1 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
4 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
5 whether we can allocate pseudos before trying to fix an address.
6 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
7 memory address is indexed or indirect.
8 (floatuns_<mode>si2_hw2): Likewise.
10 2017-12-01 Jason Merrill <jason@redhat.com>
12 * Makefile.in (TAGS): Add c-family/*.cc.
14 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
16 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
17 (call_value_insn): Likewise.
18 (sibcall_insn): Likewise.
19 (sibcall_value_insn): Likewise.
20 (movsi_aarch64): Likewise.
21 (movdi_aarch64): Likewise.
22 (add_losym_): Likewise.
23 (ldr_got_small_): Likewise.
24 (ldr_got_small_sidi): Likewise.
25 (ldr_got_small_28k_): Likewise.
26 (ldr_got_small_28k_sidi): Likewise.
27 * config/aarch64/aarch64.c (aarch64_print_address_internal):
28 Move output_addr_const to symbolic case. Add error check.
30 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32 * config/s390/predicates.md (plus16_Q_operand): New predicate.
33 * config/s390/s390.md: Disable MVC merging peephole if it would
34 disable operand forwarding.
35 (new peephole2): Split MVCs if it would turn them into up to 2
38 2017-12-01 Richard Biener <rguenther@suse.de>
40 PR tree-optimization/83232
41 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
42 detection of same access. Instead of breaking the group here
43 do not consider the duplicate. Add comment explaining real fix.
45 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
47 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
49 2017-12-01 Sudakshina Das <sudi.das@arm.com>
51 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
53 2017-12-01 Jakub Jelinek <jakub@redhat.com>
55 * function.h (struct function): Remove cilk_frame_decl,
56 is_cilk_function and calls_cilk_spawn fields.
57 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
59 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
61 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
62 * target.def: Adjust comment.
63 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
64 Don't test cilk_elemental.
66 PR tree-optimization/83233
67 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
68 bswap_stat name for the struct.
71 * tree.h (SWITCH_BREAK_LABEL_P): Define.
72 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
73 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
74 SWITCH_BREAK_LABEL_P set on the label.
75 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
76 added for default case if it was missing and not all cases covered.
77 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
78 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
80 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
81 to avoid -Wimplicit-fallthrough warning.
82 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
83 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
85 PR tree-optimization/83221
86 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
88 (init_reassoc): Formatting fix.
91 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
92 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
96 * tree-cfg.c: Include asan.h.
97 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
99 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
101 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
103 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
105 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
106 vpcmpeqd instruction.
108 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
110 * config/i386/i386.c (standard_sse_constant_opcode): Fix
111 registers type for 128bit mode.
113 2017-11-30 Jakub Jelinek <jakub@redhat.com>
115 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
117 * gengtype-state.c (read_state_pair): Likewise.
118 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
119 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
120 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
121 * ifcvt.c (noce_try_store_flag_constants): Likewise.
122 * tree-ssa-ccp.c (ccp_finalize): Likewise.
123 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
124 * builtins.c (fold_builtin_3): Likewise.
125 * graphite-scop-detection.c
126 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
127 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
130 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
133 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
134 parameter. Return true if ignore_decl_rtl_set_p is true and other
135 conditions are satisfied.
136 * asan.h (asan_protect_global): Add new parameter.
137 * varasm.c (categorize_decl_for_section): Pass true as second parameter
138 to asan_protect_global calls.
140 2017-11-30 Jim Wilson <jimw@sifive.com>
142 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
143 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
144 -mstrict-align, add info on default value. Delete redundant lines for
145 -mabi. Add missing -mexplicit-relocs docs.
147 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
149 * config/arc/arc.md (trap): New pattern.
151 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
153 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
154 instruction to end into a delay slot.
155 * config/arc/arc.md (cond_delay_insn): Check if the instruction
156 can be placed into a delay slot against reg_note.
158 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
160 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
161 labels number of usages.
163 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
165 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
167 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
169 2017-11-30 Jakub Jelinek <jakub@redhat.com>
172 * internal-fn.c (expand_mul_overflow): Optimize unsigned
173 multiplication by power of 2 constant into two shifts + comparison.
175 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
178 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
180 2017-11-30 Richard Biener <rguenther@suse.de>
182 PR tree-optimization/83202
183 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
184 allow_peel argument and guard peeling.
185 (canonicalize_loop_induction_variables): Likewise.
186 (canonicalize_induction_variables): Pass false.
187 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
188 peeling from cunrolli.
