation.
31430 2017-03-14 Richard Biener <rguenther@suse.de>
31432 PR tree-optimization/80030
31433 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
31435 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
31437 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
31438 gcc_fallthrough() instead of __attribute__((fallthrough));
31440 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
31442 * doc/gcc.texi: Remove "up" link to (DIR).
31443 * doc/gccint.texi: Ditto.
31445 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
31447 * doc/install.texi (Specific) <avr>: Remove reference to
31450 2017-03-13 Jeff Law <law@redhat.com>
31452 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
31453 attribute rather than comments.
31455 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
31456 match_scratch operand is highest.
31458 2017-03-13 Martin Liska <mliska@suse.cz>
31460 PR middle-end/78339
31461 * ipa-pure-const.c (warn_function_noreturn): If the declarations
31462 is a CHKP clone, use original declaration.
31464 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31466 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
31467 (arc_conditional_register_usage): Use a different allocation order
31468 when optimizing for size.
31469 * common/config/arc/arc-common.c (arc_option_optimization_table):
31470 Section anchors default on when optimizing for size.
31472 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31474 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
31476 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31478 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
31479 * config/arc/arc.md (cpu_facility): Add cd variant.
31480 (*movqi_insn): Add code density variant.
31481 (*movhi_insn): Likewise.
31482 (*movqi_insn): Likewise.
31483 (*addsi3_mixed): Likewise.
31484 (subsi3_insn): Likewise.
31486 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31488 * config/arc/arc.md (movsi_cond_exec): Update constraint.
31490 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31492 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
31493 expressions with MINUS and UNARY ops.
31495 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31498 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
31500 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
31501 between vec_select and vector argument.
31502 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
31503 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
31504 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
31505 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
31506 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
31507 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
31509 2017-03-13 Richard Biener <rguenther@suse.de>
31512 * params.def (vect-max-peeling-for-alignment): Fix typo.
31514 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
31516 * doc/install.texi (Specific) <mips-*-*>: Remove description of
31517 issue that only occurred with binutils below 2.18.
31519 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
31521 * doc/install.texi (Specific) <cris-axis-elf>: No longer
31522 refer to binutils 2.11/2.12 minimum.
31524 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
31526 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
31527 ftp.kernel.org and simplify binutils requirement.
31529 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
31531 * doc/invoke.texi (Warning Options): Fix spelling of link-time
31533 (Optimize Options): Ditto. Also remove redundancy.
31535 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31537 PR translation/79848
31538 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
31540 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
31541 to G_ to avoid double translation.
31543 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31545 PR translation/79923
31546 * auto-profile.c (get_combined_location): Convert leading
31547 character of diagnostics to lower case and remove trailing period.
31548 (read_profile): Likewise for various diagnostics.
31549 * config/arm/arm.c (arm_option_override): Remove trailing period
31550 from various diagnostics.
31551 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
31552 (msp430_expand_delay_cycles): Likewise.
31554 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31557 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
31558 full command-line argument, rather than just "str".
31559 (aarch64_validate_march): Likewise.
31560 (aarch64_validate_mtune): Likewise.
31562 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
31564 PR rtl-optimization/78911
31565 * lra-assigns.c (must_not_spill_p): New function.
31566 (spill_for): Use it.
31568 2017-03-10 Jakub Jelinek <jakub@redhat.com>
31570 PR tree-optimization/79981
31571 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
31572 ATOMIC_COMPARE_EXCHANGE ifn result.
31573 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
31574 IFN_ATOMIC_COMPARE_EXCHANGE.
31576 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31579 * opts.c (parse_sanitizer_options): Add missing question mark to
31580 "did you mean" message.
31582 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31584 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
31586 (VMULEUH_UNS): Likewise.
31587 (VMULOUB_UNS): Likewise.
31588 (VMULOUH_UNS): Likewise.
31589 * config/rs6000/rs6000.c (builtin_function_type): Remove
31590 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
31592 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31595 * read-rtl-function.c (function_reader::read_rtx_operand): Update
31596 x with result of extra_parsing_for_operand_code_0.
31597 (function_reader::extra_parsing_for_operand_code_0): Convert
31598 return type from void to rtx, returning x. When reading
31599 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
31600 larger size containing struct block_symbol.
31602 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
31604 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
31605 -mfloat128-hardware without -m64.
31607 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
31610 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
31611 entries to the case statement that marks unsigned arguments to
31612 overloaded functions.
31614 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
31616 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
31617 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
31619 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
31622 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
31623 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
31625 2017-03-10 Martin Liska <mliska@suse.cz>
31629 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
31630 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
31633 2017-03-10 Olivier Hainque <hainque@adacore.com>
31635 * tree-switch-conversion (array_value_type): Start by resetting
31636 candidate type to it's main variant.
31638 2017-03-10 Jakub Jelinek <jakub@redhat.com>
31640 PR rtl-optimization/79909
31641 * combine.c (try_combine): Use simplify_replace_rtx on individual
31642 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
31643 of the whole CALL_INSN_FUNCTION_USAGE.
31645 PR tree-optimization/79972
31646 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
31647 get_range_info on SSA_NAMEs. Formatting fixes.
31649 2017-03-10 Richard Biener <rguenther@suse.de>
31650 Jakub Jelinek <jakub@redhat.com>
31652 PR tree-optimization/77975
31653 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
31654 edge to be constant.
31655 (get_val_for): For constant x return it. Formatting fix.
31656 (loop_niter_by_eval): Avoid pointless looping if the next iteration
31657 would use the same bases as the current one.
31659 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31661 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
31662 instead of vec_select for V1TImode.
31663 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
31665 (VSX_LE_128): Add V1TI to this mode iterator.
31666 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
31667 (*vsx_le_perm_store_<mode>): Likewise.
31668 (pre-reload splitter for VSX stores): Likewise.
31669 (post-reload splitter for VSX stores): Likewise.
31670 (*vsx_xxpermdi2_le_<mode>): Likewise.
31671 (*vsx_lxvd2x2_le_<mode>): Likewise.
31672 (*vsx_stxvd2x2_le_<mode>): Likewise.
31674 2017-03-09 Michael Eager <eager@eagercon.com>
31676 Correct failures with --enable-checking=yes,rtl.
31678 * config/microblaze/microblaze.c (microblaze_expand_shift):
31679 Replace GET_CODE test with CONST_INT_P and INTVAL test with
31680 test for const0_rtx.
31681 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
31682 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
31684 2017-03-09 Richard Biener <rguenther@suse.de>
31686 PR tree-optimization/79977
31687 * graphite-scop-detection.c (scop_detection::merge_sese):
31688 Handle the case of extra exits to blocks dominating the entry.
31690 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
31692 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
31695 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
31697 PR rtl-optimization/79949
31698 * lra-constraints.c (process_alt_operands): Check memory when
31699 trying to predict a cycle. Print about the overall increase.
31701 2017-03-09 Richard Biener <rguenther@suse.de>
31703 PR middle-end/79971
31704 * gimple-expr.c (useless_type_conversion_p): Preserve
31705 TYPE_SATURATING for fixed-point types.
31707 2017-03-09 Richard Biener <rguenther@suse.de>
31710 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
31711 alignment of BLKmode params.
31713 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31716 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
31717 (VALL_NO_V2Q): Likewise.
31719 * config/aarch64/aarch64-simd.md
31720 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
31722 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
31723 VALL_NO_V2Q mode iterator.
31724 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
31726 2017-03-09 Martin Liska <mliska@suse.cz>
31728 PR tree-optimization/79631
31729 * tree-chkp-opt.c (chkp_is_constant_addr): Call
31730 tree_int_cst_sign_bit just for INTEGER constants.
31732 2017-03-09 Martin Liska <mliska@suse.cz>
31736 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
31739 2017-03-09 Marek Polacek <polacek@redhat.com>
31742 * tree.c (inchash::add_expr): Handle TREE_VEC.
31744 2017-03-09 Martin Liska <mliska@suse.cz>
31747 (chkp_narrow_size_and_offset): New function.
31748 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
31749 (void chkp_parse_bit_field_ref): New function.
31750 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
31751 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
31753 2017-03-09 Martin Liska <mliska@suse.cz>
31756 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
31757 (chkp_find_bounds_1): Remove gcc_unreachable.
31759 2017-03-09 Jakub Jelinek <jakub@redhat.com>
31762 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
31763 BUILT_IN_SYNC*, determine the access type from the size suffix and
31764 always build a MEM_REF with that type. Handle forgotten
31765 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
31768 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
31769 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
31770 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
31771 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
31772 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
31773 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
31774 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
31775 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
31776 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
31777 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
31778 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
31779 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
31780 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
31781 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
31782 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
31783 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
31784 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
31785 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
31786 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
31787 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
31788 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
31789 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
31790 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
31791 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
31792 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
31793 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
31794 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
31795 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
31796 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
31797 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
31798 definitions outside of __OPTIMIZE__ guarded section.
31801 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
31802 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
31803 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
31804 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
31807 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31809 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
31810 ("vfenez<mode>"): Add missing constraints.
31812 2017-03-08 Martin Sebor <msebor@redhat.com>
31815 * config/nds32/nds32.c (nds32_option_override):
31816 Fix misspelled diagnostic.
31818 2017-03-08 Jakub Jelinek <jakub@redhat.com>
31821 * gimplify.c (gimplify_omp_for): Replace index var in outer
31822 taskloop statement with an artificial variable and add
31823 OMP_CLAUSE_PRIVATE clause for it.
31825 2017-03-08 Richard Biener <rguenther@suse.de>
31827 PR tree-optimization/79955
31828 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
31829 for accesses that are completely outside of the variable.
31831 2017-03-08 Andrew Haley <aph@redhat.com>
31833 PR tree-optimization/79943
31834 * tree-ssa-loop-split.c (compute_new_first_bound): When
31835 calculating the new upper bound, (END-BEG) should be added, not
31838 2017-03-08 Jakub Jelinek <jakub@redhat.com>
31840 * config/avr/avr.md (setmemhi): Make sure match_dup
31841 operand number comes before match_scratch.
31843 2017-03-08 Richard Biener <rguenther@suse.de>
31845 PR tree-optimization/79920
31846 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
31847 with ncopies == 1 to ...
31848 (vect_transform_slp_perm_load): ... here. Properly compute
31849 all element loads by iterating VF times over the group. Do
31850 not handle ncopies (computed in a broken way) in
31851 vect_create_mask_and_perm.
31853 2017-03-08 Jakub Jelinek <jakub@redhat.com>
31856 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
31857 is a uniform vector, use uniform_vector_p return value instead of
31858 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
31860 2017-03-07 Marek Polacek <polacek@redhat.com>
31862 PR middle-end/79809
31863 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
31864 (alloca_call_type): Likewise.
31866 2017-03-07 Martin Liska <mliska@suse.cz>
31868 * gcov.c (process_args): Put comment to correct location.
31870 2017-03-07 Martin Liska <mliska@suse.cz>
31872 PR middle-end/68270
31873 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
31874 Use array_at_struct_end_p instead of DECL_CHAIN (field).
31875 (chkp_narrow_bounds_for_field): Likewise.
31876 (chkp_parse_array_and_component_ref): Pass one more argument to
31879 2017-03-07 Richard Biener <rguenther@suse.de>
31881 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
31884 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
31886 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
31887 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
31889 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31892 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
31893 to end of description.
31894 (PARAM_MAX_STORES_TO_MERGE): Likewise.
31896 2017-03-07 Jakub Jelinek <jakub@redhat.com>
31898 PR rtl-optimization/79901
31899 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
31901 (*avx512f_<code><mode>3<mask_name>): ... this.
31902 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
31903 iterator instead of VI8_AVX2_AVX512BW.
31905 PR rtl-optimization/79901
31906 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
31907 min/max expander, expand it using expand_vec_cond_expr.
31910 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
31913 2017-03-06 Jakub Jelinek <jakub@redhat.com>
31916 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
31917 to void * for PCH reasons.
31918 * dwarf2out.c (output_loc_operands, output_die): Cast
31919 v.val_vec.array to unsigned char *.
31921 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
31924 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
31927 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
31929 PR rtl-optimization/79571
31930 * lra-constraints.c (process_alt_operands): Calculate static
31931 reject and subtract it from overall when only addresses will be
31934 2017-03-06 Julia Koval <julia.koval@intel.com>
31937 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
31938 incoming stack boundary to 128 for 64-bit targets.
31940 2017-03-06 Richard Biener <rguenther@suse.de>
31942 PR tree-optimization/79894
31943 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
31944 to NULL after folding it.
31946 2017-03-06 Richard Biener <rguenther@suse.de>
31948 PR tree-optimization/79824
31949 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
31950 check disabling peeling for gaps.
31952 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
31954 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
31955 attributes): Document gettimeofday.
31957 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
31959 * config/s390/s390.c (s390_option_override_internal): Set
31960 PARAM_MIN_VECT_LOOP_BOUND
31962 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
31964 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
31965 * config/s390/s390.md: Likewise.
31967 2017-03-06 Jakub Jelinek <jakub@redhat.com>
31970 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
31971 (<avx2_avx512>_perm<mode>): Rename to ...
31972 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
31974 (<avx512>_perm<mode>_mask): Rename to ...
31975 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
31977 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
31978 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
31979 instead of VI8F_256_512.
31980 (avx512f_perm<mode>): New define_expand.
31981 (avx512f_perm<mode>_mask): Likewise.
31982 (avx512f_perm<mode>_1<mask_name>): New define_insn.
31983 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
31985 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
31987 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
31988 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
31990 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
31992 2017-03-06 Martin Liska <mliska@suse.cz>
31995 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
31996 when having a SSA NAME w/o VAR_DECL assigned to it.
31998 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32000 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
32001 msa_dpsub_<su>_d): Fix MODE for vec_select.
32003 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32005 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
32007 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
32009 2017-03-06 Richard Biener <rguenther@suse.de>
32011 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
32012 * plugin.c (register_plugin_info): Likewise.
32013 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
32015 2017-03-05 Jakub Jelinek <jakub@redhat.com>
32017 * config/i386/sse.md (sse_storehps, sse_storelps,
32018 avx_<castmode><avxsizesuffix>_<castmode>,
32019 avx512f_<castmode><avxsizesuffix>_<castmode>,
32020 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
32021 in condition that at least one operand is not a MEM.
32023 2017-03-03 Jakub Jelinek <jakub@redhat.com>
32025 PR middle-end/79805
32026 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
32027 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
32029 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
32030 gimple_call_nothrow_p flag based on whether original builtin can throw.
32031 If it can, emit following stmts on the fallthrough edge.
32032 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
32033 don't create new bb if inserting just debug stmts on the edge, try to
32034 insert them on the fallthru bb or just reset debug stmts.
32036 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
32039 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
32040 restore recog_data (including the operand rtxes inside it) around
32041 the call to get_insn_template.
32043 2017-03-03 Martin Sebor <msebor@redhat.com>
32045 PR tree-optimization/79699
32046 * context.c (context::~context): Free MPFR caches to avoid
32047 a memory leak on program exit.
32049 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32051 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
32052 Use wide_int::ulow () instead of .elt (0).
32054 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
32056 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
32057 (*pushxf): Limit oF constraint to 32bit targets and add oC
32058 constraint for 64bit targets.
32059 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
32060 (*pushdf): Change rmF constraint to rmC.
32062 2017-03-03 Martin Liska <mliska@suse.cz>
32064 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
32065 Remove unused variable.
32067 2017-03-03 Jakub Jelinek <jakub@redhat.com>
32070 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
32071 is a memory operand, increase num_memory.
32072 (ix86_expand_args_builtin): Likewise.
32074 2017-03-03 Jan Hubicka <jh@suse.cz>
32077 * ipa-devirt.c (maybe_record_node): Properly handle
32078 __cxa_pure_virtual visibility.
32080 2017-03-03 Martin Liska <mliska@suse.cz>
32082 PR tree-optimization/79803
32083 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
32085 (pass_loop_prefetch::execute): Disabled optimization if an
32086 assumption about L1 cache size is not met.
32088 2017-03-03 Martin Liska <mliska@suse.cz>
32090 PR rtl-optimization/79574
32091 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
32092 (hash_scan_set): Likewise.
32093 (dump_hash_table): Likewise.
32094 (hoist_code): Likewise.
32096 2017-03-03 Richard Biener <rguenther@suse.de>
32098 * fixed-value.c (fixed_from_string): Restore use of elt (1)
32099 in place of uhigh ().
32100 (fixed_convert_from_real): Likewise.
32102 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
32105 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
32107 2017-03-03 Richard Biener <rguenther@suse.de>
32109 PR middle-end/79818
32110 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
32111 TYPE_OVERFLOW_UNDEFINED check.
32113 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32115 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
32117 (vector_ae_<mode>_p): Likewise.
32118 (vector_nez_<mode>_p): Likewise.
32119 (vector_ne_v2di_p): Likewise.
32120 (vector_ae_v2di_p): Likewise.
32121 (vector_ne_<mode>_p): Likewise.
32122 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
32124 (vsx_tsqrt<mode>2_fe): Likewise.
32126 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
32129 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
32131 2017-03-02 Jakub Jelinek <jakub@redhat.com>
32133 PR rtl-optimization/79780
32134 * cprop.c (one_cprop_pass): When second and further conditional trap
32135 in a single basic block is turned into an unconditional trap, turn it
32136 into a deleted note to avoid RTL verification failures.
32138 2017-03-02 Richard Biener <rguenther@suse.de>
32140 * fold-const.c (const_binop): Use ulow () instead of elt (0).
32142 2017-03-02 Richard Biener <rguenther@suse.de>
32144 PR tree-optimization/79345
32146 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
32147 param and abort the walk, returning -1 if it is hit.
32148 (walk_aliased_vdefs): Take a limit param and pass it on.
32149 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
32150 defaulting to 0 and return a signed int.
32151 * tree-ssa-uninit.c (struct check_defs_data): New struct.
32152 (check_defs): New helper.
32153 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
32154 about uninitialized memory.
32155 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
32156 bogus uninitialized warning.
32157 (fixed_convert_from_real): Likewise.
32159 2017-03-02 Bin Cheng <bin.cheng@arm.com>
32161 PR tree-optimization/66768
32162 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
32163 iv_use if base object can't be determined.
32165 2017-03-02 Jakub Jelinek <jakub@redhat.com>
32167 PR tree-optimization/79345
32168 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
32169 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
32170 (get_pattern_stats): Initialize it.
32171 * genemit.c (gen_expand): Verify match_scratch numbers come after
32172 match_operand/match_dup numbers.
32173 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
32174 match_scratch numbers.
32175 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
32177 * config/s390/s390.md (trunctdsd2): Likewise.
32179 2017-03-02 Richard Biener <rguenther@suse.de>
32181 * wide-int.h (wide_int_storage::operator=): Implement in terms
32184 2017-03-02 Richard Biener <rguenther@suse.de>
32186 PR tree-optimization/79777
32187 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
32188 the to insert expression to sth existing.
32190 2017-03-01 Martin Sebor <msebor@redhat.com>
32192 PR middle-end/79692
32193 * gimple-ssa-sprintf.c
32194 (directive::known_width_and_precision): New function.
32195 (format_integer): Use it.
32196 (get_mpfr_format_length): Consider the full range of precision
32197 when computing %g output with the # flag. Set the likely byte
32198 count to 3 rather than 1 when precision is indeterminate.
32199 (format_floating): Correct the lower bound of precision.
32201 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32203 * doc/invoke.texi: Document default code model for 64-bit Linux.
32205 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32208 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
32209 udiv rather than div since input pattern is unsigned.
32211 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
32213 * config/i386/i386.c (print_reg): Warn for values of
32214 unsupported size in integer register.
32216 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
32219 * config/rs6000/predicates.md (current_file_function_operand): Do
32220 not allow self calls to be local if the function is replaceable.
32222 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
32225 * config/rs6000/altivec.h (vec_ctz and others): Change the
32226 preprocessor macro that controls conditional compilation from
32227 _ARCH_PWR9 to __POWER9_VECTOR__.
32228 (vec_all_ne): Change parameterization of __altivec_scalar_pred
32229 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
32230 control (instead of _ARCH_PWR9 control) so that template
32231 definition uses power9-specific function.
32232 (vec_any_eq): Likewise.
32233 (vec_all_ne): Change macro definition to use a power9-specific
32234 expansion under #ifdef __POWER9_VECTOR__ control (instead of
32235 _ARCH_PWR9 control).
32236 (vec_any_eq) Likewise.
32237 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
32238 expansion for CMPNEF to remove support for xvcmpnesp instruction.
32239 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
32240 support for xvcmpnedp instruction.
32241 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
32242 macro expansion so that Power9 implementation of vec_all_ne does
32243 not use the AltiVec predicate framework.
32244 (VCMPNEH_P): Likewise.
32245 (VCMPNEW_P): Likewise.
32246 (VCMPNED_P): Likewise.
32247 (VCMPNEFP_P): Likewise.
32248 (VCMPNEDP_P): Likewise.
32249 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
32250 implementation of vec_any_eq to not use AltiVec predicate
32252 (VCMPAEH_P): Likewise.
32253 (VCMPAEW_P): Likewise.
32254 (VCMPAED_P): Likewise.
32255 (VCMPAEFP_P): Likewise.
32256 (VCMPAEDP_P): Likewise.
32257 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
32258 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
32259 not use the AltiVec predicate framework.
32260 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
32261 of vec_any_eq to not use AltiVec predicate framework.
32262 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
32263 support for predefined __POWER9_VECTOR__ macro to indicate that
32264 Power9 instruction selection is enabled.
32265 (altivec_overloaded_builtins): Remove extraneous
32266 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
32267 function argument types RS6000_BTI_bool_V16QI and
32268 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
32269 entry for overloaded function argument types RS6000_BTI_bool_V4SI
32270 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
32271 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
32272 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
32273 Power9 for implementations of vec_cmpne. Change the signature for
32274 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
32275 (representing vec_all_ne) to remove the previously described first
32276 argument of type RS6000_BTI_INTSI, as this was an artifact of
32277 reliance on the AltiVec predicate framework, which is no longer
32278 used in the implementation of these functions. Add
32279 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
32280 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
32281 since, unlike the AltiVec predicate framework implementation, we
32282 do not share function descriptors between vec_alle and vec_anyeq.
32283 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
32284 set of modes that receive special treatment even when
32285 TARGET_P9_VECTOR is true. The special treatment emits code that
32286 does not depend on Power9 instructions.
32287 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
32288 define_expand to not rely on AltiVec predicate framework.
32289 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
32291 (vector_ne_v2di_p): Change this define_expand to not rely on
32292 AltiVec predicate framework.
32293 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
32295 (vector_ne_<mode>_p): Change this define_expand to not rely on
32296 AltiVec predicate framework.
32297 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
32299 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
32300 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
32301 define_insn pattern.
32302 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
32303 define_insn pattern because the xvcmpne<VSs>. instruction is not
32305 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
32306 instruction is not supported.
32308 2017-03-01 Jakub Jelinek <jakub@redhat.com>
32310 * config/nvptx/nvptx.c: Include intl.h.
32312 2017-03-01 Martin Jambor <mjambor@suse.cz>
32315 * ipa-prop.h (ipa_bits): Removed field known.
32316 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
32317 to pointers. Adjusted their comments to warn about their sharing.
32318 (ipcp_transformation_summary): Change bits to a vector of pointers.
32319 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
32320 (ipa_get_ipa_bits_for_value): Declare.
32321 * tree-vrp.h (value_range): Mark as GTY((for_user)).
32322 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
32323 (ipa_bits_hash_table): Likewise.
32324 (ipa_vr_ggc_hash_traits): Likewise.
32325 (ipa_vr_hash_table): Likewise.
32326 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
32327 being pointers and vr_known being removed.
32328 (ipa_set_jf_unknown): Likewise.
32329 (ipa_get_ipa_bits_for_value): New function.
32330 (ipa_set_jfunc_bits): Likewise.
32331 (ipa_get_value_range): New overloaded functions.
32332 (ipa_set_jfunc_vr): Likewise.
32333 (ipa_compute_jump_functions_for_edge): Use the above functions to
32334 construct bits and vr parts of jump functions.
32335 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
32336 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
32338 (ipcp_grow_transformations_if_necessary): Also allocate
32339 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
32341 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
32342 them. Fix too long lines.
32343 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
32344 vr_known being removed.
32345 (ipa_read_jump_function): Use new setter functions to construct bits
32346 and vr parts of jump functions or set them to NULL.
32347 (write_ipcp_transformation_info): Adjust for bits being pointers.
32348 (read_ipcp_transformation_info): Likewise.
32349 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
32351 Include gt-ipa-prop.h.
32352 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
32354 (ipcp_store_bits_results): Likewise.
32355 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
32356 Do not write to existing jump functions but use a temporary instead.
32358 2017-03-01 Jakub Jelinek <jakub@redhat.com>
32361 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
32362 attempt to use its first operand as BIT_FIELD_REF base.
32364 2017-03-01 Richard Biener <rguenther@suse.de>
32366 PR middle-end/79721
32367 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
32368 interpolating formula in wrapping arithmetic.
32369 (chrec_apply): Convert chrec_evaluate return value to wanted type.
32371 2017-03-01 Jakub Jelinek <jakub@redhat.com>
32373 PR tree-optimization/79734
32374 * tree-vect-generic.c (expand_vector_condition): Optimize
32375 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
32376 Handle VEC_COND_EXPR where comparison has different inner width from
32377 type's inner width.
32379 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
32381 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
32382 markup, and similar issues. Remove @opindex entries for things
32383 that aren't options. Add missing -mmpy-option entries.
32385 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32387 PR tree-optimization/79737
32388 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
32389 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
32390 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
32391 instead of byte_size. Formatting fix.
32392 (shift_bytes_in_array_right): Formatting fix.
32394 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
32397 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
32398 condition on optimize for the leaf function test.
32400 2017-02-28 Martin Liska <mliska@suse.cz>
32403 * read-rtl-function.c (function_reader::handle_unknown_directive):
32404 Bail out when one uses -flto.
32406 2017-02-28 Martin Liska <mliska@suse.cz>
32408 * common.opt: Replace space with tabular for options of <number>
32410 * config/i386/i386.opt: Show <number> value for
32411 -mlarge-data-threshold.
32412 * opts.c (print_filtered_help): Do not display number in hexadecimal
32415 2017-02-28 Martin Liska <mliska@suse.cz>
32417 * common.opt: Fix --help=option -Q for options which are of
32420 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
32422 * config/i386/i386.c (print_reg): Error out for values
32423 of 8-bit size in invalid integer register.
32425 2017-02-28 Martin Sebor <msebor@redhat.com>
32427 PR tree-optimization/79691
32428 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
32430 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32433 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
32434 gcc_unreachable with output_operand_lossage.
32436 2017-02-28 Richard Biener <rguenther@suse.de>
32438 PR tree-optimization/79740
32439 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
32441 (visit_nary_op): Insert the nary into the hashtable if we
32442 pattern-matched sth.
32443 * tree-ssa-pre.c (eliminate_insert): Robustify.
32445 2017-02-28 Richard Biener <rguenther@suse.de>
32447 PR middle-end/79731
32448 * fold-const.c (decode_field_reference): Reject out-of-bound
32451 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32453 * config/i386/i386.c: Include intl.h.
32454 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
32455 instead of just cond ? "..." : "...".
32456 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
32457 * coverage.c (read_counts_file): Likewise.
32458 * omp-offload.c: Include intl.h.
32459 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
32460 of just cond ? "..." : "...".
32461 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
32462 of just cond ? "..." : "...".
32464 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
32467 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
32469 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
32470 'tune for' CPU name.
32471 * config/arm/arm-cpu-data.h: Regenerated.
32473 2017-02-28 Richard Biener <rguenther@suse.de>
32475 PR tree-optimization/79732
32476 * tree-inline.c (expand_call_inline): Do not shadow var.
32478 2017-02-28 Richard Biener <rguenther@suse.de>
32480 PR tree-optimization/79723
32481 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
32482 address-space properly.
32484 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
32486 * doc/optinfo.texi (Optimization groups): Fix option used for
32488 * doc/invoke.texi (-fopt-info): Document "omp".
32489 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
32490 (OPTGROUP_ALL): Add OPTGROUP_OMP.
32491 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
32492 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
32493 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
32495 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
32497 * dumpfile.c (optgroup_options): Instead of "openmp", associate
32498 OPTGROUP_OMP with "omp".
32500 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
32503 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
32504 for arithmetic shift of unsigned V2DI.
32506 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
32508 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
32509 arc/linux.h headers.
32510 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
32511 (LINK_SPEC): Likewise.
32512 (ARC_TLS_EXTRA_START_SPEC): Likewise.
32513 (EXTRA_SPECS): Likewise.
32514 (STARTFILE_SPEC): Likewise.
32515 (ENDFILE_SPEC): Likewise.
32516 (LIB_SPEC): Likewise.
32517 (TARGET_SDATA_DEFAULT): Likewise.
32518 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
32519 (MULTILIB_DEFAULTS): Likewise.
32520 (DWARF2_UNWIND_INFO): Likewise.
32521 * config/arc/big.h: New file.
32522 * config/arc/elf.h: Likewise.
32523 * config/arc/linux.h: Likewise.
32524 * config/arc/t-uClibc: Remove.
32526 2017-02-27 Bin Cheng <bin.cheng@arm.com>
32528 PR tree-optimization/77536
32529 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
32530 (tree_transform_and_unroll_loop): Use above function to compute the
32531 estimated niter of unrolled loop and use it when scaling profile.
32532 Also use count info rather than frequency if it's non-zero.
32533 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
32534 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
32535 (vect_transform_loop): Call above function.
32537 2017-02-27 Richard Biener <rguenther@suse.de>
32539 PR tree-optimization/45397
32540 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
32541 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
32542 (visit_nary_op): Add pattern matching for CSEing sign-changed
32543 or truncated operations with wider ones.
32545 2017-02-27 Richard Biener <rguenther@suse.de>
32547 PR tree-optimization/79690
32548 * tree-vect-stmts.c (vectorizable_store): Use vector type
32549 built from the DR with address-space.
32551 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
32553 * doc/invoke.texi (Optimize Options): Refine the description
32554 of asan-use-after-return.
32556 2017-02-25 Alan Modra <amodra@gmail.com>
32558 PR rtl-optimization/79584
32559 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
32560 base, not ad->base_term, the reg within base. Remove assertion
32561 that ad->base == ad->base_term. Replace gen_int_mode using
32562 bogus mode with const0_rtx.
32564 2017-02-25 Jakub Jelinek <jakub@redhat.com>
32566 PR middle-end/79396
32567 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
32568 FMA_EXPR like tcc_binary or tcc_unary.
32570 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
32573 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
32575 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
32576 (output_loc_operands): Handle DW_OP_call_ref and
32577 DW_OP_GNU_variable_value.
32578 (struct variable_value_struct): New type.
32579 (struct variable_value_hasher): Likewise.
32580 (variable_value_hash): New variable.
32581 (string_types): Remove.
32582 (copy_loc_descr): New function.
32583 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
32584 (prepend_loc_descr_to_each): New function.
32585 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
32586 instead of add_loc_descr_to_each if the first argument is single
32587 location list and the second has multiple.
32588 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
32589 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
32590 when looking for variable value which doesn't have other location info.
32591 (loc_list_from_tree): Formatting fix.
32592 (gen_array_type_die): Simplify DW_AT_string_length handling.
32593 (adjust_string_types): Remove.
32594 (gen_subprogram_die): Don't call adjust_string_types nor test/set
32595 string_types. Call resolve_variable_values.
32596 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
32597 (resolve_addr_in_expr): Likewise. Add A argument.
32598 (copy_deref_exprloc): Remove deref argument. Adjust for the
32599 original expression being DW_OP_GNU_variable_value with optionally
32600 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
32601 optionally after it.
32602 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
32603 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
32604 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
32605 (variable_value_hasher::hash, variable_value_hasher::equal): New
32607 (resolve_variable_value_in_expr, resolve_variable_value,
32608 resolve_variable_values, note_variable_value_in_expr,
32609 note_variable_value): New functions.
32610 (dwarf2out_early_finish): Call note_variable_value on all toplevel
32613 2017-02-24 Jakub Jelinek <jakub@redhat.com>
32616 * opts.h (handle_generated_option): Add GENERATED_P argument.
32617 * opts-common.c (handle_option): Adjust function comment.
32618 (handle_generated_option): Add GENERATED_P argument, pass it to
32620 (control_warning_option): Pass false to handle_generated_option
32622 * opts.c (maybe_default_option): Pass true to handle_generated_option
32624 * optc-gen.awk: Likewise.
32626 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32628 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
32629 a REG, look at the REG it is a SUBREG of.
32630 (splitter for cmpeqsi_t): Ditto.
32632 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32634 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
32635 the special USEs with the pattern of the insn, not the insn itself.
32637 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
32640 * doc/invoke.texi: Document -mload-store-pairs.
32642 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32643 Sandra Loosemore <sandra@codesourcery.com>
32645 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
32646 argument isn't a CONST_INT.
32647 (nios2_alternate_compare_const): Assert op is a CONST_INT.
32648 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
32649 (nios2_validate_compare): Bypass alternate compare logic if *op2
32650 is not a CONST_INT.
32651 (ldstwm_operation_p): Return false if first_base is not a REG or
32652 if first_offset is not a CONST_INT.
32654 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32656 * config/cris/cris.md: Use correct operand in a define_peephole2.
32658 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32660 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
32662 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32664 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
32665 this_insn if it is an INSN or JUMP_INSN.
32666 (force_offsettable): Look at base, not at addr.
32667 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
32668 on things that aren't necessarily CONST_INTs.
32670 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
32672 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
32673 -mfpmath=sse is the default also for x86-32 targets with SSE2
32674 instruction set when @option{-ffast-math} is enabled
32676 2017-02-24 Jeff Law <law@redhat.com>
32678 PR rtl-optimizatoin/79286
32679 * ira.c (update_equiv_regs): Drop may_trap_p exception to
32682 2017-02-24 Richard Biener <rguenther@suse.de>
32684 PR tree-optimization/79389
32685 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
32688 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
32690 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
32691 function comment to reflect reality.
32692 (loop_exits_before_overflow): Fix typo in function description.
32694 2017-02-24 Richard Biener <rguenther@suse.de>
32696 PR tree-optimization/79389
32697 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
32698 properly that a threading opportunity exists. Detect conditional
32699 copy/constant propagation opportunities.
32701 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
32703 * config/visium/visium.md (type): Add trap.
32704 (b): New mode attribute.
32705 (*btst): Rename into...
32706 (*btst<mode>): ...this and adjust.
32707 (*cbranchsi4_btst_insn): Rename into...
32708 (*cbranch<mode>4_btst_insn): ...this and adjust.
32709 (trap): New define_insn.
32711 2017-02-23 Jakub Jelinek <jakub@redhat.com>
32713 PR tree-optimization/79389
32714 * ifcvt.c (struct noce_if_info): Add rev_cond field.
32715 (noce_reversed_cond_code): New function.
32716 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
32717 reversed_comparison_code. Formatting fix.
32718 (noce_try_store_flag): Test rev_cond != NULL in addition to
32719 reversed_comparison_code.
32720 (noce_try_store_flag_constants): Likewise.
32721 (noce_try_store_flag_mask): Likewise.
32722 (noce_try_addcc): Use rev_cond if non-NULL instead of
32723 reversed_comparison_code.
32724 (noce_try_cmove_arith): Likewise. Formatting fixes.
32725 (noce_try_minmax, noce_try_abs): Clear rev_cond.
32726 (noce_find_if_block): Initialize rev_cond.
32727 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
32728 instead of false as last argument never attempt to reverse it
32731 2017-02-23 Bin Cheng <bin.cheng@arm.com>
32733 PR tree-optimization/79663
32734 * tree-predcom.c (combine_chains): Process refs in reverse order
32735 only for ZERO length chains, and add explaining comment.
32737 2017-02-23 Jeff Law <law@redhat.com>
32739 PR tree-optimization/79578
32740 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
32741 in call to operand_equal_p.
32743 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
32746 * config/i386/cpuid.h: Fix another undefined behavior.
32748 2017-02-23 Richard Biener <rguenther@suse.de>
32750 PR tree-optimization/79683
32751 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
32752 vector types for data-refs.
32754 2017-02-23 Martin Liska <mliska@suse.cz>
32756 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
32758 2017-02-23 Jakub Jelinek <jakub@redhat.com>
32760 PR middle-end/79665
32761 * internal-fn.c (get_range_pos_neg): Moved to ...
32762 * tree.c (get_range_pos_neg): ... here. No longer static.
32763 * tree.h (get_range_pos_neg): New prototype.
32764 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
32765 are known to be in between 0 and signed maximum inclusive, try to
32766 expand both unsigned and signed divmod and use the cheaper one from
32769 2017-02-22 Jeff Law <law@redhat.com>
32771 PR tree-optimization/79578
32772 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
32773 to compare base operands.
32775 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
32778 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
32779 gpc_reg_operand instead of fpr_reg_operand.
32781 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
32783 * config/mips/mips.c (mips_return_in_memory): Force FP
32784 vector types to be returned in memory for o32 ABI.
32786 2017-02-22 Jakub Jelinek <jakub@redhat.com>
32788 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
32789 instead of DW_TAG_member for static data member declarations and don't
32790 set no_linkage_name for static inline data members.
32791 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
32794 2017-02-22 Martin Liska <mliska@suse.cz>
32796 * doc/invoke.texi: Replace inequality signs with square brackets
32799 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
32802 * lra-constraints.c (simplify_operand_subreg): Handle
32803 WORD_REGISTER_OPERATIONS targets.
32805 2017-02-22 Jakub Jelinek <jakub@redhat.com>
32808 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
32809 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
32810 elimination by swapping fld*.
32812 2017-02-22 Richard Biener <rguenther@suse.de>
32814 PR tree-optimization/79673
32815 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
32816 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
32817 irrelevant address-space qualifiers and avoiding a
32818 ADDR_SPACE_CONVERT_EXPR from fold_convert.
32820 2017-02-22 Richard Biener <rguenther@suse.de>
32822 PR tree-optimization/79666
32823 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
32824 to not symbolically negate if that may introduce undefined
32827 2017-02-22 Martin Liska <mliska@suse.cz>
32830 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
32831 * data-streamer-out.c (streamer_write_gcov_count_stream):
32833 * value-prof.c (stream_out_histogram_value): Make assert more
32834 precise based on type of counter.
32836 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
32839 * config/i386/i386.md (standard_x87sse_constant_load splitter):
32840 Use nonimmediate_operand instead of memory_operand for operand 1.
32841 (float-extend standard_x87sse_constant_load splitter): Ditto.
32843 2017-02-21 Jeff Law <law@redhat.com>
32845 PR tree-optimization/79621
32846 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
32847 blocks with edges to themselves.
32849 2017-02-21 Jakub Jelinek <jakub@redhat.com>
32852 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
32853 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
32854 Use gimple_call_builtin_p.
32857 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
32858 on temporarily removed DEBUG_INSNs.
32860 PR tree-optimization/79649
32861 * tree-loop-distribution.c (classify_partition): Give up on
32862 non-generic address space loads/stores.
32864 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
32866 * doc/loop.texi (Loop manipulation): Remove nonexistent
32867 tree_ssa_loop_version from the documentation.
32868 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
32870 2017-02-21 Jakub Jelinek <jakub@redhat.com>
32873 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
32874 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
32875 * config/rs6000/rs6000.c: Include except.h.
32876 (rs6000_expand_split_stack_prologue): Call
32877 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
32879 2017-02-21 Martin Jambor <mjambor@suse.cz>
32882 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
32883 have been analyzed.
32885 2017-02-21 Martin Jambor <mjambor@suse.cz>
32887 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
32888 for backward compatibility only.
32889 * doc/invoke.texi (Option Summary): Remove all references to
32890 -fipa-cp-alignment.
32892 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
32896 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32898 * lra-constraints.c (curr_insn_transform): Handle
32899 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
32901 2017-02-21 Martin Liska <mliska@suse.cz>
32903 * config/i386/i386.opt: Replace -masm-dialect with -masm.
32905 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
32907 PR translation/79638
32908 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
32910 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
32913 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
32914 (arm_function_ok_for_sibcall): Return false for an indirect call by
32915 descriptor if all the argument registers are used.
32916 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
32917 alignment of the function.
32919 2017-02-21 Jakub Jelinek <jakub@redhat.com>
32921 PR tree-optimization/61441
32922 * simplify-rtx.c (simplify_const_unary_operation): For
32923 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
32924 the sNaN unmodified.
32926 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
32928 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32929 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32930 instead of SYSTEM_HEADER_DIR.
32932 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
32933 Martin Liška <mliska@suse.cz>
32935 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
32936 Fix typos and grammar, use active voice, and clarify.
32938 2017-02-20 Marek Polacek <polacek@redhat.com>
32940 PR middle-end/79537
32941 * gimplify.c (gimplify_expr): Handle unused *&&L;.
32944 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
32946 2017-02-20 Jakub Jelinek <jakub@redhat.com>
32949 * config/i386/i386.c (ix86_expand_builtin): Handle
32950 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
32951 ix86_builtins_isa[fcode].isa as a requirement of those
32952 flags and any other flag in the bitmask.
32953 (ix86_init_mmx_sse_builtins): Use 0 instead of
32954 ~OPTION_MASK_ISA_64BIT as mask.
32955 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
32956 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
32957 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
32958 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
32960 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32963 * lra-constraints.c (split_reg): Check requested split mode
32964 is supported by the register.
32966 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32968 * lra-constraints.c (simplify_operand_subreg): Remove early
32971 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32974 * lra-constraints.c (curr_insn_transform): Tighten condition
32975 for converting SUBREG reloads from OP_OUT to OP_INOUT.
32977 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32980 * lra-constraints.c (curr_insn_transform): Handle
32981 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
32983 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
32986 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
32988 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
32990 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
32993 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
32996 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32998 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
33001 2017-02-18 Jakub Jelinek <jakub@redhat.com>
33003 * final.c (last_columnnum, override_columnnum): New variables.
33004 (final_start_function): Set last_columnnum, pass it to begin_prologue
33005 hook and pass 0 to dwarf2out_begin_prologue.
33006 (final_scan_insn): Update override_columnnum. Pass last_columnnum
33007 to source_line debug hook.
33008 (notice_source_line): Compute last_columnnum and for debug_column_info
33009 return true on column changes.
33010 * debug.h (struct gcc_debug_hooks): Add column argument to
33011 source_line and begin_prologue hooks.
33012 (debug_nothing_int_charstar_int_bool): Remove prototype.