190 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
192 * combine.c (try_combine): Print a message to dump file whenever
193 I0, I1, or I2 cannot be combined into I3.
195 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
197 PR rtl-optimization/83156
198 PR rtl-optimization/82621
199 * combine.c (try_combine): Don't split an I2 if one of the dests is
200 set again before I3. Allow unused dests.
202 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
204 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
206 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
208 PR rtl-optimization/80818
209 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
210 recursively. Use insn code for clobber.
211 (lra_set_insn_recog_data): Pass the new arg to
212 collect_non_operand_hard_regs.
213 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
215 (lra_update_insn_regno_info): Pass insn to
216 add_regs_to_insn_regno_info.
218 2017-11-29 Jim Wilson <jimw@sifive.com>
219 Andrew Waterman <andrew@sifive.com>
221 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
222 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
223 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
224 use SINGLE_SHIFT_COST.
225 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
226 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
228 2017-11-29 Julia Koval <julia.koval@intel.com>
230 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
231 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
232 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
233 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
234 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
235 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
236 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
237 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
238 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
239 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
240 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
241 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
242 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
243 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
244 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
245 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
246 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
247 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
248 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
249 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
250 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
251 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
252 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
253 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
254 __builtin_ia32_vpshldv_v2di_mask,
255 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
256 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
257 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
259 2017-11-29 Julia Koval <julia.koval@intel.com>
261 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
262 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
263 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
264 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
265 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
266 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
267 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
268 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
269 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
270 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
271 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
272 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
273 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
274 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
275 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
276 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
277 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
278 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
279 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
280 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
281 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
283 2017-11-29 Daniel Cederman <cederman@gaisler.com>
285 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
286 movsi_pic_gotdata_op instruction as a load for the UT699 errata
289 2017-11-29 Martin Aberg <maberg@gaisler.com>
291 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
292 to prevent b2bst errata sequence.
293 (sqrtdf2_fix): Likewise.
295 2017-11-29 Daniel Cederman <cederman@gaisler.com>
297 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
298 (div_sqrt_insn_p): New function.
299 (sparc_do_work_around_errata): Insert NOP instructions to
300 prevent sequences that could trigger the TN-0013 errata for
301 certain LEON3 processors.
302 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
303 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
304 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
305 (in_branch_delay): Prevent div and sqrt in delay slot if
307 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
309 2017-11-29 Daniel Cederman <cederman@gaisler.com>
311 * config/sparc/sparc.c (atomic_insn_p): New function.
312 (sparc_do_work_around_errata): Insert NOP instructions to
313 prevent sequences that could trigger the TN-0010 errata for
315 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
316 instruction referable in atomic_insns_p.
318 2017-11-29 Daniel Cederman <cederman@gaisler.com>
320 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
321 (atomic_compare_and_swap_leon3_1): Likewise.
324 2017-11-29 Daniel Cederman <cederman@gaisler.com>
326 * config/sparc/sparc.c (fpop_insn_p): New function.
327 (sparc_do_work_around_errata): Insert NOP instructions to
328 prevent sequences that could trigger the TN-0012 errata for
330 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
331 * config/sparc/sparc.md (fix_gr712rc): New attribute.
332 (in_branch_annul_delay): Prevent floating-point instructions
333 in delay slot of annulled integer branch.
335 2017-11-29 Richard Biener <rguenther@suse.de>
337 PR tree-optimization/83202
338 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
340 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
341 nodes vectorized to the same stmts multiple times.
342 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
343 (vect_analyze_slp_instance): Adjust.
344 (scalar_stmts_to_slp_tree_map_t): New typedef.
345 (vect_schedule_slp_instance): Add a map recording the SLP node
346 representing the vectorized stmts for a set of scalar stmts.
347 Avoid code-generating redundancies.
348 (vect_schedule_slp): Allocate map and pass it down.
350 2017-11-29 Nathan Sidwell <nathan@acm.org>
353 * tree.c (build_complex_type): Fix canonicalization. Only fill in
356 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
358 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
359 cases for printing LDP/STP memory addresses.
360 (aarch64_print_address_internal): Renamed from
361 aarch64_print_operand_address, added parameter, add Pmode check.
362 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
363 (aarch64_print_operand_address): Indirect to
364 aarch64_print_address_internal.
365 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
366 'y' operand output specifier.
368 2017-11-29 Jakub Jelinek <jakub@redhat.com>
371 * tree.c (build_simple_mem_ref_loc): Handle
372 get_addr_base_and_unit_offset returning a MEM_REF.
375 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
378 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
381 2017-11-28 David Malcolm <dmalcolm@redhat.com>
383 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
385 (layout::move_to_column): Likewise.