33013 (debug_nothing_int_int_charstar,
33014 debug_nothing_int_int_charstar_int_bool): New prototypes.
33015 (dwarf2out_begin_prologue): Add column argument.
33016 * debug.c (do_nothing_debug_hooks): Adjust source_line and
33017 begin_prologue hooks.
33018 (debug_nothing_int_charstar_int_bool): Remove.
33019 (debug_nothing_int_int_charstar,
33020 debug_nothing_int_int_charstar_int_bool): New functions.
33021 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
33022 through to dwarf2out_source_line.
33023 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
33024 (dwarf2out_source_line): Add column argument, emit it if requested.
33025 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
33027 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
33028 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
33029 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
33030 through to dwarf2out_begin_prologue.
33031 (vmsdbgout_source_line): Add column argument, pass it through to
33032 dwarf2out_source_line.
33033 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
33034 dbxout_source_line caller.
33035 (dbxout_source_line): Add column argument.
33037 * common.opt (gno-column-info, gcolumn-info): New options.
33038 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
33039 (check_die): Also test for multiple DW_AT_decl_column attributes.
33040 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
33041 DW_AT_decl_column if requested.
33042 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
33044 (gen_variable_die): Likewise.
33045 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
33046 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
33049 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
33050 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
33051 (ix86_handle_option): Handle OPT_m3dnowa.
33052 * doc/invoke.texi (-m3dnowa): Document.
33053 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
33054 -m3dnowa instead of -m3dnow -march=athlon.
33057 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
33058 instead of gcc_assert for K, r and R code checks. Formatting fixes.
33060 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33063 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
33064 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
33065 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
33066 generator for vsx_xxpermdi_<mode>_be.
33067 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
33068 force big-endian semantics.
33069 (vsx_xxpermdi_<mode>_be): New define_expand with same
33070 implementation as previous version of vsx_xxpermdi_<mode>.
33072 2017-02-17 Jakub Jelinek <jakub@redhat.com>
33074 PR tree-optimization/79327
33075 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
33076 variable, its initialization and use.
33078 2017-02-17 Julia Koval <julia.koval@intel.com>
33080 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
33081 (OPTION_MASK_ISA_PKU_UNSET): New.
33082 (ix86_handle_option): Handle -mrdpid.
33083 * config/i386/cpuid.h (bit_RDPID): New.
33084 * config/i386/driver-i386.c (host_detect_local_cpu):
33085 Detect RDPID feature.
33086 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
33087 * config/i386/i386-c.c (ix86_target_macros_internal):
33089 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
33090 (ix86_valid_target_attribute_inner_p): Add "rdpid".
33091 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
33092 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
33093 * config/i386/i386.md (define_insn "rdpid"): New.
33094 * config/i386/i386.opt Add -mrdpid.
33095 * config/i386/immintrin.h (_rdpid_u32): New.
33097 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
33099 PR rtl-optimization/79541
33100 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
33101 instead of transforming it into USE.
33103 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
33105 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
33106 If HONOR_SNANS (SFmode) force the input to a register.
33107 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
33108 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
33109 an frsp or similar insn.
33111 2017-02-17 Martin Liska <mliska@suse.cz>
33113 PR rtl-optimization/79577
33114 * params.def (selsched-max-sched-times): Increase minimum to 1.
33116 2017-02-17 Martin Liska <mliska@suse.cz>
33118 PR rtl-optimization/79574
33119 * gcse.c (want_to_gcse_p): Prevent integer overflow.
33121 2017-02-17 Martin Liska <mliska@suse.cz>
33123 PR tree-optimization/79529
33124 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
33125 ssa_defined_default_def_p to handle cases which are implicitly
33127 * tree-ssa.c (ssa_defined_default_def_p): New function.
33128 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
33129 which are implicitly defined.
33130 * tree-ssa.h (ssa_defined_default_def_p): Declare.
33132 2017-02-17 Richard Biener <rguenther@suse.de>
33134 PR middle-end/79576
33135 * params.def (max-ssa-name-query-depth): Limit to 10.
33137 2017-02-17 Richard Biener <rguenther@suse.de>
33139 PR tree-optimization/79552
33140 * tree-ssa-structalias.c (visit_loadstore): Properly verify
33143 2017-02-17 Richard Biener <rguenther@suse.de>
33146 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
33148 2017-02-17 Marek Polacek <polacek@redhat.com>
33150 PR middle-end/79536
33151 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
33152 (fold_negate_expr): New wrapper.
33154 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
33156 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
33157 Correct terminology and de-emphasize pre-standard behavior.
33159 2017-02-16 Alan Modra <amodra@gmail.com>
33161 PR rtl-optimization/79286
33162 * ira.c (def_dominates_uses): New function.
33163 (update_equiv_regs): Don't create an equivalence for insns that
33164 may trap where the register def does not dominate the use.
33166 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
33168 PR rtl-optimization/78127
33169 * lra.c (lra): Call lra_eliminate before finish the loop after
33172 2017-02-16 Richard Biener <rguenther@suse.de>
33174 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
33176 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
33177 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
33178 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
33179 (isl_val_int_from_wi): New function.
33180 (extract_affine_gmp): Rename to ...
33181 (extract_affine_wi): ... this, take a widest_int.
33182 (extract_affine_int): Just wrap extract_affine_wi.
33183 (add_param_constraints): Use isl_val_int_from_wi.
33184 (add_loop_constraints): Likewise, and extract_affine_wi.
33186 2017-02-15 Jeff Law <law@redhat.com>
33188 PR middle-end/79521
33189 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
33190 ira_init_register_move_cost_if_necessary.
33192 2017-02-15 Martin Sebor <msebor@redhat.com>
33194 PR middle-end/32003
33195 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
33196 removed in a prior commit.
33198 2017-02-15 Bin Cheng <bin.cheng@arm.com>
33200 PR tree-optimization/79347
33201 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
33202 counters during peeling.
33204 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
33206 * Makefile.in (site.exp): Remove "set ISLVER".
33208 2017-02-15 Jakub Jelinek <jakub@redhat.com>
33211 * real.c (real_from_integer): Call real_convert even for decimal.
33213 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
33216 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
33218 2017-02-14 Andrew Pinski <apinski@cavium.com>
33220 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
33221 cores and change the partno/implementer to be correct.
33222 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
33223 the 'B" as the implementer.
33224 * config/aarch64/aarch64-tune.md: Regenerate.
33226 2017-02-14 Carl Love <cel@us.ibm.com>
33228 * config/rs6000/rs6000.c: Add case statement entry to make the
33229 xvcvuxdsp built-in argument unsigned.
33230 * config/rs6000/vsx.md: Fix the source and return operand types so they
33231 match the instruction definitions from the ISA document. Fix typo
33232 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
33235 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
33238 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
33239 member early_clobber_alts.
33240 * lra-lives.c (reg_early_clobber_p): New.
33241 (process_bb_lives): Use it.
33242 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
33243 (debug_operand_data): Initialize early_clobber_alts.
33244 (setup_operand_alternative): Set up early_clobber_alts.
33245 (collect_non_operand_hard_regs): Ditto. Pass early clobber
33246 alternatives to new_insn_reg.
33247 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
33249 (lra_update_insn_regno_info): Pass the new arg.
33251 2017-02-14 Jakub Jelinek <jakub@redhat.com>
33253 PR middle-end/79505
33254 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
33255 (new_oacc_loop_raw): Don't clear already cleared fields.
33258 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
33259 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
33260 _mm512_prefetch_i64gather_ps): New inline functions and macros.
33262 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
33265 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
33267 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
33270 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
33271 the extra instruction to the right place to store 128-bit constant
33274 2017-02-14 Martin Sebor <msebor@redhat.com>
33276 PR middle-end/79448
33277 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
33278 warning for strings of unknown length.
33280 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
33282 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
33284 2017-02-14 Jeff Law <law@redhat.com>
33287 * ira-costs.c (scan_one_insn): Initialize register move costs
33288 for pseudos seen in USE/CLOBBER insns.
33290 PR tree-optimization/79095
33291 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
33292 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
33293 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
33294 if the operands are known to be not equal, then the resulting range
33296 (intersect_ranges): If the new range is ~[0,0] and the old range is
33297 wide, then prefer ~[0,0].
33298 * tree-vrp.c (overflow_comparison_p_1): New function.
33299 (overflow_comparison_p): New function.
33300 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
33301 if NAME is used in an overflow test.
33302 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
33303 overflow check that can be expressed as an equality test, then adjust
33304 ops to be that equality test.
33306 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33308 * config/s390/s390-builtin-types.def: Remove flags argument.
33309 * config/s390/s390.c (s390_init_builtins): Likewise.
33311 2017-02-14 Martin Liska <mliska@suse.cz>
33313 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
33314 vector. Fix trailing white spaces.
33316 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
33318 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
33321 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33323 PR rtl-optimization/68664
33324 * config/arm/arm.c (arm_sched_can_speculate_insn):
33325 New function. Declare prototype.
33326 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
33328 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33330 PR rtl-optimization/68664
33331 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
33333 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
33335 2017-02-14 Amit Pawar <amit.pawar@amd.com>
33337 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
33338 max skip bytes for function, loop and jump.
33340 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33342 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
33343 ABS_EXPR for gimple dump.
33345 2017-02-14 Jakub Jelinek <jakub@redhat.com>
33348 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
33350 PR tree-optimization/79408
33351 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
33352 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
33353 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
33354 also if rhs1 is INTEGER_CST.
33356 2017-02-14 Richard Biener <rguenther@suse.de>
33358 PR middle-end/79432
33359 * tree-into-ssa.c (insert_phi_nodes): When the function can
33360 have abnormal edges rewrite SSA names with broken use-def
33361 dominance out of SSA and register them for PHI insertion.
33363 2017-02-13 Martin Sebor <msebor@redhat.com>
33365 PR middle-end/79496
33366 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
33367 clearing info.nowrite flag when snprintf size argument is a range.
33369 2017-02-13 Jakub Jelinek <jakub@redhat.com>
33371 * cprop.c (cprop_jump): Add missing space in string literal.
33372 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
33373 (get_constraint_for_component_ref): Likewise.
33374 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
33375 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
33376 * lra-constraints.c (process_alt_operands): Likewise.
33377 * ipa-inline.c (inline_small_functions): Likewise.
33378 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
33379 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
33380 * trans-mem.c (diagnose_tm_1_op): Likewise.
33381 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
33382 (grid_parallel_clauses_gridifiable): Likewise.
33384 * config/nvptx/mkoffload.c (process): Add space in between
33387 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
33388 "MOD4_SSE_REGS" and "ALL_REGS".
33390 * spellcheck.c (test_data): Add , in between "foo" and "food".
33392 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33395 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
33396 boundary crossing check and subsequent code generation agree.
33398 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33400 * config/aarch64/aarch64.c (has_memory_op): Delete.
33401 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
33404 2017-02-13 Jakub Jelinek <jakub@redhat.com>
33406 PR rtl-optimization/79388
33407 PR rtl-optimization/79450
33408 * combine.c (distribute_notes): When removing TEM_INSN for which
33409 corresponding dest has last value recorded, invalidate that last
33412 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33414 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
33415 of explicit '@'. Add missing assembly comment marker on branch costs
33418 2017-02-13 Nathan Sidwell <nathan@acm.org>
33420 * gengtype-lex.l (<in_struct>): Add '/'.
33422 2017-02-13 Martin Liska <mliska@suse.cz>
33425 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
33427 2017-02-13 Richard Biener <rguenther@suse.de>
33429 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
33431 * configure: Re-generate.
33432 * config.in: Likewise.
33433 * graphite-dependences.c: Simplify as if
33434 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
33435 * graphite-isl-ast-to-gimple.c: Likewise.
33436 * graphite-optimize-isl.c: Likewise.
33437 * graphite-poly.c: Likewise.
33438 * graphite-sese-to-poly.c: Likewise.
33439 * graphite.h: Likewise.
33440 * toplev.c: Include isl/version.h and use isl_version () for
33441 printing the ISL version.
33442 * doc/install.texi: Update ISL requirement.
33444 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
33446 * doc/standards.texi (Standards): Update reference to
33449 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
33451 * doc/extend.texi (Named Address Spaces): sourceware.org now
33453 * doc/install.texi (Binaries): Ditto.
33456 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
33458 * doc/cpp.texi: Replace "stringify"/"stringification" with C
33459 standard terminology "stringize"/"stringizing" throughout.
33460 * doc/cppinternals.texi: Likewise.
33462 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
33464 * doc/extend.texi: Fix some spelling mistakes and typos.
33465 * doc/invoke.texi: Likewise.
33467 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
33470 * params.def (inline-min-speedup) Change from 10 to 8.
33472 2017-02-11 Jakub Jelinek <jakub@redhat.com>
33474 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
33477 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
33480 * ipa-inline-analysis.c (get_minimal_bb): New function.
33481 (record_modified): Use it.
33482 (remap_edge_change_prob): Handle also ancestor functions.
33484 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
33486 * doc/contrib.texi (Contributors): Remove broken link into
33487 the Mauve CVS repository.
33489 2017-02-11 Jakub Jelinek <jakub@redhat.com>
33491 PR middle-end/79454
33492 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
33493 result computation whenever lhs doesn't have vector mode, not
33494 just when it has BLKmode.
33496 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
33498 * doc/makefile.texi (profiledbootstrap): Refer to the
33499 installation instructions only in textual form.
33501 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33504 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
33506 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
33508 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
33509 (Specific): Update mingw-w64 reference.
33511 (Specific): Remove broken link to Renesas RX processor.
33513 2017-02-10 Richard Biener <rguenther@suse.de>
33515 * toplev.c (process_options): Do not mention obsolete graphite
33516 options when printing sorry message about missing graphite support.
33517 Mention -floop-nest-optimize.
33519 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
33521 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
33522 (vtst_p16): Likewise.
33523 (vtstq_p8): Likewise.
33524 (vtstq_p16): Likewise.
33526 (vtstq_p64): Likewise.
33527 * config/arm/arm_neon.h (vgetq_lane_p64): New.
33528 (vset_lane_p64): New.
33529 (vsetq_lane_p64): New.
33531 2017-02-10 Jakub Jelinek <jakub@redhat.com>
33533 PR tree-optimization/79411
33534 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
33535 stmt operands are SSA_NAMEs used in abnormal phis.
33536 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
33539 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
33542 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
33545 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
33547 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
33549 2017-02-09 Jakub Jelinek <jakub@redhat.com>
33551 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
33555 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
33556 not arbitrary TREE_CONSTANT.
33559 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
33560 "omp declare target link" attribute unless is_global_var.
33561 * omp-offload.c (find_link_var_op): Likewise.
33563 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
33564 Chung-Lin Tang <cltang@codesourcery.com>
33566 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
33568 (gimplify_adjust_omp_clauses): Don't delete TILE.
33569 (gimplify_omp_for): Deal with TILE.
33570 * internal-fn.c (expand_GOACC_TILE): New function.
33571 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
33573 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
33574 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
33576 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
33577 avoid DIV for outermost collapse var.
33578 (expand_oacc_for): Insert tile element loop as needed. Adjust.
33579 Remove out of date comments, fix whitespace.
33580 * omp-general.c (omp_extract_for_data): Deal with tiling.
33581 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
33582 adjust OLF_DIM_BASE value.
33583 (struct omp_for_data): Add tiling field.
33584 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
33585 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
33586 for auto loops. Remove default auto determining, moved to
33587 oacc_loop_fixed_partitions.
33588 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
33589 stmts, add e_mask field.
33590 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
33591 (oacc_thread_numbers): Use oacc_dim_call.
33592 (oacc_xform_tile): New.
33593 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
33594 (finish_oacc_loop): Adjust for ifns vector.
33595 (oacc_loop_discover_walk): Append loop abstraction sites to list,
33596 add case for GOACC_TILE fns.
33597 (oacc_loop_xform_loop): Delete.
33598 (oacc_loop_process): Iterate over call list directly, and add
33599 handling for GOACC_TILE fns.
33600 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
33602 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
33603 vector partitioning to outer loops. Assign 2 partitions to loops
33604 when available. Add TILE handling.
33605 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
33606 (execite_oacc_device_lower): Process GOACC_TILE fns,
33607 ignore unknown specs.
33608 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
33609 * tree.c (omp_clause_num_ops): Adjust TILE ops.
33610 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
33612 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
33614 * configure.ac (ACX_BUGURL): Update.
33615 * configure: Regenerate.
33617 2017-02-09 Richard Biener <rguenther@suse.de>
33619 PR tree-optimization/69823
33620 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
33621 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
33623 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
33625 * config/arc/arc-c.def: Add __NPS400__ definition.
33626 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
33627 (TARGET_NPS400): Define.
33629 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
33631 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
33633 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
33634 pointer, arch_info.
33635 (arc_cpu_types): Fill the arch_info field with a pointer into the
33636 arc_arch_types table.
33637 (arc_selected_cpu): Declare.
33638 * config/arc/arc.c (arc_selected_cpu): Make global.
33639 (arc_selected_arch): Delete.
33640 (arc_base_cpu): Delete.
33641 (arc_override_options): Remove references to deleted variables,
33642 update access to arch information.
33643 (ARC_OPT): Update access to arch information.
33644 (ARC_OPTX): Likewise.
33645 * config/arc/arc.h (arc_base_cpu): Remove declaration.
33646 (TARGET_ARC600): Update access to arch information.
33647 (TARGET_ARC601): Likewise.
33648 (TARGET_ARC700): Likewise.
33649 (TARGET_EM): Likewise.
33650 (TARGET_HS): Likewise.
33651 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
33654 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
33657 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
33658 condition/operands for integer GE/LE/GEU/LEU operations.
33660 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
33662 PR translation/79397
33663 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
33666 2017-02-08 Martin Jambor <mjambor@suse.cz>
33669 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
33670 whether allocation happened.
33671 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
33672 nothing was allocated.
33674 2017-02-08 Jakub Jelinek <jakub@redhat.com>
33676 PR tree-optimization/79408
33677 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
33678 constant, but SSA_NAME with a known integer range, use the minimum
33679 of that range instead of op1 to determine if modulo can be replaced
33680 with its first operand.
33682 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33684 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
33686 2017-02-08 Richard Biener <rguenther@suse.de>
33688 PR tree-optimization/71824
33689 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
33690 Check all loops contained in the merged region.
33692 2017-02-07 Andrew Pinski <apinski@cavium.com>
33694 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
33696 2017-02-07 Andrew Pinski <apinski@cavium.com>
33698 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
33699 (thunderxt88): Likewise.
33700 (thunderxt81): Disable LSE and change v8.1 to v8.
33701 (thunderxt83): Likewise.
33703 2017-02-07 Jakub Jelinek <jakub@redhat.com>
33704 Richard Biener <rguenther@suse.de>
33706 PR middle-end/79399
33707 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
33708 type from int to size_t.
33709 * ira-costs.c (struct_costs_size): Change type from int to size_t.
33711 2017-02-07 Jakub Jelinek <jakub@redhat.com>
33713 PR rtl-optimization/79386
33714 * cprop.c (bypass_conditional_jumps): Initialize
33715 bypass_last_basic_block already before splitting bbs after
33716 unconditional traps...
33717 (bypass_conditional_jumps): ... rather than here.
33720 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
33721 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
33722 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
33723 fix -masm=intel patterns.
33725 2017-02-07 Richard Biener <rguenther@suse.de>
33727 PR tree-optimization/79256
33728 PR middle-end/79278
33729 * builtins.c (get_object_alignment_2): Use min_align_of_type
33730 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
33731 and ADJUST_FIELD_ALIGN.
33733 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
33735 * doc/tm.texi: Regenerate.
33736 * stor-layout.c (layout_decl): Adjust.
33737 (update_alignment_for_field): Likewise.
33738 (place_field): Likewise.
33739 (min_align_of_type): Likewise.
33740 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
33741 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
33742 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
33743 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
33744 * config/frv/frv.c (frv_adjust_field_align): Likewise.
33745 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
33746 * config/i386/i386.c (x86_field_alignment): Likewise.
33747 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
33748 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
33749 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
33750 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
33751 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
33752 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
33756 2017-01-30 Richard Biener <rguenther@suse.de>
33758 PR tree-optimization/79256
33759 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
33760 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
33763 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
33765 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
33766 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
33767 builtins to SImode and emit a zero-extend, if necessary.
33769 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
33771 * docs/invoke.texi (RISC-V Options): Alphabetize.
33773 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
33775 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
33778 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
33780 * config/riscv/riscv.c: New file.
33781 * common/config/riscv/riscv-common.c: Likewise.
33782 * config.gcc: Likewise.
33783 * config/riscv/constraints.md: Likewise.
33784 * config/riscv/elf.h: Likewise.
33785 * config/riscv/generic.md: Likewise.
33786 * config/riscv/linux.h: Likewise.
33787 * config/riscv/multilib-generator: Likewise.
33788 * config/riscv/peephole.md: Likewise.
33789 * config/riscv/pic.md: Likewise.
33790 * config/riscv/predicates.md: Likewise.
33791 * config/riscv/riscv-builtins.c: Likewise.
33792 * config/riscv/riscv-c.c: Likewise.
33793 * config/riscv/riscv-ftypes.def: Likewise.
33794 * config/riscv/riscv-modes.def: Likewise.
33795 * config/riscv/riscv-opts.h: Likewise.
33796 * config/riscv/riscv-protos.h: Likewise.
33797 * config/riscv/riscv.h: Likewise.
33798 * config/riscv/riscv.md: Likewise.
33799 * config/riscv/riscv.opt: Likewise.
33800 * config/riscv/sync.md: Likewise.
33801 * config/riscv/t-elf-multilib: Likewise.
33802 * config/riscv/t-linux: Likewise.
33803 * config/riscv/t-linux-multilib: Likewise.
33804 * config/riscv/t-riscv: Likewise.
33805 * configure.ac: Likewise.
33806 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
33807 Waterman as RISC-V maintainers.
33808 * doc/install.texi: Add RISC-V entries.
33809 * doc/invoke.texi: Add RISC-V options section.
33810 * doc/md.texi: Add RISC-V constraints section.
33811 * configure: Regenerated.
33813 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
33816 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
33817 false values to be constant vectors with all 0 or all 1 bits set.
33818 (vcondu<mode><mode>): Likewise.
33819 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
33821 (fpmask_comparison_operator): Update comment.
33822 (vecint_comparison_operator): New predicate.
33823 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
33824 vector conditionals when the true and false values are constant
33825 vectors with all 0 bits or all 1 bits set.
33827 2017-02-06 Martin Sebor <msebor@redhat.com>
33829 PR tree-optimization/79376
33830 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
33832 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
33834 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
33835 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
33836 to simplify split condition.
33838 2017-02-06 Jakub Jelinek <jakub@redhat.com>
33840 * omp-expand.c (oxpand_omp_atomic_fetch_op,
33841 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
33844 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
33846 PR rtl-optimization/68664
33847 * target.def (can_speculate_insn): New hook.
33848 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
33849 * doc/tm.texi: Regenerate.
33850 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
33851 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
33852 (rs6000_sched_can_speculate_insn): New function.
33854 2017-02-06 Jakub Jelinek <jakub@redhat.com>
33856 PR tree-optimization/79284
33857 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
33858 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
33859 vectorizable_mask_load_store, vectorizable_operation,
33860 vect_is_simple_cond, get_same_sized_vectype): Use it instead
33861 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
33862 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
33863 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
33864 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
33865 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
33866 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
33867 is_gimple_assign (stmt). Replace another such test with
33868 is_gimple_assign (stmt).
33870 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
33873 * config/avr/avr.c (rtl-iter.h): Include it.
33874 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
33875 (avr_legitimate_combined_insn): ...and implementation.
33877 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
33879 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
33880 * config/s390/s390.c (s390_const_operand_ok)
33881 (s390_canonicalize_comparison, s390_extract_part)
33882 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
33883 (s390_contiguous_bitmask_p, s390_rtx_costs)
33884 (legitimize_pic_address): Likewise.
33885 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
33886 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
33887 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
33888 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
33889 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
33891 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
33893 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
33894 REGNO($0) == REGNO($1).
33896 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33898 * config/s390/linux.h(SIZE_TYPE): Add comment.
33900 2017-02-06 Julian Brown <julian@codesourcery.com>
33901 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33902 Virendra Pathak <virendra.pathak@broadcom.com>
33904 * config/aarch64/aarch64-cores.def: Change the scheduler
33906 * config/aarch64/aarch64.md: Include thunderx2t99.md.
33907 * config/aarch64/thunderx2t99.md: New file.
33909 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
33911 * doc/standards.texi (Go Language): Update link to language
33914 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
33916 * tree-eh.c (lower_resx): Sanitize profile.
33917 (cleanup_empty_eh_move_lp): Likewise.
33919 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
33922 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
33924 * cfgloopmanip.h (loop_version): Update prototype.
33925 * modulo-sched.c (sms_schedule): Update call of loop_version.
33926 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
33927 * tree-parloops.c (gen_parallel_loop): Likewise.
33928 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
33929 * tree-ssa-loop-split.c (split_loop): Likewise.
33930 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
33931 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
33933 2017-02-05 Martin Liska <mliska@suse.cz>
33936 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
33938 (print_operand_address): Initialize a struct to zero.
33940 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
33942 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
33943 garbage collector only in textual form.
33945 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
33947 * doc/extend.texi (x86 specific memory model extensions for
33948 transactional memory): Simplify a phrase.
33950 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
33953 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
33954 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
33955 (atomic_storedi_1): Likewise.
33957 2017-02-04 Jakub Jelinek <jakub@redhat.com>
33959 PR tree-optimization/79338
33960 * tree-parloops.c (gather_scalar_reductions): Don't call
33961 vect_analyze_loop_form for loop->inner before destroying loop's
33964 2017-02-03 Martin Sebor <msebor@redhat.com>
33966 PR tree-optimization/79327
33967 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
33968 when precision has resulted in leading zeros.
33969 (format_integer): Adjust the likely counter to assume an unknown
33970 argument that may be zero is non-zero.
33972 2017-02-03 Jason Merrill <jason@redhat.com>
33975 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
33976 avoid copying non-taken branch.
33978 2017-02-03 Jakub Jelinek <jakub@redhat.com>
33980 PR tree-optimization/79340
33981 * tree-vect-loop.c (vectorizable_reduction): Release
33982 vec_defs elements after safe_splicing them into other vectors.
33985 PR tree-optimization/79327
33986 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
33987 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
33989 (format_integer): Use wide_int_to_tree instead of build_int_cst
33990 + to_?hwi. If argmin is NULL, just set argmin and argmax to
33991 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
33992 of shortest and longest sequence.
33994 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
33996 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
33997 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
33999 2017-02-03 Walter Lee <walt@tilera.com>
34002 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
34003 after initial stackframe link reg save.
34004 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
34006 2017-02-03 Jakub Jelinek <jakub@redhat.com>
34009 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
34010 wu for stxssp alternative.
34012 2017-02-03 Martin Sebor <msebor@redhat.com>
34014 PR tree-optimization/79352
34015 * gimple-fold.c (get_range_strlen): Add argument.
34016 (get_range_strlen): Change return type to bool.
34017 (get_maxval_strlen): Pass in a dummy argument.
34018 * gimple-fold.h (get_range_strlen): Change return type to bool.
34019 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
34020 * tree.h (array_at_struct_end_p): Add argument.
34021 * tree.c (array_at_struct_end_p): Handle it.
34023 2017-02-03 Martin Liska <mliska@suse.cz>
34026 * multiple_target.c (create_dispatcher_calls): Redirect edge
34027 from a caller of a dispatcher.
34028 (expand_target_clones): Make the clones local.
34029 (ipa_target_clone): Do both target clones and resolvers.
34030 (ipa_dispatcher_calls): Remove the pass.
34031 (pass_dispatcher_calls::gate): Likewise.
34032 (make_pass_dispatcher_calls): Likewise.
34033 * passes.def (pass_target_clone): Put as very first IPA early
34036 2017-02-03 Martin Liska <mliska@suse.cz>
34038 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
34039 in case of a function with ifunc attribute.
34041 2017-02-03 Martin Liska <mliska@suse.cz>
34043 * cgraph.c (cgraph_node::dump): Dump function version info.
34044 * symtab.c (symtab_node::dump_base): Add missing new line.
34046 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
34048 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
34049 (ifcombine_ifandif): Use it.
34051 2017-02-03 Martin Liska <mliska@suse.cz>
34053 * doc/invoke.texi: Document default value for
34054 use-after-scope-direct-emission-threshold.
34056 2017-02-03 Martin Liska <mliska@suse.cz>
34058 PR tree-optimization/79339
34059 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
34060 (format_floating): Likewise.
34062 2017-02-03 Martin Liska <mliska@suse.cz>
34065 * ipa-prop.c (ipa_node_params_t::insert): Remove current
34067 (ipa_node_params_t::remove): Likewise.
34068 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
34069 initialization from removed ipa_node_params_t::insert.
34070 (ipa_node_params::~ipa_node_params): Move from removed
34071 ipa_node_params_t::release.
34072 * symbol-summary.h (symbol_summary::m_released): New member.
34073 Do not release a summary twice. Do not allow to call finalizer
34074 for types of a summary that live in GGC memory.
34076 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34078 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
34081 2017-02-02 Martin Sebor <msebor@redhat.com>
34083 PR middle-end/79275
34084 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
34085 (format_string): Tighten up the range of output for non-constant
34086 strings and correct the expected range for wide non-constant strings.
34088 2017-02-02 Martin Sebor <msebor@redhat.com>
34090 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
34092 PR middle-end/32003
34093 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
34095 (-fdump-tree-@var): Add to index and document how to come up
34096 with pass-specific option and dump file names.
34097 (-fdump-passes): Clarify where to look for output.
34099 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
34101 PR middle-end/77445
34102 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
34103 statistics of the analyzed path; allow threading for speed when
34104 any of BBs along the path are optimized for speed.
34106 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
34108 PR middle-end/78468
34109 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
34110 settings of the virtual registers.
34113 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
34115 * explow.c (get_dynamic_stack_size): Take known alignment of stack
34116 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
34119 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34121 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
34122 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
34124 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34126 * config/s390/s390.md: Add missing comments with the expanded
34128 * config/s390/vector.md: Likewise.
34129 * config/s390/vx-builtins.md: Likewise.
34131 2017-02-02 Jakub Jelinek <jakub@redhat.com>
34134 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
34135 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
34136 conditions on a single line.
34138 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34140 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
34141 __S390_VX__ to __VX__.
34143 2017-02-01 Andrew Pinski <apinski@cavium.com>
34145 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
34146 stmt_info to record_stmt_cost.
34147 (vect_get_known_peeling_cost): Pass stmt_info if known to
34149 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
34150 cpu_vector_cost field into
34151 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
34152 field into vec_int_stmt_cost and vec_fp_stmt_cost.
34153 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
34154 splitting of scalar_stmt_cost and vec_stmt_cost.
34155 (thunderx_vector_cost): Likewise.
34156 (cortexa57_vector_cost): LIkewise.
34157 (exynosm1_vector_cost): Likewise.
34158 (xgene1_vector_cost): Likewise.
34159 (thunderx2t99_vector_cost): Improve after the splitting of the two
34161 (aarch64_builtin_vectorization_cost): Update for the splitting of
34162 scalar_stmt_cost and vec_stmt_cost.
34164 2017-02-01 Torvald Riegel <triegel@redhat.com>
34165 Richard Henderson <rth@redhat.com>
34167 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
34168 conditional on existance of a fast atomic load.
34169 * optabs-query.c (can_atomic_load_p): New function.
34170 * optabs-query.h (can_atomic_load_p): Declare it.
34171 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
34172 no fast atomic load is available for the particular size of access.
34173 (expand_atomic_compare_and_swap): Likewise.
34174 (expand_atomic_load): Likewise.
34175 (expand_atomic_store): Likewise.
34176 (expand_atomic_fetch_op): Likewise.
34177 * testsuite/lib/target-supports.exp
34178 (check_effective_target_sync_int_128): Remove x86 because it provides
34179 no fast atomic load.
34180 (check_effective_target_sync_int_128_runtime): Likewise.
34182 2017-02-01 Richard Biener <rguenther@suse.de>
34184 * graphite.c: Include tree-vectorizer.h for find_loop_location.
34185 (graphite_transform_loops): Provide opt-info for optimized nests.
34186 * tree-parloop.c (parallelize_loops): Provide opt-info for
34187 parallelized loops.
34189 2017-02-01 Richard Biener <rguenther@suse.de>
34191 PR middle-end/79315
34192 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
34193 was not set before.
34195 2017-02-01 Richard Biener <rguenther@suse.de>
34197 PR tree-optimization/71824
34198 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
34199 Verify the loops are valid in the merged SESE region.
34200 (scop_detection::can_represent_loop_1): Check analyzing the
34201 evolution of the number of iterations in the region succeeds.
34203 2017-01-31 Ian Lance Taylor <iant@golang.org>
34205 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
34206 REG_ARGS_SIZE note to 32-bit push insns and call insn.
34208 2017-01-31 David Malcolm <dmalcolm@redhat.com>
34210 PR preprocessor/79210
34211 * input.c (get_substring_ranges_for_loc): Replace line_width
34212 assertion with error-handling.
34214 2017-01-31 Richard Biener <rguenther@suse.de>
34216 PR tree-optimization/77318
34217 * graphite-sese-to-poly.c (extract_affine): Fix assert.
34218 (create_pw_aff_from_tree): Take loop parameter.
34219 (add_condition_to_pbb): Pass loop of the condition to
34220 create_pw_aff_from_tree.
34222 2017-01-31 Jakub Jelinek <jakub@redhat.com>
34224 * config/s390/s390.c (s390_asan_shadow_offset): New function.
34225 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
34227 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
34231 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
34233 (convert_int_to_float128): Likewise.
34234 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
34235 (convert_int_to_float128): Likewise.
34236 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
34237 (UNSPEC_IEEE128_CONVERT): Likewise.
34238 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
34239 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
34240 Use local variables for IBM extended format.
34241 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
34242 (fix_trunc<mode>si2_fprs): Likewise.
34243 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
34244 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
34245 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
34246 to know that we can now have integers of all sizes in vector
34248 (fix<uns>_<mode>di2_hw): Likewise.
34249 (float<uns>_<mode>si2_hw): Likewise.
34250 (fix_<mode>si2_hw): Likewise.
34251 (fixuns_<mode>si2_hw): Likewise.
34252 (float<uns>_<mode>di2_hw): Likewise.
34253 (float_<mode>di2_hw): Likewise.
34254 (float_<mode>si2_hw): Likewise.
34255 (floatuns_<mode>di2_hw): Likewise.
34256 (floatuns_<mode>si2_hw): Likewise.
34257 (xscvqp<su>wz_<mode>): Delete, no longer used.
34258 (xscvqp<su>dz_<mode>): Likewise.
34259 (xscv<su>dqp_<mode>): Likewise.
34260 (ieee128_mfvsrd_64bit): Likewise.
34261 (ieee128_mfvsrd_32bit): Likewise.
34262 (ieee128_mfvsrwz): Likewise.
34263 (ieee128_mtvsrw): Likewise.
34264 (ieee128_mtvsrd_64bit): Likewise.
34265 (ieee128_mtvsrd_32bit): Likewise.
34267 2017-01-31 Martin Liska <mliska@suse.cz>
34270 * ipa-prop.c (ipa_free_all_node_params): Call release method
34271 instead of ~sumbol_summary to not to trigger double times
34274 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
34276 PR tree-optimization/71691
34277 * bitmap.h (class auto_bitmap): New.
34278 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
34279 is_maybe_undefined instead of ssa_undefined_value_p.
34281 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34283 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
34284 __S390_ARCH_LEVEL__ to __ARCH__.
34286 2017-01-31 Jakub Jelinek <jakub@redhat.com>
34288 PR tree-optimization/79267
34289 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
34290 if should_remove_lhs_p is true.
34292 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
34295 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
34296 (add_alignment_attribute): New.
34297 (base_type_die): Add alignment attribute.
34298 (subrange_type_die): Likewise.
34299 (modified_type_die): Likewise.
34300 (gen_array_type_die): Likewise.
34301 (gen_descr_array_type_die: Likewise.
34302 (gen_enumeration_type_die): Likewise.
34303 (gen_subprogram_die): Likewise.
34304 (gen_variable_die): Likewise.
34305 (gen_field_die): Likewise.
34306 (gen_ptr_to_mbr_type_die): Likewise.
34307 (gen_struct_or_union_type_die): Likewise.
34308 (gen_subroutine_type_die): Likewise.
34309 (gen_typedef_die): Likewise.
34310 (base_type_cmp): Compare alignment attribute.
34312 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34315 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
34316 (setb_unsigned) New pattern for setb with CCUNS.
34317 * config/rs6000/rs6000.c (expand_block_compare): Use a different
34318 subfc./subfe sequence to avoid overflow problems. Generate a
34319 shorter sequence with cmpld/setb for power9.
34320 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
34321 for generating subfc. instruction.
34322 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
34323 now uses this instruction.
34325 2017-01-30 Ian Lance Taylor <iant@google.com>
34328 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
34329 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
34331 2017-01-30 Martin Sebor <msebor@redhat.com>
34333 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
34334 Move constant to the right of a relational operator.
34335 (get_mpfr_format_length, format_character, format_string): Ditto.
34336 (should_warn_p, maybe_warn): Same.
34338 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
34340 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
34343 * asan.c (get_translation_unit_decl): Remove function.
34344 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
34346 2017-01-30 Martin Liska <mliska@suse.cz>
34348 PR gcov-profile/79259
34349 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
34350 -fprofile-generate.
34352 2017-01-30 Martin Liska <mliska@suse.cz>
34355 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
34356 Initialize variables with NULL value.
34358 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
34361 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
34363 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
34365 2017-01-30 Richard Biener <rguenther@suse.de>
34367 PR tree-optimization/79276
34368 * tree-vrp.c (process_assert_insertions): Properly adjust common
34369 when removing a duplicate.
34371 2017-01-30 Richard Biener <rguenther@suse.de>
34373 PR tree-optimization/79256
34374 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
34375 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
34377 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
34379 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
34382 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
34383 ("*r<noxa>sbg_<mode>_sll_bitmask")
34384 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
34385 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
34386 Use contiguous_bitmask_nowrap_operand.
34388 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34391 * config/rs6000/altivec.h (vec_xl): Revise #define.
34392 (vec_xst): Likewise.
34394 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
34396 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
34398 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
34400 PR rtl-optimization/79194
34401 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
34402 traps before call to bypass_conditional_jumps.
34404 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
34406 PR tree-optimization/71374
34407 * lra-constraints.c (check_conflict_input_operands): New.
34408 (match_reload): Use it.
34410 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
34413 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
34414 account to calculate conflict_set.
34416 2017-01-27 Bin Cheng <bin.cheng@arm.com>
34418 PR rtl-optimization/78559
34419 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
34420 other_insn in combine.
34422 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
34424 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
34425 uint16_type_node for BT_UINT16.
34427 2017-01-27 David Malcolm <dmalcolm@redhat.com>
34429 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
34430 "RTL Tests" to menu.
34431 (GIMPLE Tests): New node.
34432 (RTL Tests): New node.
34434 2017-01-27 Richard Biener <rguenther@suse.de>
34436 PR tree-optimization/79245
34437 * tree-loop-distribution.c (distribute_loop): Apply cost
34438 modeling also to detected patterns.
34440 2017-01-27 Richard Biener <rguenther@suse.de>
34442 PR tree-optimization/71433
34443 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
34444 (compare_assert_loc): New function.
34445 (process_assert_insertions): Sort and optimize assert locations
34446 to remove duplicates and push down identical assertions on
34447 edges to their destination block.
34449 2017-01-27 Richard Biener <rguenther@suse.de>
34451 PR tree-optimization/79244
34452 * tree-vrp.c (remove_range_assertions): Forcefully propagate
34453 out SSA names even if abnormal.
34455 2017-01-27 Jakub Jelinek <jakub@redhat.com>
34457 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
34458 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
34459 instead of MPFR_RNDN.
34461 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
34464 * arm.c (arm_option_override): Don't call build_target_option_node
34465 until after doing all option overrides.
34466 (arm_valid_target_attribute_tree): Likewise.
34468 2017-01-27 Martin Liska <mliska@suse.cz>
34470 * doc/invoke.texi (-fprofile-arcs): Document profiling support
34471 for {cd}tors and C++ {cd}tors.
34473 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
34475 * config/s390/s390.md ("*setmem_long_and")
34476 ("*setmem_long_and_31z"): Use zero_extend instead of and.
34478 2017-01-26 Martin Sebor <msebor@redhat.com>
34480 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
34483 2017-01-26 Martin Sebor <msebor@redhat.com>
34485 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
34486 HAVE_DFmode before using XFmode or DFmode.
34487 (parse_directive): Avoid using the z length modifier to avoid
34488 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
34490 PR middle-end/78703
34491 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
34492 to accept adjustment as an array.
34493 (get_int_range): New function.
34494 (struct directive): Make width and prec arrays.
34495 (directive::set_width, directive::set_precision): Call get_int_range.
34496 (format_integer, format_floating): Handle width and precision ranges.
34497 (format_string, parse_directive): Same.
34499 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34502 * dwarf2out.c (generate_skeleton_bottom_up): For children with
34503 comdat_type_p set, just clone them, but keep the children in the
34507 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
34508 which have direct callers with -fvar-tracking-assignments enabled
34510 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
34511 inside of type units.
34513 2017-01-26 Martin Sebor <msebor@redhat.com>
34515 PR middle-end/78703
34516 * gimple-ssa-sprintf.c (struct result_range): Add likely and
34518 (struct format_result): Replace number_chars, number_chars_min,
34519 and number_chars_max with a single member of struct result_range.
34521 (format_result::operator+=): Adjust.
34522 (struct fmtresult): Remove bounded. Handle likely and unlikely
34524 (fmtresult::adjust_for_width_or_precision): New function.
34525 (fmtresult:type_max_digits): New function.
34526 (bytes_remaining): Handle likely and unlikely counters.
34527 (min_bytes_remaining): Remove.
34528 (format_percent): Simplify.
34529 (format_integer, format_floating): Set likely and unlikely counters.
34530 (get_string_length, format_character, format_string): Same.
34531 (format_plain, should_warn_p): New function.
34532 (maybe_warn): Call should_warn_p. Update diagnostic messages
34533 and handle those for all directives, including plain strings.
34534 (format_directive): Handle likely and unlikely counters.
34535 Remove unnecessary quoting from diagnostics. Add an informational
34537 (add_bytes): Remove.