387 2017-11-28 Jakub Jelinek <jakub@redhat.com>
390 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
391 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
393 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
394 Martin Jambor <mjambor@suse.cz>
397 * tree.h (expr_type_first_operand_type_p): Declare
398 * tree.c (expr_type_first_operand_type_p): New function.
399 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
400 (ipa_value_from_jfunc): Adjust declaration.
401 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
402 Use it as result type for arithmetics, unless it is NULL in which case
403 be more conservative.
404 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
405 ipa_get_jf_pass_through_result.
406 (propagate_vals_across_pass_through): Likewise.
407 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
408 is to propagate_vals_across_pass_through.
409 (propagate_constants_across_call): Pass PARM_TYPE to
410 propagate_scalar_across_jump_function.
411 (find_more_scalar_values_for_callers_subset): Pass parameter type to
412 ipa_value_from_jfunc.
413 (cgraph_edge_brings_all_scalars_for_node): Likewise.
414 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
415 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
416 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
417 target_type, pass it to ipa_value_from_jfunc.
418 (update_indirect_edges_after_inlining): Pass parameter type to
419 try_make_edge_direct_simple_call.
421 2017-11-28 Jeff Law <law@redhat.com>
423 * gimple-ssa-evrp-analyze.c
424 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
425 refine ranges if scev_initialized_p returns true.
426 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
428 2017-11-28 Julia Koval <julia.koval@intel.com>
430 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
431 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
432 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
433 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
434 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
435 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
436 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
437 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
438 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
439 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
440 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
441 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
442 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
443 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
444 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
445 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
446 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
447 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
448 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
449 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
450 __builtin_ia32_vpshrd_v2di_mask): New builtins.
451 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
453 2017-11-28 Julia Koval <julia.koval@intel.com>
455 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
456 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
457 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
458 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
459 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
460 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
461 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
462 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
463 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
464 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
465 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
466 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
467 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
468 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
469 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
470 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
471 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
472 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
473 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
474 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
475 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
476 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
477 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
478 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
479 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
480 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
481 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
482 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
483 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
484 __builtin_ia32_vpshld_v2di_mask): New builtins.
485 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
486 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
488 2017-11-28 Richard Biener <rguenther@suse.de>
490 PR tree-optimization/80776
491 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
493 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
495 (evrp_range_analyzer::record_ranges_from_incoming_edges):
496 If the incoming edge is an effective fallthru because the other
497 edge only reaches a __builtin_unreachable () then record ranges
498 derived from the controlling condition in SSA info.
499 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
500 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
502 2017-11-28 Olivier Hainque <hainque@adacore.com>
504 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
505 on mingw build hosts.
507 2017-11-28 Jakub Jelinek <jakub@redhat.com>
510 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
512 2017-11-28 Richard Biener <rguenther@suse.de>
515 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
516 copies generated from memcpy use a character array as reference
519 2017-11-28 Julia Koval <julia.koval@intel.com>
520 Sebastian Peryt <sebastian.peryt@intel.com>
522 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
523 c-family/c-cilkplus.o, c-family/array-notation-common.o,
524 cilk-common.o, cilk.h, cilk-common.c): Remove.
526 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
527 * builtins.c (is_builtin_name): Remove cilkplus condition.
528 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
529 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
530 cilk-builtins.def, cilkplus.def): Remove.
531 * cif-code.def (CILK_SPAWN): Remove.
532 * cilk-builtins.def: Delete.
533 * cilk-common.c: Ditto.
535 * cilkplus.def: Ditto.
536 * config/darwin.h (fcilkplus): Delete.
537 * cppbuiltin.c: Ditto.
538 * doc/extend.texi: Remove cilkplus doc.
539 * doc/generic.texi: Ditto.
540 * doc/invoke.texi: Ditto.
541 * doc/passes.texi: Ditto.
542 * gcc.c (fcilkplus): Remove.
543 * gengtype.c (cilk.h): Remove.
544 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
546 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
548 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
550 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
551 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
553 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
554 inline_read_section): Ditto.
555 * ipa-inline-analysis.c (cilk.h): Remove.
556 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
557 * lto-wrapper.c (merge_and_complain, append_compiler_options,
558 append_linker_options): Remove condition for fcilkplus.
559 * lto/lto-lang.c (cilk.h): Remove.
560 (lto_init): Remove condition for fcilkplus.
561 * omp-expand.c (expand_cilk_for_call): Delete.