34538 (pass_sprintf_length::compute_format_length): Simplify.
34539 (try_substitute_return_value): Handle likely and unlikely counters.
34541 2017-01-26 Carl Love <cel@us.ibm.com>
34543 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
34544 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
34546 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
34549 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
34550 endianess for subregs into account.
34551 * lra-constraints.c (lra_constraints): Do risky transformations
34552 always on the first iteration.
34553 * lra-lives.c (check_pseudos_live_through_calls): Add arg
34554 last_call_used_reg_set.
34555 (process_bb_lives): Define and use last_call_used_reg_set.
34556 * lra.c (lra): Always continue after lra_constraints on the first
34559 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
34561 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
34563 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
34565 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34567 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
34568 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
34569 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
34570 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
34571 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
34572 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
34573 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
34574 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
34575 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
34577 2017-01-26 Marek Polacek <polacek@redhat.com>
34580 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
34581 for the third operand.
34583 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34585 PR middle-end/79236
34586 * omp-low.c (struct omp_context): Add simt_stmt field.
34587 (scan_omp_for): Return omp_context *.
34588 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
34589 context to the _simt_ SIMD stmt.
34590 (lower_omp_for): For combined SIMD with sibling _simt_
34591 SIMD, make sure to use the same decls in _looptemp_
34592 clauses as in the sibling.
34594 2017-01-26 David Sherwood <david.sherwood@arm.com>
34596 PR middle-end/79212
34597 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
34600 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34603 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
34604 emit fld b; fld a; if possible.
34606 * brig-builtins.def: Update copyright years.
34607 * config/arm/arm_acle_builtins.def: Update copyright years.
34609 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
34612 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
34613 constraint instead of o for the stxsd instruction.
34615 2017-01-25 Carl Love <cel@us.ibm.com>
34617 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
34618 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
34620 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
34622 * doc/invoke.texi (C++ Dialect Options): Fix typo.
34624 2017-01-25 Richard Biener <rguenther@suse.de>
34626 PR tree-optimization/69264
34627 * target.def (vector_alignment_reachable): Improve documentation.
34628 * doc/tm.texi: Regenerate.
34629 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
34631 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
34632 earlier changes with respect to TYPE_USER_ALIGN.
34633 (vector_alignment_reachable_p): Likewise. Improve dumping.
34635 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34638 * config/arm/arm.md (xordi3): Force constant operand into a register
34641 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34643 * doc/invoke.texi (-fstore-merging): Correct default optimization
34644 levels at which it is enabled.
34645 (-O): Move -fstore-merging from list to...
34648 2017-01-25 Richard Biener <rguenther@suse.de>
34651 * omp-expand.c: Include debug.h.
34652 (expand_omp_taskreg): Make sure to generate early debug before
34653 outlining anything from a function.
34654 (expand_omp_target): Likewise.
34655 (grid_expand_target_grid_body): Likewise.
34657 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
34660 * asan.c (get_translation_unit_decl): New function.
34661 (asan_add_global): Extract modules file name from globals
34662 TRANSLATION_UNIT_DECL name.
34664 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
34667 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
34668 for long calls with APCS frame and VFP.
34670 2017-01-24 David Malcolm <dmalcolm@redhat.com>
34672 * cfg.c (original_copy_tables_initialized_p): New function.
34673 * cfg.h (original_copy_tables_initialized_p): New decl.
34674 * cfgrtl.c (relink_block_chain): Guard the call to
34675 free_original_copy_tables with a call to
34676 original_copy_tables_initialized_p.
34677 * cgraph.h (symtab_node::native_rtl_p): New decl.
34678 * cgraphunit.c (symtab_node::native_rtl_p): New function.
34679 (symtab_node::needed_p): Don't assert for early assembly output
34680 for __RTL functions.
34681 (cgraph_node::finalize_function): Set "force_output" for __RTL
34683 (cgraph_node::analyze): Bail out early for __RTL functions.
34684 (analyze_functions): Update assertion to support __RTL functions.
34685 (cgraph_node::expand): Bail out early for __RTL functions.
34686 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
34688 * function.h (struct function): Update comment for field
34690 * gimple-expr.c: Include "tree-pass.h".
34691 (gimple_has_body_p): Return false for __RTL functions.
34692 * Makefile.in (OBJS): Add run-rtl-passes.o.
34693 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
34695 (gcc::pass_manager::get_clean_slate): New accessor.
34696 * passes.c: Include "insn-addr.h".
34697 (should_skip_pass_p): Add logging. Update logic for running
34698 "expand" to be compatible with both __GIMPLE and __RTL. Guard
34699 property-provider override so it is only done for gimple passes.
34701 (skip_pass): New function.
34702 (execute_one_pass): Call skip_pass when skipping passes.
34703 * read-md.c (md_reader::read_char): Support filtering
34704 the input to a subset of line numbers.
34705 (md_reader::md_reader): Initialize fields
34706 m_first_line and m_last_line.
34707 (md_reader::read_file_fragment): New function.
34708 * read-md.h (md_reader::read_file_fragment): New decl.
34709 (md_reader::m_first_line): New field.
34710 (md_reader::m_last_line): New field.
34711 * read-rtl-function.c (function_reader::create_function): Only
34712 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
34713 curr_properties. Set DECL_INITIAL to a dummy block.
34714 (read_rtl_function_body_from_file_range): New function.
34715 * read-rtl-function.h (read_rtl_function_body_from_file_range):
34717 * run-rtl-passes.c: New file.
34718 * run-rtl-passes.h: New file.
34720 2017-01-24 Jeff Law <law@redhat.com>
34722 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
34725 2017-01-24 Bin Cheng <bin.cheng@arm.com>
34727 PR tree-optimization/79159
34728 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
34729 (record_nonwrapping_iv): Improve boundary using above function if no
34730 value range information.
34732 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
34733 Martin Jambor <mjambor@suse.cz>
34735 * brig-builtins.def: New file.
34736 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
34737 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
34738 (DEF_HSAIL_SAT_BUILTIN): Likewise.
34739 (DEF_HSAIL_INTR_BUILTIN): Likewise.
34740 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
34741 * builtin-types.def (BT_INT8): New.
34742 (BT_INT16): Likewise.
34743 (BT_UINT8): Likewise.
34744 (BT_UINT16): Likewise.
34745 (BT_FN_ULONG): Likewise.
34746 (BT_FN_UINT_INT): Likewise.
34747 (BT_FN_UINT_ULONG): Likewise.
34748 (BT_FN_UINT_LONG): Likewise.
34749 (BT_FN_UINT_PTR): Likewise.
34750 (BT_FN_ULONG_PTR): Likewise.
34751 (BT_FN_INT8_FLOAT): Likewise.
34752 (BT_FN_INT16_FLOAT): Likewise.
34753 (BT_FN_UINT32_FLOAT): Likewise.
34754 (BT_FN_UINT16_FLOAT): Likewise.
34755 (BT_FN_UINT8_FLOAT): Likewise.
34756 (BT_FN_UINT64_FLOAT): Likewise.
34757 (BT_FN_UINT16_UINT32): Likewise.
34758 (BT_FN_UINT32_UINT16): Likewise.
34759 (BT_FN_UINT16_UINT16_UINT16): Likewise.
34760 (BT_FN_INT_PTR_INT): Likewise.
34761 (BT_FN_UINT_PTR_UINT): Likewise.
34762 (BT_FN_LONG_PTR_LONG): Likewise.
34763 (BT_FN_ULONG_PTR_ULONG): Likewise.
34764 (BT_FN_VOID_UINT64_UINT64): Likewise.
34765 (BT_FN_UINT8_UINT8_UINT8): Likewise.
34766 (BT_FN_INT8_INT8_INT8): Likewise.
34767 (BT_FN_INT16_INT16_INT16): Likewise.
34768 (BT_FN_INT_INT_INT): Likewise.
34769 (BT_FN_UINT_FLOAT_UINT): Likewise.
34770 (BT_FN_FLOAT_UINT_UINT): Likewise.
34771 (BT_FN_ULONG_UINT_UINT): Likewise.
34772 (BT_FN_ULONG_UINT_PTR): Likewise.
34773 (BT_FN_ULONG_ULONG_ULONG): Likewise.
34774 (BT_FN_UINT_UINT_UINT): Likewise.
34775 (BT_FN_VOID_UINT_PTR): Likewise.
34776 (BT_FN_UINT_UINT_PTR: Likewise.
34777 (BT_FN_UINT32_UINT64_PTR): Likewise.
34778 (BT_FN_INT_INT_UINT_UINT): Likewise.
34779 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
34780 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
34781 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
34782 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
34783 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
34784 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
34785 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
34786 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
34787 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
34788 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
34789 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
34790 * doc/frontends.texi: List BRIG FE.
34791 * doc/install.texi (Testing): Add BRIG tesring requirements.
34792 * doc/invoke.texi (Overall Options): Mention BRIG.
34793 * doc/standards.texi (Standards): Doucment BRIG HSA version.
34795 2017-01-24 Richard Biener <rguenther@suse.de>
34797 PR translation/79208
34798 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
34800 2017-01-24 Martin Jambor <mjambor@suse.cz>
34803 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
34804 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
34805 and known_contexts.
34807 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
34809 PR middle-end/79123
34810 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
34811 casts from signed to unsigned really don't have a range.
34813 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
34815 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
34816 GMP_RNDx for compatiblity.
34818 2017-01-24 Martin Liska <mliska@suse.cz>
34821 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
34822 that would prevent us to call alloca with -1 as argument.
34824 2017-01-24 Jakub Jelinek <jakub@redhat.com>
34826 * dwarf2out.c (output_compilation_unit_header, output_file_names):
34827 Avoid -Wformat-security warning.
34829 2017-01-23 Andrew Pinski <apinski@cavium.com>
34831 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
34834 2017-01-23 Martin Sebor <msebor@redhat.com>
34836 PR middle-end/78703
34837 * gimple-ssa-sprintf.c (warn_level): New global.
34838 (format_integer): Use it here and throughout the rest of the file.
34839 Use the same switch to compute sign as base.
34840 (maybe_warn): New function.
34841 (format_directive): Factor out warnings into maybe_warn.
34842 Add debugging output. Use warn_level.
34843 (add_bytes): Use warn_level.
34844 (pass_sprintf_length::compute_format_length): Add debugging output.
34845 (try_substitute_return_value): Same.
34846 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
34848 PR middle-end/78703
34849 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
34850 (struct fmtresult, format_integer, format_floating): Adjust.
34851 (fmtresult::fmtresult): Set max correctly in two argument ctor.
34852 (get_string_length, format_string,format_directive): Same.
34853 (pass_sprintf_length::compute_format_length): Same.
34854 (try_substitute_return_value): Simplify slightly.
34856 PR middle-end/78703
34857 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
34858 (fmtresult::operator+=): Outlined.
34859 (struct fmtresult): Add ctors.
34860 (struct conversion_spec): Rename...
34861 (struct directive): ...to this. Add and remove data members.
34862 (directive::set_width, directive::set_precision): New functions.
34863 (format_percent): Use fmtresult ctor.
34864 (get_width_and_precision): Remove.
34865 (format_integer): Make naming changes. Avoid computing width and
34867 (format_floating): Same. Adjust indentation.
34868 (format_character, format_none): New functions.
34869 (format_string): Moved character handling to format_character.
34870 (format_directive): Remove arguments, change return type.
34871 (parse_directive): New function.
34872 (pass_sprintf_length::compute_format_length): Move directive
34873 parsing to parse_directive.
34875 2017-01-23 Jakub Jelinek <jakub@redhat.com>
34877 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
34878 (assign_assembler_name_if_needed): ... this.
34879 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
34880 (assign_assembler_name_if_needed): ... this.
34881 (free_lang_data_in_cgraph): Adjust callers.
34882 * cgraphunit.c (cgraph_node::analyze): Likewise.
34883 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
34886 2017-01-23 Richard Biener <rguenther@suse.de>
34888 PR tree-optimization/79088
34889 PR tree-optimization/79188
34890 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
34891 resetting loop bounds after last path deletion. Reset loop
34892 bounds of the target loop, make code match the comments.
34893 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
34894 Make sure loops need no fixups.
34896 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
34898 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
34899 exponent support with double type for first argument.
34900 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
34901 type returned by __builtin_vec_extract_sig,
34902 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
34903 functions from "vector int" to "vector unsigned int" or from
34904 "vector long long int" to "vector unsigned long long int".
34905 Changed type returned by __builtin_vec_extract_exp,
34906 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
34907 functions from "vector int" to "vector unsigned int" or from
34908 "vector long long int" to "vector unsigned long long int".
34909 Changed return type of __builtin_vec_test_data_class,
34910 __builtin_vec_test_data_class_sp, and
34911 __builtin_vec_test_data_class_dp from "vector int" to
34912 "vector bool int" or from "vector long long int" to "vector bool
34913 long long int" and changed second argument type from "unsigned
34914 int" to "int". Added new overloaded function forms "vector float
34915 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
34916 "vector float __builtin_vec_insert_exp_sp (vector float, vector
34917 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
34918 double, vector unsigned long long int)" and "vector double
34919 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
34920 long int)". Changed return type of
34921 __builtin_scalar_test_data_class and
34922 __builtin_scalar_test_data_class_sp and
34923 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
34924 int" and changed second argument from "unsigned int" to "int".
34925 Changed type returned by __builtin_scalar_test_neg,
34926 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
34927 from "int" to "bool int". Added new overloaded function form
34928 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
34929 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
34930 exponent double-precision with floating point first argument.
34931 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
34932 documentation of scalar_test_data_class, scalar_test_neg,
34933 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
34934 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
34935 vec_test_data_class built-in functions to reflect refinements in
34936 their type signatures.
34938 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
34940 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
34942 (aarch64_elf_asm_destructor): Likewise.
34944 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
34946 PR rtl-optimization/78634
34947 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
34948 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
34949 * ifcvt.c (noce_try_cmove): Add missing cost check.
34951 PR rtl-optimization/71724
34952 * combine.c (if_then_else_cond): Look for situations where it is
34953 beneficial to undo the work of one of the recursive calls.
34955 2017-01-23 Bin Cheng <bin.cheng@arm.com>
34957 PR tree-optimization/70754
34958 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
34959 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
34960 combined stmt before it if not NULL.
34961 (combine_chains): Process refs reversely and compute dominance point
34964 2017-01-23 Martin Liska <mliska@suse.cz>
34966 PR tree-optimization/79196
34967 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
34968 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
34970 (strlen_optimize_stmt): Call the renamed function.
34972 2017-01-23 Michael Matz <matz@suse.de>
34974 PR tree-optimization/78384
34975 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
34977 2017-01-23 Richard Biener <rguenther@suse.de>
34979 PR tree-optimization/79186
34980 * tree-vrp.c (register_new_assert_for): Make sure we've seen
34981 both incoming edges before moving an assert.
34983 2017-01-23 Martin Jambor <mjambor@suse.cz>
34985 * ipa-prop.c (load_from_param_1): Removed.
34986 (load_from_unmodified_param): Bits from load_from_param_1 put back
34988 (load_from_param): Removed.
34989 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
34990 with stmt. Reverted back to use of load_from_unmodified_param.
34992 2017-01-23 Martin Jambor <mjambor@suse.cz>
34995 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
34996 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
34997 field a pointer to garbage collected vector, mark lattices and
34998 ipcp_orig_node with GTY((skip)).
34999 (ipa_get_param_count): Adjust to descriptors being a pointer.
35000 (ipa_get_param): Likewise.
35001 (ipa_get_type): Likewise.
35002 (ipa_get_param_move_cost): Likewise.
35003 (ipa_set_param_used): Likewise.
35004 (ipa_get_controlled_uses): Likewise.
35005 (ipa_set_controlled_uses): Likewise.
35006 (ipa_is_param_used): Likewise.
35007 (ipa_node_params_t): Move into garbage collector. New methods insert
35009 (ipa_node_params_sum): Annotate wth GTY(()).
35010 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
35012 (ipa_load_from_parm_agg): Adjust declaration.
35013 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
35014 * ipa-profile.c (ipa_profile): Likewise.
35015 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
35016 (ipa_populate_param_decls): Make descriptors parameter garbage
35018 (ipa_dump_param): Adjust to descriptors being a pointer.
35019 (ipa_alloc_node_params): Likewise.
35020 (ipa_initialize_node_params): Likewise.
35021 (load_from_param_1): Make descriptors parameter garbage collected.
35022 (load_from_unmodified_param): Likewise.
35023 (load_from_param): Likewise.
35024 (ipa_load_from_parm_agg): Likewise.
35025 (ipa_node_params::~ipa_node_params): Removed.
35026 (ipa_free_all_node_params): Remove call to delete operator.
35027 (ipa_node_params_t::insert): New.
35028 (ipa_node_params_t::remove): Likewise.
35029 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
35030 copy known_csts and known_contexts vectors.
35031 (ipa_read_node_info): Adjust to descriptors being a pointer.
35032 (ipcp_modif_dom_walker): Make m_descriptors field garbage
35034 (ipcp_transform_function): Make descriptors variable garbage
35037 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
35039 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
35040 * config/i386/avx512dqintrin.h: Ditto.
35041 * config/i386/avx512fintrin.h: Ditto.
35042 * config/i386/i386.c: Handle new builtins.
35043 * config/i386/i386-builtin.def: Add new builtins.
35044 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
35045 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
35047 2017-01-23 Jakub Jelinek <jakub@redhat.com>
35048 Martin Liska <mliska@suse.cz>
35050 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
35051 * asan.c (asan_expand_poison_ifn): Support stores and use
35052 appropriate ASAN report function.
35053 * internal-fn.c (expand_ASAN_POISON_USE): New function.
35054 * internal-fn.def (ASAN_POISON_USE): Declare.
35055 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
35056 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
35057 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
35058 ASAN_POISON calls w/o LHS.
35059 * tree-ssa.c (execute_update_addresses_taken): Create clobber
35060 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
35061 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
35062 * gimplify.c (asan_poison_variables): Add attribute
35063 use_after_scope_memory to variables that really needs to live
35065 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
35066 having the attribute.
35068 2017-01-23 Martin Liska <mliska@suse.cz>
35070 * asan.c (create_asan_shadow_var): New function.
35071 (asan_expand_poison_ifn): Likewise.
35072 * asan.h (asan_expand_poison_ifn): New declaration.
35073 * internal-fn.c (expand_ASAN_POISON): Likewise.
35074 * internal-fn.def (ASAN_POISON): New builtin.
35075 * sanopt.c (pass_sanopt::execute): Expand
35076 asan_expand_poison_ifn.
35077 * tree-inline.c (copy_decl_for_dup_finish): Make function
35079 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
35080 * tree-ssa.c (is_asan_mark_p): New function.
35081 (execute_update_addresses_taken): Rewrite local variables
35082 (identified just by use-after-scope as addressable) into SSA.
35084 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
35086 * doc/install.texi (Specific): opensource.apple.com uses https
35087 now. Remove trailing slash.
35089 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
35091 * README.Portability: Remove note on an Irix compatibility issue.
35093 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
35095 * gcov.c (INCLUDE_ALGORITHM): Define.
35096 (INCLUDE_VECTOR): Define.
35097 No longer include <vector> and <algorithm> directly.
35099 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
35101 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
35103 * doc/invoke.texi (Code Gen Options): Ditto.
35105 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
35108 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
35110 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
35112 rtl-optimization/79125
35113 * cprop.c (local_cprop_pass): Handle cases where we make an
35114 unconditional trap.
35116 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
35120 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
35121 read from, for big endian.
35123 2017-01-20 Jiong Wang <jiong.wang@arm.com>
35125 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
35126 register pauth builtins for LP64 only.
35128 2017-01-20 Marek Polacek <polacek@redhat.com>
35131 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
35134 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
35136 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
35138 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
35139 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
35140 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
35142 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35145 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
35146 in big-endian mode when they are not a single duplicated value.
35148 2017-01-20 Richard Biener <rguenther@suse.de>
35150 * BASE-VER: Bump to 7.0.1.
35152 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
35154 * omp-low.c (omplow_simd_context): New struct. Use it...
35155 (lower_rec_simd_input_clauses): ...here and...
35156 (lower_rec_input_clauses): ...here to hold common data. Adjust all
35157 references to idx, lane, max_vf, is_simt.
35159 2017-01-20 Graham Markall <graham.markall@embecosm.com>
35161 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
35164 2017-01-20 Martin Jambor <mjambor@suse.cz>
35166 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
35167 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
35169 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
35170 (GTFILES): Rename hsa.c to hsa-common.c.
35171 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
35172 * hsa-dump.c: Likewise.
35173 * hsa-gen.c: Likewise.
35174 * hsa-regalloc.c: Likewise.
35175 * ipa-hsa.c: Likewise.
35176 * omp-expand.c: Likewise.
35177 * omp-low.c: Likewise.
35178 * toplev.c: Likewise.
35180 2017-01-20 Marek Polacek <polacek@redhat.com>
35183 * doc/invoke.texi: Document -Wduplicated-branches.
35184 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
35185 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
35186 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
35187 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
35188 return 0 only when not OEP_LEXICOGRAPHIC.
35189 (fold_build_cleanup_point_expr): Use the expression
35190 location when building CLEANUP_POINT_EXPR.
35191 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
35192 * tree.c (add_expr): Handle error_mark_node.