562 (expand_omp_taskreg, expand_omp_for_static_chunk,
563 expand_omp_for): Remove cilkplus
565 (expand_cilk_for): Delete.
566 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
567 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
568 execute_lower_omp, diagnose_sb_0): Ditto.
569 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
570 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
571 * tree-nested.c: Ditto.
572 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
573 (dump_generic_node): Ditto.
574 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
575 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
577 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
579 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
581 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
582 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
583 for vector mode and !TARGET_SIMD.
585 2017-11-28 Jakub Jelinek <jakub@redhat.com>
587 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
589 * tree.h (SWITCH_LABELS): Remove.
590 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
591 assert SWITCH_BODY is non-NULL.
592 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
594 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
596 PR tree-optimization/80788
597 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
598 has TREE_OVERFLOW set, call drop_tree_overflow.
600 2017-11-28 Richard Biener <rguenther@suse.de>
602 PR tree-optimization/83158
603 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more
606 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
609 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
610 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
612 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
614 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
615 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
617 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
620 * varasm.c (compare_constant): Take the mode of the constants into
621 account when comparing floating point constants.
623 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
625 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
626 from explicit instantiation of debug_helper.
627 * vec.h (DEFINE_DEBUG_VEC): Ditto.
629 2017-11-27 Richard Biener <rguenther@suse.de>
631 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
634 2017-11-27 Richard Biener <rguenther@suse.de>
636 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
637 (type_hash_canon): Also clear int_cst_hash_table entry for
639 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
641 2017-11-27 Tamar Christina <tamar.christina@arm.com>
643 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
644 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
645 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
647 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
649 * hash-map.h (gt_cleare_cache): Avoid UB.
651 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
653 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
654 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
655 (decide_unroll_runtime_iterations): Likewise.
656 (decide_unroll_stupid): Likewise.
658 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
661 * config/i386/i386.md: Add a loop with incssp.
663 2017-11-27 Martin Jambor <mjambor@suse.cz>
665 PR tree-optimization/81248
666 * tree-sra.c (splice_param_accesses): Remove size check.
667 (decide_one_param_reduction): Fix size check.
668 * gimple-pretty-print.c (dump_profile): Silence warning.
669 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
671 2017-11-27 Jakub Jelinek <jakub@redhat.com>
674 * dbxout.c (lastlineno): New variable.
675 (dbx_debug_hooks): Use dbxout_switch_text_section as
676 switch_text_section debug hook.
677 (dbxout_function_end): Switch to current_function_section
678 rather than function_section. If crtl->has_bb_partition,
679 output just one N_FUN, depending on in_cold_section_p.
680 (dbxout_source_line): Remember last lineno in lastlineno.
681 (dbxout_switch_text_section): New function.
682 (dbxout_function_decl): Adjust dbxout_block caller.
683 (dbx_block_with_cold_children): New function.
684 (dbxout_block): Return true if any LBRAC/RBRAC have been
685 emitted. Use dbx_block_with_cold_children at depth == 0
686 in second partition. Add PARENT_BLOCKNUM argument, pass
687 it optionally adjusted to children. Output LBRAC/RBRAC
688 around recursive call only if the block is in the current
689 partition, if not and anything was output, emit empty
691 * final.c (final_scan_insn): Compute cold_function_name
692 before calling switch_text_section debug hook. Call
693 that hook even if dwarf2out_do_frame if not emitting
697 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
700 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
702 PR rtl-optimization/82488
703 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
705 2017-11-26 Julia Koval <julia.koval@intel.com>
707 * config/i386/i386.c (processor_target_table): Add skylake_cost for
709 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
712 2017-11-26 Julia Koval <julia.koval@intel.com>
714 * config/i386/driver-i386.c (host_detect_local_cpu):
715 Detect skylake-avx512.
717 2017-11-26 Julia Koval <julia.koval@intel.com>
719 * config.gcc: Add -march=cannonlake.
720 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
721 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
722 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
723 (PTA_CANNONLAKE): New.
724 (processor_target_table): Add cannonlake.
725 (ix86_option_override_internal): Ditto.
726 (fold_builtin_cpu): Ditto.
727 (get_builtin_code_for_version): Handle cannonlake.
728 (M_INTEL_COREI7_CANNONLAKE): New.
729 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
730 * doc/invoke.texi: Add -march=cannonlake.
732 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
734 * plugin.c (add_new_plugin): Use platform-specific library extensions.
735 (try_init_one_plugin): Alternative implementation for MinGW.
736 * Makefile.in (plugin_implib): New.
737 (gengtype-lex.c): Fix broken AIX workaround.