35194 2017-01-20 Martin Liska <mliska@suse.cz>
35197 * tree-profile.c (init_ic_make_global_vars): Do not call
35199 (gimple_init_gcov_profiler): Likewise.
35201 2017-01-20 Martin Liska <mliska@suse.cz>
35204 * cgraph.h (maybe_create_reference): Remove argument and
35206 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
35208 * ipa-cp.c (create_specialized_node): Likewise.
35209 * symtab.c (symtab_node::maybe_create_reference): Handle
35210 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
35212 2017-01-20 Martin Liska <mliska@suse.cz>
35214 * read-rtl-function.c (function_reader::create_function): Use
35215 build_decl instread of build_decl_stat.
35217 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
35219 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
35220 * config/i386/avx512dqintrin.h: Ditto.
35221 * config/i386/avx512fintrin.h: Ditto.
35222 * config/i386/i386-builtin-types.def: Add new types.
35223 * config/i386/i386.c: Handle new types.
35224 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
35225 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
35226 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
35227 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
35228 (__builtin_ia32_kshiftridi): New.
35229 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
35231 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
35235 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
35236 define to rs6000_init_stack_protect_guard.
35237 (rs6000_init_stack_protect_guard): New function.
35239 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
35240 Yunqiang Su <yunqiang.su@imgtec.com>
35242 * config.gcc (supported_defaults): Add madd4.
35243 (with_madd4): Add validation.
35244 (all_defaults): Add madd4.
35245 * config/mips/mips.opt (mmadd4): New option.
35246 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
35248 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
35250 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
35251 (ISA_HAS_FUSED_MADD4): Likewise.
35252 * doc/invoke.texi (-mmadd4): Document the new option.
35253 * doc/install.texi (--with-madd4): Document the new option.
35255 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35257 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
35258 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
35259 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
35260 (aarch64_init_pauth_hint_builtins): New.
35261 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
35262 (aarch64_expand_builtin): Expand new builtins.
35264 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35266 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
35267 * combine-stack-adj.c (no_unhandled_cfa): Handle
35268 REG_CFA_TOGGLE_RA_MANGLE.
35269 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
35270 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
35271 info for return address signing.
35272 (aarch64_expand_epilogue): Likewise.
35274 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35276 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
35277 * config/aarch64/aarch64-protos.h
35278 (aarch64_return_address_signing_enabled): New declaration.
35279 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
35281 (aarch64_expand_prologue): Sign return address before it's pushed onto
35283 (aarch64_expand_epilogue): Authenticate return address fetched from
35285 (aarch64_override_options): Sanity check for ILP32 and ISA level.
35286 (aarch64_attributes): New function attributes for "sign-return-address".
35287 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
35288 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
35289 ("*do_return"): Generate combined instructions according to key index.
35290 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
35291 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
35293 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
35294 * config/aarch64/aarch64.opt (msign-return-address=): New.
35295 * doc/extend.texi (AArch64 Function Attributes): Documents
35296 "sign-return-address=".
35297 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
35299 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
35301 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
35302 overall option summary.
35304 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35306 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
35307 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
35308 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
35309 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
35311 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
35313 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
35314 -mpower9-minmax by default for -mcpu=power9.
35315 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
35316 128-bit floating point.
35318 2017-01-20 Alan Modra <amodra@gmail.com>
35320 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
35321 optimizing for size.
35323 2017-01-20 Alan Modra <amodra@gmail.com>
35326 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
35327 for strcmp and strncmp from corresponding builtin decl.
35329 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
35331 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
35332 instead of i386/rtems-64.h.
35333 * config/i386/rtems-64.h: Remove.
35335 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
35339 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
35341 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
35343 2017-01-19 Tamar Christina <tamar.christina@arm.com>
35345 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
35346 Change int to HOST_WIDE_INT.
35347 * config/aarch64/aarch64-protos.h
35348 (aarch64_simd_gen_const_vector_dup): Likewise.
35349 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
35351 2017-01-19 David Malcolm <dmalcolm@redhat.com>
35353 * langhooks-def.h (lhd_type_for_size): New decl.
35354 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
35355 * langhooks.c (lhd_type_for_size): New function, taken from
35358 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
35360 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
35361 define_bypass for CR latency.
35362 (power9-cracked-alu): Update bypass latency and remove power9-branch.
35363 (power9-alu2): Add define_bypass for CR latency.
35365 (power9-mul): Update insn latency.
35366 (power9-mul-compare): Update insn latency, bypass latency and remove
35369 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35371 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
35373 * config/aarch64/aarch64.md
35374 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
35375 aarch64_nopcrelative_literal_loads.
35376 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
35378 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
35380 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
35381 TARGET_LOONGSON_3A.
35382 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
35384 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
35387 * config.gcc (supported_defaults): Add lxc1-sxc1.
35388 (with_lxc1_sxc1): Add validation.
35389 (all_defaults): Add lxc1-sxc1.
35390 * config/mips/mips.opt (mlxc1-sxc1): New option.
35391 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
35393 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
35394 __mips_no_lxc1_sxc1.
35395 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
35396 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
35397 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
35399 2017-01-19 Richard Biener <rguenther@suse.de>
35401 PR tree-optimization/72488
35402 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
35403 sure to restore SSA info.
35404 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
35406 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
35408 PR rtl-optimization/79121
35409 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
35410 of the inner type when shifting an extended value.
35412 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
35415 * symtab.c (symtab_node::equal_address_to): Fix comparing of
35416 interposable aliases.
35418 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
35421 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
35422 Use the evmergelohi instruction.
35423 (mov_si<mode>_e500_subreg4_2_le): Likewise.
35424 (mov_sitf_e500_subreg8_2_be): Likewise.
35425 (mov_sitf_e500_subreg12_2_le): Likewise.
35426 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
35427 (mov_si<mode>_e500_subreg4_2_be): Likewise.
35428 (mov_sitf_e500_subreg8_2_le): Likewise.
35429 (mov_sitf_e500_subreg12_2_be): Likewise.
35431 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35433 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
35434 attribute from vecsimple to vecperm.
35435 (altivec_vbpermq2): Likewise.
35437 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35440 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
35442 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35443 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
35444 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
35445 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
35446 case where N arg is SIZE_MAX.
35447 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
35448 (cmpstrsi): Add pattern.
35450 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
35452 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
35453 __builtin_vec_revb builtins.
35454 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
35455 built-in functions to support generation of the ISA 3.0 XXBR<x>
35456 vector byte reverse instructions.
35457 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
35458 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
35459 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
35460 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
35461 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
35462 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
35463 (P9V_BUILTIN_VEC_REVB): Likewise.
35464 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
35465 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
35466 (p9_xxbrq_v16qi): Likewise.
35467 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
35468 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
35469 (p9_xxbrh_v8hi): Likewise.
35470 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
35471 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
35472 vec_revb built-in functions.
35474 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
35476 PR rtl-optimization/78952
35477 * config/i386/i386.md (any_extract): New code iterator.
35478 (*insvqi_2): Use any_extract for source operand.
35479 (*insvqi_3): Use any_shiftrt for source operand.
35481 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
35483 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
35485 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
35487 2017-01-18 Matthias Klose <doko@ubuntu.com>
35489 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
35491 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35493 * config/rs6000/altivec.h (vec_bperm): Change #define.
35494 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
35495 (altivec_vbpermq2): New define_insn.
35496 (altivec_vbpermd): Likewise.
35497 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
35498 function interface.
35499 (VBPERMD): Likewise.
35500 (VBPERM): New polymorphic function interface.
35501 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
35502 Add entries for P9V_BUILTIN_VEC_VBPERM.
35503 * doc/extend.texi: Add interfaces for vec_bperm.
35505 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35507 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
35508 first letter of error messages.
35509 (s390_resolve_overloaded_builtin): Likewise.
35510 * config/s390/s390.c (s390_expand_builtin): Likewise.
35511 (s390_invalid_arg_for_unprototyped_fn): Likewise.
35512 (s390_valid_target_attribute_inner_p): Likewise.
35513 * config/s390/s390.md ("tabort"): Likewise.
35515 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
35517 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
35518 (ISA_AVOID_DIV_HILO): New macro.
35519 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
35520 (ISA_HAS_DDIV): Likewise.
35522 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35524 * doc/invoke.texi (fabi-version): Correct number of occurrences.
35526 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35528 * doc/invoke.texi (fabi-version): Spelling fix.
35530 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35533 * doc/invoke.texi (fabi-version): Mention mangling fix for
35536 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35539 * doc/invoke.texi (fabi-version): Document discriminator mangling.
35541 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
35544 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
35545 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
35547 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
35549 (stack_protect_test): Ditto.
35550 * config/rs6000/rs6000.opt (mstack-protector-guard=,
35551 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
35553 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
35554 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
35555 -mstack-protector-guard-offset=.
35556 (RS/6000 and PowerPC Options): Ditto.
35558 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
35560 * config/i386/i386.h (MASK_CLASS_P): New define.
35561 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
35562 there are no registers from different register sets also when
35563 mask registers are used. Update function comment.
35564 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
35565 to (*k/*r) and (*k/*km) alternatives.
35567 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
35569 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
35570 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
35571 (EH_RETURN_HANDLER_RTX): New define.
35572 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
35573 Force frame pointer in EH return functions.
35574 (aarch64_expand_epilogue): Add barrier for eh_return.
35575 (aarch64_final_eh_return_addr): Remove.
35576 (aarch64_eh_return_handler_rtx): New function.
35577 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
35579 (aarch64_eh_return_handler_rtx): New prototype.
35581 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35583 * config/rs6000/altivec.h (vec_rlmi): New #define.
35584 (vec_vrlnm): Likewise.
35585 (vec_rlnm): Likewise.
35586 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
35587 (UNSPEC_VRLNM): Likewise.
35588 (VIlong): New mode iterator.
35589 (altivec_vrl<VI_char>mi): New define_insn.
35590 (altivec_vrl<VI_char>nm): Likewise.
35591 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
35593 (VRLDNM): Likewise.
35594 (RLNM): New polymorphic function entry.
35595 (VRLWMI): New monomorphic function entry.
35596 (VRLDMI): Likewise.
35597 (RLMI): New polymorphic function entry.
35598 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
35599 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
35600 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
35603 2017-01-17 Jakub Jelinek <jakub@redhat.com>
35606 * dwarf2out.c (field_byte_offset): Restore the
35607 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
35608 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
35609 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
35612 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
35615 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
35617 2017-01-17 Jakub Jelinek <jakub@redhat.com>
35620 * dwarf2out.c (add_data_member_location_attribute): For constant
35621 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
35622 instead of DW_AT_data_member_location, DW_AT_bit_offset and
35623 DW_AT_byte_size attributes.
35625 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
35627 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
35628 after forcing to constant memory when the code model is medium.
35630 2017-01-17 Julia Koval <julia.koval@intel.com>
35633 * config/i386/avx512fintrin.h
35634 (_mm512_i32gather_ps): Change __addr type to void const*.
35635 (_mm512_mask_i32gather_ps): Ditto.
35636 (_mm512_i32gather_pd): Ditto.
35637 (_mm512_mask_i32gather_pd): Ditto.
35638 (_mm512_i64gather_ps): Ditto.
35639 (_mm512_mask_i64gather_ps): Ditto.
35640 (_mm512_i64gather_pd): Ditto.
35641 (_mm512_mask_i64gather_pd): Ditto.
35642 (_mm512_i32gather_epi32): Ditto.
35643 (_mm512_mask_i32gather_epi32): Ditto.
35644 (_mm512_i32gather_epi64): Ditto.
35645 (_mm512_mask_i32gather_epi64): Ditto.
35646 (_mm512_i64gather_epi32): Ditto.
35647 (_mm512_mask_i64gather_epi32): Ditto.
35648 (_mm512_i64gather_epi64): Ditto.
35649 (_mm512_mask_i64gather_epi64): Ditto.
35650 (_mm512_i32scatter_ps): Change __addr type to void*.
35651 (_mm512_mask_i32scatter_ps): Ditto.
35652 (_mm512_i32scatter_pd): Ditto.
35653 (_mm512_mask_i32scatter_pd): Ditto.
35654 (_mm512_i64scatter_ps): Ditto.
35655 (_mm512_mask_i64scatter_ps): Ditto.
35656 (_mm512_i64scatter_pd): Ditto.
35657 (_mm512_mask_i64scatter_pd): Ditto.
35658 (_mm512_i32scatter_epi32): Ditto.
35659 (_mm512_mask_i32scatter_epi32): Ditto.
35660 (_mm512_i32scatter_epi64): Ditto.
35661 (_mm512_mask_i32scatter_epi64): Ditto.
35662 (_mm512_i64scatter_epi32): Ditto.
35663 (_mm512_mask_i64scatter_epi32): Ditto.
35664 (_mm512_i64scatter_epi64): Ditto.
35665 (_mm512_mask_i64scatter_epi64): Ditto.
35666 * config/i386/avx512pfintrin.h
35667 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
35668 (_mm512_mask_prefetch_i32gather_ps): Ditto.
35669 (_mm512_mask_prefetch_i64gather_pd): Ditto.
35670 (_mm512_mask_prefetch_i64gather_ps): Ditto.
35671 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
35672 (_mm512_prefetch_i32scatter_ps): Ditto.
35673 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
35674 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
35675 (_mm512_prefetch_i64scatter_pd): Ditto.
35676 (_mm512_prefetch_i64scatter_ps): Ditto.
35677 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
35678 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
35679 * config/i386/avx512vlintrin.h
35680 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
35681 (_mm_mmask_i32gather_ps): Ditto.
35682 (_mm256_mmask_i32gather_pd): Ditto.
35683 (_mm_mmask_i32gather_pd): Ditto.
35684 (_mm256_mmask_i64gather_ps): Ditto.
35685 (_mm_mmask_i64gather_ps): Ditto.
35686 (_mm256_mmask_i64gather_pd): Ditto.
35687 (_mm_mmask_i64gather_pd): Ditto.
35688 (_mm256_mmask_i32gather_epi32): Ditto.
35689 (_mm_mmask_i32gather_epi32): Ditto.
35690 (_mm256_mmask_i32gather_epi64): Ditto.
35691 (_mm_mmask_i32gather_epi64): Ditto.
35692 (_mm256_mmask_i64gather_epi32): Ditto.
35693 (_mm_mmask_i64gather_epi32): Ditto.
35694 (_mm256_mmask_i64gather_epi64): Ditto.
35695 (_mm_mmask_i64gather_epi64): Ditto.
35696 (_mm256_i32scatter_ps): Change __addr type to void*.
35697 (_mm256_mask_i32scatter_ps): Ditto.
35698 (_mm_i32scatter_ps): Ditto.
35699 (_mm_mask_i32scatter_ps): Ditto.
35700 (_mm256_i32scatter_pd): Ditto.
35701 (_mm256_mask_i32scatter_pd): Ditto.
35702 (_mm_i32scatter_pd): Ditto.
35703 (_mm_mask_i32scatter_pd): Ditto.
35704 (_mm256_i64scatter_ps): Ditto.
35705 (_mm256_mask_i64scatter_ps): Ditto.
35706 (_mm_i64scatter_ps): Ditto.
35707 (_mm_mask_i64scatter_ps): Ditto.
35708 (_mm256_i64scatter_pd): Ditto.
35709 (_mm256_mask_i64scatter_pd): Ditto.
35710 (_mm_i64scatter_pd): Ditto.
35711 (_mm_mask_i64scatter_pd): Ditto.
35712 (_mm256_i32scatter_epi32): Ditto.
35713 (_mm256_mask_i32scatter_epi32): Ditto.
35714 (_mm_i32scatter_epi32): Ditto.
35715 (_mm_mask_i32scatter_epi32): Ditto.
35716 (_mm256_i32scatter_epi64): Ditto.
35717 (_mm256_mask_i32scatter_epi64): Ditto.
35718 (_mm_i32scatter_epi64): Ditto.
35719 (_mm_mask_i32scatter_epi64): Ditto.
35720 (_mm256_i64scatter_epi32): Ditto.
35721 (_mm256_mask_i64scatter_epi32): Ditto.
35722 (_mm_i64scatter_epi32): Ditto.
35723 (_mm_mask_i64scatter_epi32): Ditto.
35724 (_mm256_i64scatter_epi64): Ditto.
35725 (_mm256_mask_i64scatter_epi64): Ditto.
35726 (_mm_i64scatter_epi64): Ditto.
35727 (_mm_mask_i64scatter_epi64): Ditto.
35728 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
35729 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
35730 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
35731 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
35732 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
35733 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
35734 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
35735 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
35736 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
35737 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
35738 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
35739 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
35740 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
35741 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
35742 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
35743 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
35744 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
35745 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
35746 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
35747 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
35748 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
35749 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
35750 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
35751 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
35752 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
35753 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
35754 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
35755 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
35756 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
35757 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
35758 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
35759 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
35760 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
35761 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
35762 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
35763 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
35764 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
35765 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
35766 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
35767 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
35768 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
35769 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
35770 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
35771 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
35772 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
35773 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
35774 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
35775 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
35776 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
35777 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
35778 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
35779 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
35780 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
35781 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
35782 definitions accordingly.
35784 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
35785 Kuan-Lin Chen <kuanlinchentw@gmail.com>
35788 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
35791 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
35794 * ira-conflicts.c (ira_build_conflicts): Update total conflict
35795 hard regs for inner regno.
35797 2017-01-17 Martin Liska <mliska@suse.cz>
35800 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
35801 assumption and add comment.
35803 2017-01-17 Nathan Sidwell <nathan@acm.org>
35805 * ipa-visibility.c (localize_node): New function, broken out of ...
35806 (function_and_variable_visibility): ... here. Call it.
35808 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
35810 PR middle-end/77445
35811 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
35812 correctly set frequency of oudgoing edge.
35813 (duplicate_thread_path): Fix profile updating.
35815 2017-01-17 Jakub Jelinek <jakub@redhat.com>
35818 * configure.ac: Add GCC_BASE_VER.
35819 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
35820 version from BASE-VER file.
35821 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
35822 (gcc.o): Depend on $(BASEVER).
35823 * common.opt (dumpfullversion): New option.
35824 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
35825 * doc/invoke.texi: Document -dumpfullversion.
35826 * doc/install.texi: Document --with-gcc-major-version-only.
35827 * configure: Regenerated.
35829 2017-01-17 Richard Biener <rguenther@suse.de>
35831 PR tree-optimization/71433
35832 * tree-vrp.c (register_new_assert_for): Merge same asserts
35833 on all incoming edges.
35834 (process_assert_insertions_for): Handle insertions at the
35837 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
35839 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
35840 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
35842 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
35845 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
35848 2017-01-17 Alan Modra <amodra@gmail.com>
35851 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
35852 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
35853 symbolic stack limit when pic.
35855 2017-01-16 Martin Sebor <msebor@redhat.com>
35857 PR tree-optimization/78608
35858 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
35860 2017-01-16 Jeff Law <law@redhat.com>
35863 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
35864 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
35865 for several include directories that may be relative to sysroot.
35866 * config/i386/x-mingw32 (gplus_includedir): Define.
35867 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
35868 (native_system_includedir): Likewise.
35869 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
35870 override if TARGET_SYSTEM_ROOT is defined.
35871 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
35873 PR tree-optimization/79090
35874 PR tree-optimization/33562
35875 PR tree-optimization/61912
35876 PR tree-optimization/77485
35877 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
35878 and computed trims into the dump file.
35880 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
35882 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
35884 2017-01-16 Jakub Jelinek <jakub@redhat.com>
35887 * gimplify.c (gimplify_init_constructor): If want_value and
35888 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
35892 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
35893 sequence. Formatting fixes.
35894 (doloop_optimize): Formatting fixes.
35897 * gcc.c (debug_level_greater_than_spec_func): New function.
35898 (static_spec_functions): Add debug-level-gt spec function.
35899 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
35901 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
35902 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
35903 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
35904 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
35905 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
35906 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
35908 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
35910 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
35911 QImode fixups to general and mask registers only.
35913 2017-01-16 Carl Love <cel@us.ibm.com>
35915 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
35916 for built-in functions
35917 vector signed char vec_nabs (vector signed char)
35918 vector signed short vec_nabs (vector signed short)
35919 vector signed int vec_nabs (vector signed int)
35920 vector signed long long vec_nabs (vector signed long long)
35921 vector float vec_nabs (vector float)
35922 vector double vec_nabs (vector double)
35923 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
35925 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
35926 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
35927 * doc/extend.texi: Update the documentation file for the new built-in
35930 2017-01-16 Martin Sebor <msebor@redhat.com>
35932 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
35935 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35937 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
35938 UNSPEC_VSX__XXSPLTD to require special splat handling.
35940 2017-01-16 David Malcolm <dmalcolm@redhat.com>
35943 * system.h: Poison strndup.
35945 2017-01-16 Alan Modra <amodra@gmail.com>
35948 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
35951 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
35953 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
35955 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
35957 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
35958 call recog here. Assert that INSN_CODE (insn) is non-negative.
35960 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
35963 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
35965 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
35966 in the currently scheduled RTL fragment.
35968 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
35970 PR rtl-optimization/78751
35971 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
35974 2017-01-14 Jeff Law <law@redhat.com>
35976 PR tree-optimization/79090
35977 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
35978 variable length stores.
35979 (compute_trims): Delete dead assignment to *trim_tail.
35980 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
35983 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
35985 PR rtl-optimization/78626
35986 PR rtl-optimization/78727
35987 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
35988 of a block, and split such blocks after everything else is finished.
35990 2017-01-14 Alan Modra <amodra@gmail.com>
35993 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
35994 target legitimate_combined_insn.
35995 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
35996 (rs6000_legitimate_combined_insn): New function.
35997 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
35999 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
36000 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
36001 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
36003 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
36005 * doc/frontends.texi (G++ and GCC): Remove references to Java.
36007 2017-01-13 Jeff Law <law@redhat.com>
36009 PR tree-optimization/33562
36010 PR tree-optimization/61912
36011 PR tree-optimization/77485
36012 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
36014 (delete_dead_assignment): Likewise.
36015 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
36016 statement to delete_dead_call and delete_dead_assignment.
36018 2017-01-13 David Malcolm <dmalcolm@redhat.com>
36021 * substring-locations.c (format_warning_va): Strengthen case 1 so
36022 that both endpoints of the substring must be within the format
36023 range for just the substring to be printed.
36025 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
36027 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
36028 * config/i386/i386.c (ix86_target_string): Add missing options
36029 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
36030 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
36031 flags_other and ix86_target_other to flags2_other. Display unknown
36033 (ix86_valid_target_attribute_inner_p): Add missing options and
36034 reorder options by implied ISAs, as in ix86_target_string.
36036 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
36038 * hash-table.h (hash_table::too_empty_p): New function.
36039 (hash_table::expand): Use it.
36040 (hash_table::traverse): Likewise.
36041 (hash_table::empty_slot): Use sizeof (value_type) instead of
36042 sizeof (PTR) to convert bytes to elements. Shrink the table
36043 if the current size is excessive for the current number of
36046 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
36048 * ira-costs.c (record_reg_classes): Break from the inner loop
36049 early once alt_fail is known to be true. Update outer loop
36050 handling accordingly.
36052 2017-01-13 Jeff Law <law@redhat.com>
36054 * tree-ssa-dse.c (decrement_count): New function.
36055 (increment_start_addr, maybe_trim_memstar_call): Likewise.
36056 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
36057 when we know the partially dead statement is a mem* function.
36059 PR tree-optimization/61912
36060 PR tree-optimization/77485
36061 * tree-ssa-dse.c: Include expr.h.
36062 (maybe_trim_constructor_store): New function.
36063 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
36065 PR tree-optimization/33562
36066 PR tree-optimization/61912
36067 PR tree-optimization/77485
36068 * doc/invoke.texi: Document new dse-max-object-size param.
36069 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
36070 * tree-ssa-dse.c: Include params.h.
36071 (dse_store_status): New enum.
36072 (initialize_ao_ref_for_dse): New, partially extracted from
36074 (valid_ao_ref_for_dse, normalize_ref): New.
36075 (setup_live_bytes_from_ref, compute_trims): Likewise.
36076 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
36077 (maybe_trim_partially_dead_store): Likewise.
36078 (maybe_trim_complex_store): Likewise.
36079 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
36080 Track what bytes live from the original store. Return tri-state
36081 for dead, partially dead or live.
36082 (dse_dom_walker): Add constructor, destructor and new private members.
36083 (delete_dead_call, delete_dead_assignment): New extracted from
36085 (dse_optimize_stmt): Make a member of dse_dom_walker.
36086 Use initialize_ao_ref_for_dse.
36088 PR tree-optimization/33562
36089 PR tree-optimization/61912
36090 PR tree-optimization/77485
36091 * sbitmap.h (bitmap_count_bits): Prototype.
36092 (bitmap_clear_range, bitmap_set_range): Likewise.
36093 * sbitmap.c (bitmap_clear_range): New function.
36094 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
36096 2017-01-13 Martin Liska <mliska@suse.cz>
36099 * function.c (set_cfun): Add new argument force.
36100 * function.h (set_cfun): Likewise.
36101 * ipa-inline-transform.c (inline_call): Use the function when
36102 strict alising from is dropped for function we inline to.
36104 2017-01-13 Richard Biener <rguenther@suse.de>
36106 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
36107 for dumping GIMPLE INTEGER_CSTs.
36109 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
36111 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
36112 to 201112L since C++17.
36114 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
36117 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
36118 if -fsanitize=kernel-address is present.
36120 2017-01-13 Richard Biener <rguenther@suse.de>
36122 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
36123 as _Literal ( type ) number in case usual suffixes do not
36124 preserve all information.
36126 2017-01-13 Richard Biener <rguenther@suse.de>
36128 PR tree-optimization/77283
36129 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
36130 and ssa-iterators.h.
36131 (is_feasible_trace): Implement a cost model based on joiner
36134 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
36137 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
36138 char or short to __float128/_Float128 directly.
36140 2017-01-12 Martin Sebor <msebor@redhat.com>
36142 to -Wformat-overflow.
36143 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
36144 (min_bytes_remaining): Same.
36145 (get_string_length): Same.
36146 (format_string): Same.
36147 (format_directive): Same.
36149 (pass_sprintf_length::handle_gimple_call): Same.
36151 2017-01-12 Jakub Jelinek <jakub@redhat.com>
36153 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
36154 info.nowrite calls with no lhs that can't throw. Return bool
36155 whether gsi_remove has been called or not.
36156 (pass_sprintf_length::handle_gimple_call): Return bool whether
36157 try_substitute_return_value called gsi_remove. Formatting fix.
36158 (pass_sprintf_length::execute): Don't use gsi_remove if
36159 handle_gimple_call returned true.
36162 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
36163 be removed due to side-effects, don't remove following barrier nor
36164 turn the successor edge into fallthru edge.
36166 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36169 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
36170 element-reversing loads and stores as not swappable.
36172 2017-01-12 Nathan Sidwell <nathan@acm.org>
36173 Nicolai Stange <nicstange@gmail.com>
36175 * combine.c (try_combine): Don't ignore result of overlap checking
36176 loop. Combine overlap & asm check into single loop.
36178 2017-01-12 Richard Biener <rguenther@suse.de>
36180 * tree-pretty-print.c (dump_generic_node): Provide -gimple
36181 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
36183 2017-01-12 Richard Biener <rguenther@suse.de>
36185 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
36186 and TS_TARGET_OPTION directly derive from TS_BASE.
36187 * tree-core.h (tree_optimization_option): Derive from tree_base.
36188 (tree_target_option): Likewise.
36190 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
36192 * config/i386/i386.c (memory_address_length): Increase len
36193 only when rip_relative_addr_p returns false.
36195 2017-01-11 Julia Koval <julia.koval@intel.com>
36197 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
36198 (OPTION_MASK_ISA_SGX_SET): New.
36199 (ix86_handle_option): Handle OPT_msgx.
36200 * config.gcc: Added sgxintrin.h.
36201 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
36202 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
36203 * config/i386/i386.c (ix86_target_string): Add -msgx.
36205 (ix86_option_override_internal): Handle new options.
36206 (ix86_valid_target_attribute_inner_p): Add sgx.
36207 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
36208 * config/i386/i386.opt: Add msgx.
36209 * config/i386/sgxintrin.h: New file.
36210 * config/i386/x86intrin.h: Add sgxintrin.h.
36212 2017-01-11 Jakub Jelinek <jakub@redhat.com>
36215 * fold-const.c (maybe_nonzero_address): Return 1 for function
36217 (tree_single_nonzero_warnv_p): Don't handle function local objects
36221 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
36224 2017-01-11 David Malcolm <dmalcolm@redhat.com>
36227 * opts.c: Include "spellcheck.h"
36228 (struct string_fragment): New struct.
36229 (struct edit_distance_traits<const string_fragment &>): New
36231 (get_closest_sanitizer_option): New function.
36232 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
36234 2017-01-11 Jakub Jelinek <jakub@redhat.com>
36236 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
36238 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
36239 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
36240 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
36241 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
36242 for initial die_offset if dwarf_split_debug_info.
36243 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
36244 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
36246 (output_skeleton_debug_sections): Formatting fix. Use
36247 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
36248 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
36250 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
36252 * config/arm/cortex-a53.md: Add bypasses for
36253 cortex_a53_r2f_cvt.
36254 (cortex_a53_r2f): Only use for transfers.
36255 (cortex_a53_f2r): Likewise.
36256 (cortex_a53_r2f_cvt): Add reservation for conversions.
36257 (cortex_a53_f2r_cvt): Likewise.
36259 2017-01-11 Tamar Christina <tamar.christina@arm.com>
36261 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
36262 to all inlined functions, change static to extern.
36264 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
36267 * config/arm/arm.c (legitimize_pic_address): Handle reference to
36269 (arm_assemble_integer): Likewise.
36271 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
36273 * config.gcc: Use new awk script to check CPU, FPU and architecture
36274 parameters for --with-... options.
36275 * config/arm/parsecpu.awk: New file
36276 * config/arm/arm-cpus.in: New file.
36277 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
36279 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
36281 * config/arm/t-arm: Update dependency rules.
36282 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
36283 of processing .def files.
36284 * config/arm/genopt.sh: Deleted.
36285 * config/arm/gentune.sh: Deleted.
36286 * config/arm/arm-cores.def: Deleted.
36287 * config/arm/arm-arches.def: Deleted.
36288 * config/arm/arm-fpus.def: Deleted.
36289 * config/arm/arm-tune.md: Regenerated.
36290 * config/arm/arm-tables.opt: Regenerated.
36291 * config/arm/arm-cpu.h: New generated file.
36292 * config/arm/arm-cpu-data.h: New generated file.
36293 * config/arm/arm-cpu-cdata.h: New generated file.
36295 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
36298 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
36300 (input_varpool_node): Unpack dynamically_initialized bit.
36302 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
36304 PR rtl-optimization/79032
36305 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
36306 the alignment of the adjusted memory reference against that of MODE,
36307 instead of the alignment of the original memory reference.
36309 2017-01-11 Martin Jambor <mjambor@suse.cz>
36311 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
36313 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
36314 decorated functions.
36316 2017-01-11 Richard Biener <rguenther@suse.de>
36318 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
36319 set range/nonnull info for PHI results. Do not set it on
36320 stmts marked for removal.
36322 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
36324 * expr.c (store_field): In the bitfield case, fetch the return value
36325 from the registers before applying a single big-endian adjustment.
36326 Always do a final load for a BLKmode value not larger than a word.
36328 2017-01-10 David Malcolm <dmalcolm@redhat.com>
36331 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
36332 that we correctly handle column numbers greater than
36333 LINE_MAP_MAX_COLUMN_NUMBER.
36335 2017-01-10 Martin Sebor <msebor@redhat.com>
36337 PR middle-end/78245
36338 * gimple-ssa-sprintf.c (get_destination_size): Call
36339 {init,fini}object_sizes.
36340 * tree-object-size.c (addr_object_size): Adjust.
36341 (pass_through_call): Adjust.
36342 (pass_object_sizes::execute): Adjust.
36343 * tree-object-size.h (fini_object_sizes): Declare.
36345 2017-01-10 Martin Sebor <msebor@redhat.com>
36347 PR tree-optimization/78775
36348 * builtins.c (get_size_range): Move...
36349 * calls.c: ...to here.
36350 (alloc_max_size): Accept zero argument.
36351 (operand_signed_p): Remove.
36352 (maybe_warn_alloc_args_overflow): Call get_size_range.
36353 * calls.h (get_size_range): Declare.
36355 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
36357 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
36358 from TI's devices.csv file as of September 2016.
36359 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
36361 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
36363 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
36364 * doc/invoke.texi: Likewise.
36365 * doc/md.texi: Likewise.
36366 * doc/objc.texi: Likewise.
36368 2017-01-10 Joshua Conner <joshconner@google.com>
36370 * config/arm/fuchsia-elf.h: New file.
36371 * config/fuchsia.h: New file.
36372 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
36373 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
36375 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
36377 2016-01-10 Richard Biener <rguenther@suse.de>
36379 PR tree-optimization/79034
36380 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
36381 Propagate out degenerate PHIs in the joiner.
36383 2017-01-10 Martin Liska <mliska@suse.cz>
36385 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
36386 (sort_congruence_classes_by_decl_uid): Likewise.
36387 (sort_congruence_class_groups_by_decl_uid): Likewise.
36388 (sem_item_optimizer::merge_classes): Sort class, groups in these
36389 classes and members in the groups by DECL_UID of declarations.
36390 This would make merge operations stable.
36392 2017-01-10 Martin Liska <mliska@suse.cz>
36394 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
36395 usage of m_classes_vec.
36396 (sem_item_optimizer::~sem_item_optimizer): Likewise.
36397 (sem_item_optimizer::get_group_by_hash): Likewise.
36398 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
36399 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
36400 (sem_item_optimizer::verify_classes): Likewise.
36401 (sem_item_optimizer::process_cong_reduction): Likewise.
36402 (sem_item_optimizer::dump_cong_classes): Likewise.
36403 (sem_item_optimizer::merge_classes): Likewise.
36404 * ipa-icf.h (congruence_class_hash): Rename from
36405 congruence_class_group_hash. Remove declaration of m_classes_vec.
36407 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
36409 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
36410 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
36411 * config.gcc: Add avx512vpopcntdqintrin.h.
36412 * config/i386/avx512vpopcntdqintrin.h: New.
36413 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
36414 * config/i386/i386-builtin-types.def: Add new types.
36415 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
36416 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
36417 __builtin_ia32_vpopcountq_v8di_mask): New.
36418 * config/i386/i386-c.c (ix86_target_macros_internal): Define
36419 __AVX512VPOPCNTDQ__.
36420 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
36421 (PTA_AVX512VPOPCNTDQ): Define.
36422 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
36423 TARGET_AVX512VPOPCNTDQ_P): Define.
36424 * config/i386/i386.opt: Add mavx512vpopcntdq.
36425 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
36426 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
36428 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36430 PR middle-end/77484
36431 * predict.def (PRED_CALL): Set to 67.
36433 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
36435 * expr.c (store_field): In the bitfield case, if the value comes from
36436 a function call and is of an aggregate type returned in registers, do
36437 not modify the field mode; extract the value in all cases if the mode
36438 is BLKmode and the size is not larger than a word.
36440 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
36443 * config/i386/cpuid.h: Fix undefined behavior.
36445 2017-01-04 Jeff Law <law@redhat.com>
36447 PR tree-optimization/79007
36448 PR tree-optimization/67955
36449 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
36450 conservative for pt.null when flag_non_call_exceptions is on.
36452 2017-01-09 Jakub Jelinek <jakub@redhat.com>
36454 PR translation/79019
36455 PR translation/79020
36456 * params.def (PARAM_INLINE_MIN_SPEEDUP,
36457 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
36458 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
36460 * config/avr/avr.opt (maccumulate-args): Likewise.
36461 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
36462 * common.opt (freport-bug): Likewise.
36463 * cif-code.def (CIF_FINAL_ERROR): Likewise.
36464 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
36465 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
36466 translatable string.
36467 * config/i386/i386.c (function_value_32): Likewise.
36468 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
36469 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
36471 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
36472 * common/config/msp430/msp430-common.c (msp430_handle_option):
36474 * symtab.c (symtab_node::verify_base): Likewise.
36475 * opts.c (set_debug_level): Likewise.
36476 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
36477 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
36478 missing whitespace to translatable strings.
36479 * config/avr/avr.md (bswapsi2): Fix typo in comment.
36480 * config/sh/superh.h: Likewise.
36481 * config/i386/xopintrin.h: Likewise.
36482 * config/i386/znver1.md: Likewise.
36483 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
36484 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
36485 * double-int.h (struct double_int): Likewise.
36486 * double-int.c (div_and_round_double): Likewise.
36487 * wide-int.cc: Likewise.
36488 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
36489 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
36490 * cfgcleanup.c (crossjumps_occured): Renamed to ...
36491 (crossjumps_occurred): ... this.
36492 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
36495 PR tree-optimization/78899
36496 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
36497 returning bool return struct loop *, NULL for failure and the new
36499 (versionable_outer_loop_p): Don't version outer loop if it has
36500 dont_vectorized bit set.
36501 (tree_if_conversion): When versioning outer loop, ensure
36502 tree_if_conversion is performed also on the inner loop of the
36503 non-vectorizable outer loop copy.
36504 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
36505 LOOP_VECTORIZED in inner loop of the scalar outer loop and
36506 prevent vectorization of it.
36507 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
36508 the outer loop vectorization of the non-scalar version is attempted
36509 before vectorization of the inner loop in scalar version. If
36510 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
36511 vectorization of its inner loop.
36512 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
36513 has 2 inner loops, rename also on edges from bb whose single pred
36514 is outer_loop->header. Fix typo in function comment.
36516 2017-01-09 Martin Sebor <msebor@redhat.com>
36519 * asan.c (asan_emit_stack_protection): Increase local buffer size
36520 to avoid snprintf truncation warning.
36522 2017-01-09 Andrew Pinski <apinski@cavium.com>
36524 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
36525 to reference thunderx2t99 for the tuning structure
36526 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
36528 (thunderx2t99_extra_costs): This.
36529 * config/aarch64/aarch64-tune.md: Regenerate.
36530 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
36531 (vulcan_addrcost_table): This.
36532 (vulcan_regmove_cost): Rename to ...
36533 (thunderx2t99_regmove_cost): This.
36534 (vulcan_vector_cost): Rename to ...
36535 (thunderx2t99_vector_cost): this.
36536 (vulcan_branch_cost): Rename to ...
36537 (thunderx2t99_branch_cost): This.
36538 (vulcan_tunings): Rename to ...
36539 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
36540 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
36542 2017-01-09 Martin Jambor <mjambor@suse.cz>
36546 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
36547 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
36548 (propagate_vr_accross_jump_function): Use the above function for all
36549 value range computations for pass-through jump functions and type
36550 converasion from explicit value range values.
36551 (ipcp_propagate_stage): Do not attempt to deduce types of formal
36552 parameters from TYPE_ARG_TYPES.
36553 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
36554 (ipa_write_node_info): Stream type of the actual argument.
36555 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
36557 2017-01-09 Martin Liska <mliska@suse.cz>
36560 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
36561 (lookup_compiler): Do not show error message with have_E.
36563 2017-01-09 Jakub Jelinek <jakub@redhat.com>
36565 PR tree-optimization/78938
36566 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
36567 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
36568 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
36569 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
36572 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36574 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
36577 2017-01-09 Richard Biener <rguenther@suse.de>
36579 PR tree-optimization/78997
36580 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
36581 name condition properly.
36583 2017-01-09 Richard Biener <rguenther@suse.de>
36586 * dwarf2out.c (is_cxx): New overload with context.
36587 (is_naming_typedef_decl): Use it.
36589 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
36591 * invoke.texi (Option Summary): Correct spacing in option lists
36592 and add line breaks to fix over-long lines.
36594 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
36596 PR middle-end/17660
36598 * extend.texi (Common Variable Attributes): Add xref to GCC
36599 Internals manual to explain mode attribute keywords.
36601 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
36604 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
36605 and Preprocessor Options.
36606 (Options for Linking): Document -pthread here....
36607 (RS/6000 and PowerPC Options): ...not here.
36608 (Solaris 2 Options): ...or here.
36609 * doc/cppopts.texi: Document -pthread.
36611 2017-01-08 Martin Sebor <msebor@redhat.com>
36613 PR middle-end/77708
36614 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
36615 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
36616 New member functions.
36617 (format_directive): Used them.
36619 (pass_sprintf_length::handle_gimple_call): Same.
36620 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
36621 to avoid truncation for any argument.
36622 (extract_affine_mul): Same.
36623 * tree.c (get_file_function_name): Same.
36625 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36627 PR middle-end/77484
36628 * predict.def (PRED_INDIR_CALL): Set to 86.
36630 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
36632 PR preprocessor/54124
36633 * doc/cppopts.texi: Reformat -d subtable to list the full name
36634 of the options. Add cross-reference to the docs for the general
36635 compiler -d options.
36636 * doc/invoke.texi (Developer Options): Add cross-reference to the
36637 preprocessor-specific -d option documentation.
36639 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
36641 PR preprocessor/13498
36642 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
36643 redudant material, and reflect new command-line options.
36644 (System Headers): Likewise.
36646 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
36648 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
36649 -isystem, and -idirafter. Copy-edit.
36650 * doc/cppopts.texi: Copy-edit. Remove contradiction about
36651 default for -ftrack-macro-expansion. Delete obsolete and
36652 badly-formatted implementation details about -fdebug-cpp output.
36653 * doc/cppwarnopts.texi: Copy-edit.
36655 2017-01-07 David Malcolm <dmalcolm@redhat.com>
36658 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
36659 that the transition from a max line width >= 1<<10 to narrower
36660 lines works correctly.
36662 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
36664 * doc/options.texi (PerFunction): New.
36665 * opt-functions.awk (switch_flags): Map both Optimization and
36666 PerFunction to CL_OPTIMIZATION.
36667 * opth-gen.awk: Test for PerFunction flag along with
36669 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
36670 it only when the latter is present. Skip those that don't in
36671 the hash function generator.
36672 * common.opt (fvar-tracking): Mark as PerFunction instead of
36674 (fvar-tracking-assignments): Likewise.
36675 (fvar-tracking-assignments-toggle): Likewise.
36676 (fvar-tracking-uninit): Likewise.
36678 2017-01-07 Jakub Jelinek <jakub@redhat.com>
36680 PR translation/79018
36681 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
36684 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
36687 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
36688 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
36689 TARGET_LONG_JUMP_TABLE_OFFSETS.
36690 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
36691 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
36692 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
36693 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
36694 * config/m68k/m68k.md (tablejump expander): Likewise.
36695 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
36696 TARGET_LONG_JUMP_TABLE_OFFSETS.
36697 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
36698 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
36700 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
36701 David Holsgrove <david.holsgrove@xilinx.com>
36703 * common/config/microblaze/microblaze-common.c
36704 (TARGET_EXCEPT_UNWIND_INFO): Remove.
36705 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
36707 * config/microblaze/microblaze.c (microblaze_must_save_register)
36708 (microblaze_expand_epilogue, microblaze_return_addr): Handle
36710 (microblaze_eh_return): New function.
36711 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
36712 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
36713 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
36714 * config/microblaze/microblaze.md (eh_return): New pattern.
36716 2017-01-06 Jakub Jelinek <jakub@redhat.com>
36718 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
36719 GCC_DIAGNOSTIC_STRINGIFY): Define.
36721 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
36723 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36725 * config/arm/arm.md (<mcrr>): New.
36727 * config/arm/arm.c (arm_arch5te): New.
36728 (arm_option_override): Set arm_arch5te.
36729 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
36731 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
36732 (arm_mcrr_qualifiers): ... this. New.
36733 (MRRC_QUALIFIERS): Define to...
36734 (arm_mrrc_qualifiers): ... this. New.
36735 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
36737 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
36738 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
36739 (MRRCI, mrrc, MRRC): New.
36740 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
36741 VUNSPEC_MRRC2): New.
36743 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36745 * config/arm/arm.md (<mcr>): New.
36747 * config/arm/arm.c (arm_coproc_builtin_available): Add
36748 support for mcr, mrc, mcr2 and mrc2.
36749 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
36750 (arm_mcr_qualifiers): ... this. New.
36751 (MRC_QUALIFIERS): Define to ...
36752 (arm_mrc_qualifiers): ... this. New.
36753 (MCR_QUALIFIERS): Define to ...
36754 (arm_mcr_qualifiers): ... this. New.
36755 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
36757 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
36758 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
36759 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
36760 VUNSPEC_MRC2): New.
36762 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36764 * config/arm/arm.md (*ldc): New.
36768 * config/arm/arm.c (arm_coproc_builtin_available): Add
36769 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
36770 (arm_coproc_ldc_stc_legitimate_address): New.
36771 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
36772 'qualifier_const_pointer'.
36773 (LDC_QUALIFIERS): Define to...
36774 (arm_ldc_qualifiers): ... this. New.
36775 (STC_QUALIFIERS): Define to...
36776 (arm_stc_qualifiers): ... this. New.
36777 * config/arm/arm-protos.h
36778 (arm_coproc_ldc_stc_legitimate_address): New.
36779 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
36780 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
36781 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
36782 stc2, stcl, stc2l): New.
36783 * config/arm/constraints.md (Uz): New.
36784 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
36785 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
36786 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
36787 VUNSPEC_STC2L): New.
36789 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36791 * config/arm/arm.md (<cdp>): New.
36792 * config/arm/arm.c (neon_const_bounds): Rename this ...
36793 (arm_const_bounds): ... this.
36794 (arm_coproc_builtin_available): New.
36795 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
36796 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
36797 (CDP_QUALIFIERS): Define to...
36798 (arm_cdp_qualifiers): ... this. New.
36800 (arm_expand_builtin_args): Add case for 6 arguments.
36801 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
36802 (arm_const_bounds): ... this.
36803 (arm_coproc_builtin_available): New.
36804 * config/arm/arm_acle.h (__arm_cdp): New.
36806 * config/arm/arm_acle_builtins.def (cdp): New.
36808 * config/arm/iterators.md (CDPI,CDP,cdp): New.
36809 * config/arm/neon.md: Rename all 'neon_const_bounds' to
36810 'arm_const_bounds'.
36811 * config/arm/types.md (coproc): New.
36812 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
36813 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
36814 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
36815 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
36817 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36819 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
36820 (UBINOP_QUALIFIERS): New.
36822 (acle_builtin_data): New. Change comment.
36823 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
36824 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
36825 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
36826 arm_acle_builtins.def.
36827 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
36828 (arm_init_acle_builtins): New.
36829 (CRC32_BUILTIN): Remove.
36830 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
36831 crc32cb, crc32ch and crc32cw.
36832 (arm_init_crc32_builtins): Remove.
36833 (arm_init_builtins): Use arm_init_acle_builtins rather
36834 than arm_init_crc32_builtins.
36835 (arm_expand_acle_builtin): New.
36836 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
36837 * config/arm/arm_acle_builtins.def: New.
36839 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36841 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
36842 (arm_builtin_datum): ... this.
36843 (arm_init_neon_builtin): Rename to ...
36844 (arm_init_builtin): ... this. Add a new parameters PREFIX
36845 and USE_SIG_IN_NAME.
36846 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
36847 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
36848 'arm_builtin_datum'.
36849 (arm_init_vfp_builtins): Likewise.
36850 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
36851 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
36852 (arm_expand_neon_args): Rename to ...
36853 (arm_expand_builtin_args): ... this. Rename builtin_arg
36854 enum values and differentiate between ARG_BUILTIN_MEMORY
36855 and ARG_BUILTIN_NEON_MEMORY.
36856 (arm_expand_neon_builtin_1): Rename to ...
36857 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
36858 values, arm_expand_builtin_args and add bool parameter NEON.
36859 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
36860 (arm_expand_vfp_builtin): Likewise.
36861 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
36863 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36865 PR middle-end/77484
36866 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
36867 * predict.c (tree_estimate_probability_bb): Reverse direction of
36868 polymorphic call predictor.
36870 2017-01-06 David Malcolm <dmalcolm@redhat.com>
36872 * passes.c (execute_one_pass): Split out pass-skipping logic into...
36873 (determine_pass_name_match): ...this new function and...
36874 (should_skip_pass_p): ...this new function.
36876 2017-01-06 Nathan Sidwell <nathan@acm.org>
36878 * ipa-visibility.c (function_and_variable_visibility): Reformat
36879 comments and long lines. Remove extrneous if.
36880 * symtab.c (symtab_node::make_decl_local): Fix code format.
36881 (symtab_node::set_section_for_node): Fix comment typo.
36883 2017-01-06 Martin Liska <mliska@suse.cz>
36886 * lra-constraints.c: Rename invariant to lra_invariant.
36887 * predict.c (set_even_probabilities): Initialize e to NULL.
36889 2017-01-05 Martin Sebor <msebor@redhat.com>
36891 PR tree-optimization/78910
36892 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
36893 (format_integer): Correct off-by-one error in the handling
36894 of precision with negative numbers in signed conversions..
36896 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
36898 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
36900 2017-01-05 Jakub Jelinek <jakub@redhat.com>
36902 PR tree-optimization/71016
36903 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
36904 factor_out_conditional_conversion. Formatting fix.
36905 (factor_out_conditional_conversion): Add cond_stmt argument.
36906 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
36907 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
36910 2017-01-05 David Malcolm <dmalcolm@redhat.com>
36912 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
36913 read-rtl-function.o, and selftest-rtl.o.
36914 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
36915 (selftest::aarch64_test_loading_full_dump): New function.
36916 (selftest::aarch64_run_selftests): New function.
36917 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
36918 selftest::aarch64_run_selftests.
36919 * config/i386/i386.c
36920 (selftest::ix86_test_loading_dump_fragment_1): New function.
36921 (selftest::ix86_test_loading_call_insn): New function.
36922 (selftest::ix86_test_loading_full_dump): New function.
36923 (selftest::ix86_test_loading_unspec): New function.
36924 (selftest::ix86_run_selftests): Call the new functions.
36925 * emit-rtl.c (maybe_set_max_label_num): New function.
36926 * emit-rtl.h (maybe_set_max_label_num): New decl.
36927 * function.c (instantiate_decls): Guard call to
36928 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
36929 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
36931 * gensupport.c (gen_reader::gen_reader): Pass "false"
36932 for new "compact" param of rtx_reader.
36933 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
36934 rather than an empty string for NULL strings.
36935 * read-md.c: Potentially include config.h rather than bconfig.h.
36936 Wrap include of errors.h with #ifdef GENERATOR_FILE.
36937 (have_error): New global, copied from errors.c.
36938 (md_reader::read_name): Rename to...
36939 (md_reader::read_name_1): ...this, adding "out_loc" param,
36940 and converting "missing name or number" to returning false, rather
36942 (md_reader::read_name): Reimplement in terms of read_name_1.
36943 (md_reader::read_name_or_nil): New function.
36944 (md_reader::read_string): Handle "(nil)" by returning NULL.
36945 (md_reader::md_reader): Add new param "compact".
36946 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
36947 (md_reader::read_file): New method.
36948 * read-md.h (md_reader::md_reader): Add new param "compact".
36949 (md_reader::read_file): New method.
36950 (md_reader::is_compact): New accessor.
36951 (md_reader::read_name): Convert return type from void to file_location.
36952 (md_reader::read_name_or_nil): New decl.
36953 (md_reader::read_name_1): New decl.
36954 (md_reader::m_compact): New field.
36955 (noop_reader::noop_reader): Pass "false" for new "compact" param
36957 (rtx_reader::rtx_reader): Add new "compact" param.
36958 (rtx_reader::read_rtx_operand): Make virtual and convert return
36959 type from void to rtx.
36960 (rtx_reader::read_until): New decl.
36961 (rtx_reader::handle_any_trailing_information): New virtual function.
36962 (rtx_reader::postprocess): New virtual function.
36963 (rtx_reader::finalize_string): New virtual function.
36964 (rtx_reader::m_in_call_function_usage): New field.
36965 (rtx_reader::m_reuse_rtx_by_id): New field.
36966 * read-rtl-function.c: New file.
36967 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
36968 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
36969 (selftest::verify_three_block_rtl_cfg): New decl.
36970 * read-rtl-function.h: New file.
36971 * read-rtl.c: Potentially include config.h rather than bconfig.h.
36972 For host, include function.h, memmodel.h, and emit-rtl.h.
36973 (one_time_initialization): New function.
36974 (struct compact_insn_name): New struct.
36975 (compact_insn_names): New array.
36976 (find_code): Handle insn codes in compact dumps.
36977 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
36978 (bind_subst_iter_and_attr): Likewise.
36979 (add_condition_to_string): Likewise.
36980 (add_condition_to_rtx): Likewise.
36981 (apply_attribute_uses): Likewise.
36982 (add_current_iterators): Likewise.
36983 (apply_iterators): Likewise.
36984 (initialize_iterators): Guard usage of apply_subst_iterator with
36985 #ifdef GENERATOR_FILE.
36986 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
36987 (md_reader::read_mapping): Likewise.
36988 (add_define_attr_for_define_subst): Likewise.
36989 (add_define_subst_attr): Likewise.
36990 (read_subst_mapping): Likewise.
36991 (check_code_iterator): Likewise.
36992 (rtx_reader::read_rtx): Likewise. Move one-time initialization
36994 (one_time_initialization): New function.
36995 (rtx_reader::read_until): New method.
36996 (read_flags): New function.
36997 (parse_reg_note_name): New function.
36998 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
36999 Handle reuse_rtx ids.
37000 Wrap iterator lookup within #ifdef GENERATOR_FILE.
37001 Add parsing support for RTL dumps, mirroring the special-cases in
37002 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
37003 values, and calling handle_any_trailing_information.
37004 (rtx_reader::read_rtx_operand): Convert return type from void
37005 to rtx, returning return_rtx. Handle case 'e'. Call
37006 finalize_string on XSTR and XTMPL fields.
37007 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
37008 "(nil)" values were omitted. Call the postprocess vfunc on the
37010 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
37011 class ctor. Initialize m_in_call_function_usage. Call
37012 one_time_initialization.
37013 * rtl-tests.c (selftest::test_uncond_jump): Call
37014 set_new_first_and_last_insn.
37015 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
37016 * selftest-rtl.c: New file.
37017 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
37018 (selftest::get_insn_by_uid): New decl.
37019 * selftest-run-tests.c (selftest::run_tests): Call
37020 read_rtl_function_c_tests.
37021 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
37022 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
37025 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
37027 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
37028 operands in a special way. Assert that pos+len <= mode precision.
37030 2017-01-05 Jakub Jelinek <jakub@redhat.com>
37032 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
37033 3 argument Alias with unlimited for the negative form.
37034 (fno-vect-cost-model): Removed.
37036 2017-01-05 Martin Liska <mliska@suse.cz>
37038 * hsa-gen.c (gen_hsa_divmod): New function.
37039 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
37041 2017-01-05 Martin Liska <mliska@suse.cz>
37044 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
37047 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37049 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
37050 small constant length operands.
37052 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37054 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
37055 between loop iterations.
37057 2017-01-05 Martin Liska <mliska@suse.cz>
37060 * gimplify.c (gimplify_decl_expr): Compare to
37061 asan_poisoned_variables instread of checking flags.
37062 (gimplify_target_expr): Likewise.
37063 (gimplify_expr): Likewise.
37064 (gimplify_function_tree): Conditionally initialize
37065 asan_poisoned_variables.
37067 2017-01-04 Jeff Law <law@redhat.com>
37069 PR tree-optimizatin/78812
37070 * rtl.h (contains_mem_rtx_p): Prototype.
37071 * ifcvt.c (containts_mem_rtx_p): Move from here to...
37072 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
37073 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
37074 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
37075 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
37077 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37079 * input.c (assert_char_at_range): Default-initialize actual_range.
37081 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37083 * df-scan.c (df_ref_create_structure): Make regno unsigned,
37084 to match the caller.
37086 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37088 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
37089 insns after final jump in test to emit dummy move.
37091 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37093 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
37094 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
37096 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37098 * multiple_target.c (create_dispatcher_calls): Init e_next.
37099 * tree-ssa-loop-split.c (split_loop): Init border.
37100 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
37103 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
37108 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
37109 (altivec_register_operand): Do not return true if the operand
37110 contains a SUBREG mixing SImode and SFmode.
37111 (vsx_register_operand): Likewise.
37112 (vsx_reg_sfsubreg_ok): New predicate.
37113 (vfloat_operand): Do not return true if the operand contains a
37114 SUBREG mixing SImode and SFmode.
37115 (vint_operand): Likewise.
37116 (vlogical_operand): Likewise.
37117 (gpc_reg_operand): Likewise.
37118 (int_reg_operand): Likewise.
37119 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
37120 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
37121 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
37123 (rs6000_emit_move_si_sf_subreg): New helper function.
37124 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
37125 fixup SUBREGs involving SImode and SFmode.
37126 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
37127 numbers for the new peephole2 optimization.
37128 (peephole2 for SFmode unions): New peephole2 to optimize cases in
37129 the GLIBC math library that do AND/IOR/XOR operations on single
37130 precision floating point.
37131 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
37132 target macros to say whether we need to avoid SUBREGs mixing
37134 (TARGET_ALLOW_SF_SUBREG): Likewise.
37135 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
37136 (UNSPEC_SI_FROM_SF): Likewise.
37137 (iorxor): Change spacing.
37138 (and_ior_xor): New iterator for AND, IOR, and XOR.
37139 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
37140 (movdi_from_sf_zero_ext): Likewise.
37141 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
37142 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
37143 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
37144 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
37145 (fms<mode>4): Likewise.
37146 (fnma<mode>4): Likewise.
37147 (fnms<mode>4): Likewise.
37148 (nfma<mode>4): Likewise.
37149 (nfms<mode>4): Likewise.
37151 2017-01-04 Marek Polacek <polacek@redhat.com>
37154 * doc/invoke.texi: Document -Wpointer-compare.
37156 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37158 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
37161 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
37162 descriptions for -gdwarf-5 and emit them as uleb128 instead of
37165 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
37168 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
37169 documentation of the powerpc_popcntb_ok attribute.
37170 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
37171 code to issue warning messages if a requested CPU configuration is
37172 not supported by the binary (assembler and loader) toolchain.
37173 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
37174 made to define a built-in function that has been disabled.
37175 (paired_init_builtins): Add assertion to prevent ICE if attempt is
37176 made to define a built-in function that has been disabled.
37177 (altivec_init_builtins): Add comment explaining why definition
37178 of the DST built-in functions is not preceded by an assertion
37179 check. Add assertions to prevent ICE if attempts are made to
37180 define an altivec predicate or an abs* built-in function that has
37182 (htm_init_builtins): Add comment explaining why definition of the
37183 htm built-in functions is not preceded by an assertion check.
37185 2017-01-04 Jeff Law <law@redhat.com>
37187 PR tree-optimizatin/67955
37188 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
37189 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
37190 the points-to solution does not include pt_null. Use DECL_PT_UID
37193 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
37195 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
37196 Use gen_int_mode instead of gen_lopwart for const_int operands.
37198 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37200 PR tree-optimization/71563
37201 * match.pd: Simplify X << Y into X if Y is known to be 0 or
37202 out of range value - has low bits known to be zero.
37204 2017-01-04 Alan Modra <amodra@gmail.com>
37206 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
37207 * configure: Regenerate.
37208 * config.in: Regenerate.
37210 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37213 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
37214 a substring of the message, but strcmp with the whole message. Ifdef
37215 ENABLE_NLS, translate the message first using dgettext.
37217 2017-01-03 Jeff Law <law@redhat.com>
37219 PR tree-optimizatin/78856
37220 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
37221 (mark_threaded_blocks): Remove code to truncate thread paths that
37222 cross multiple loop headers. Instead invalidate the cached loop
37223 iteration information and handle case of a thread path walking
37224 into an irreducible region.
37226 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
37229 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
37230 assertions. Add support for doing the signbit if the IEEE 128-bit
37231 floating point value is in a GPR.
37232 * config/rs6000/rs6000.md (Fsignbit): Delete.
37233 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
37234 Update the length attribute if the value is in a GPR.
37235 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
37236 the sign or zero extension instruction, since the value is always 0/1.
37237 (signbit<mode>2_dm2): Delete using <Fsignbit>.
37240 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
37241 extracting SImode to a GPR register so that we can generate a
37242 store, limit the vector to be in a traditional Altivec register
37243 for the vextuwrx instruction.
37245 2017-01-03 Ian Lance Taylor <iant@google.com>
37247 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
37249 2017-01-03 Martin Sebor <msebor@redhat.com>
37251 PR tree-optimization/78696
37252 * gimple-ssa-sprintf.c (format_floating): Correct handling of
37253 precision. Use MPFR for %f for greater fidelity. Correct handling
37255 (pass_sprintf_length::compute_format_length): Set width and precision
37256 specified by asrerisk to void_node for vararg functions.
37257 (try_substitute_return_value): Adjust dump output.
37259 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
37261 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
37263 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
37265 * doc/invoke.texi (SPARC options): Document -mlra as the default.
37266 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
37267 -mlra/-mno-lra was passed to the compiler.
37269 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
37271 PR rtl-optimization/65618
37272 * emit-rtl.c (try_split): Move initialization of "before" and
37273 "after" to just before the call to emit_insn_after_setloc.
37275 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
37277 * doc/md.texi (Standard Names): Remove reference to Java frontend.
37279 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
37281 * dwarf2out.c (gen_enumeration_type_die): When
37282 -gno-strict-dwarf, add a DW_AT_encoding attribute.
37284 2017-01-03 Jakub Jelinek <jakub@redhat.com>
37286 PR tree-optimization/78965
37287 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
37288 Change first argument from const call_info & to call_info &. For %n
37289 set info.nowrite to false.
37291 PR middle-end/78901
37292 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
37293 possibly throwing calls.
37295 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
37296 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
37297 and fns handling, rather than in a separate case SSA_NAME.
37299 2017-01-02 Jeff Law <law@redhat.com>
37301 * config/darwin-driver.c (darwin_driver_init): Const-correctness
37302 fixes for first_period and second_period variables.
37304 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
37307 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
37308 (*insvqi_1): New insn pattern.
37309 (*insvqi_1_mem_rex64): Ditto.
37310 (*insvqi_2): Ditto.
37311 (*insvqi_3): Rename from *insvqi.
37313 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
37315 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
37317 * doc/cfg.texi (Edges): Remove reference to Java.
37318 (Maintaining the CFG): Ditto.
37320 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37322 PR middle-end/77674
37323 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
37324 transparent aliases.
37326 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37328 PR middle-end/77484
37329 * predict.def (PRED_CALL): Update hitrate.
37330 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
37331 * predict.c (tree_estimate_probability_bb): Split CALL predictor
37332 into direct/indirect/polymorphic variants.
37334 2017-01-01 Jakub Jelinek <jakub@redhat.com>
37336 Update copyright years.
37338 * gcc.c (process_command): Update copyright notice dates.
37339 * gcov-dump.c (print_version): Ditto.
37340 * gcov.c (print_version): Ditto.
37341 * gcov-tool.c (print_version): Ditto.
37342 * gengtype.c (create_file): Ditto.
37343 * doc/cpp.texi: Bump @copying's copyright year.
37344 * doc/cppinternals.texi: Ditto.
37345 * doc/gcc.texi: Ditto.
37346 * doc/gccint.texi: Ditto.
37347 * doc/gcov.texi: Ditto.
37348 * doc/install.texi: Ditto.
37349 * doc/invoke.texi: Ditto.
37351 Copyright (C) 2017 Free Software Foundation, Inc.
37353 Copying and distribution of this file, with or without modification,
37354 are permitted in any medium without royalty provided the copyright
37355 notice and this notice are preserved.