738 * configure: Regenerate.
739 * doc/plugins.texi: Document support for MinGW.
741 2017-11-25 Jakub Jelinek <jakub@redhat.com>
743 PR rtl-optimization/81553
744 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
745 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
746 is a shift where C1 has different mode than the whole shift, use C1's
747 mode for MULT rather than the shift's mode.
750 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
751 builtins not enabled in the currently selected ISA.
753 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
755 PR tree-optimization/71026
756 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
757 (insert_reciprocals): Change to insert reciprocals before a division
758 by a square and to insert the square of a reciprocal.
759 (execute_cse_reciprocals_1): Change to consider division by a square.
760 (register_division_in): Add importance parameter.
762 2017-11-24 Richard Biener <rguenther@suse.de>
764 PR tree-optimization/82402
765 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
766 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
768 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
770 * match.pd (0-ptr): New transformation.
772 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
775 * ipa-inline.c (inline_small_functions): Set current badnes correctly
776 when skipping checking.
778 2017-11-24 Richard Biener <rguenther@suse.de>
780 PR tree-optimization/83128
781 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
782 (vn_reference_lookup_3): Likewise.
784 2017-11-24 Jakub Jelinek <jakub@redhat.com>
787 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
788 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
791 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
792 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
793 check for the latter with gimple_call_builtin_p. Do not handle
794 BUILT_IN_STPNCPY_CHK which is not a pass through call.
796 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
798 * config/arm/arm_neon.h: Fix pragma GCC push_options before
801 2017-11-23 Julia Koval <julia.koval@intel.com>
803 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
804 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
805 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
806 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
807 _mm512_maskz_expandloadu_epi16): New intrinsics.
808 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
809 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
810 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
811 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
812 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
813 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
814 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
815 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
816 _mm256_maskz_expandloadu_epi8): New intrinsics.
817 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
818 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
819 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
820 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
821 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
822 * config/i386/sse.md (VI248_VLBW): New iterator.
823 (expand<mode>_mask, expand<mode>_maskz): New patterns.
825 2017-11-23 Julia Koval <julia.koval@intel.com>
827 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
828 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
829 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
830 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
831 _mm512_mask_compressstoreu_epi16): New.
832 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
833 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
834 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
835 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
836 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
837 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
838 _mm256_mask_compressstoreu_epi8): New.
839 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
840 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
841 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
842 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
843 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
844 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
845 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
846 __builtin_ia32_compresshi128_mask,
847 __builtin_ia32_compressstoreuqi512_mask,
848 __builtin_ia32_compressstoreuhi512_mask,
849 __builtin_ia32_compressstoreuqi256_mask,
850 __builtin_ia32_compressstoreuqi128_mask,
851 __builtin_ia32_compressstoreuhi256_mask,
852 __builtin_ia32_compressstoreuhi128_mask): New builtins.
853 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
855 (ix86_expand_special_args_builtin): Handle new types.
856 (s4fma_expand): Handle new builtin array.
857 * config/i386/immintrin.h: Include new headers.
858 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
859 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
861 2017-11-23 Jakub Jelinek <jakub@redhat.com>
863 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
864 fixes. Declare temp and g variables at the top in order to avoid
865 {} in most of the cases.
867 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
869 * match.pd (ptr-0): New transformation.
871 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
873 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
874 (struct addr_mode_cost_table): New.
875 (struct tune_params): Add field addr_mode_costs.
876 * config/arm/arm.c (generic_addr_mode_costs): New.
877 (arm_slowmul_tune): Initialise addr_mode_costs field.
878 (arm_fastmul_tune): Likewise.
879 (arm_strongarm_tune): Likewise.
880 (arm_xscale_tune): Likewise.
881 (arm_9e_tune): Likewise.
882 (arm_marvell_pj4_tune): Likewise.
883 (arm_v6t2_tune): Likewise.
884 (arm_cortex_tune): Likewise.
885 (arm_cortex_a8_tune): Likewise.
886 (arm_cortex_a7_tune): Likewise.
887 (arm_cortex_a15_tune): Likewise.
888 (arm_cortex_a35_tune): Likewise.
889 (arm_cortex_a53_tune): Likewise.
890 (arm_cortex_a57_tune): Likewise.
891 (arm_exynosm1_tune): Likewise.
892 (arm_xgene1_tune): Likewise.
893 (arm_cortex_a5_tune): Likewise.
894 (arm_cortex_a9_tune): Likewise.
895 (arm_cortex_a12_tune): Likewise.
896 (arm_cortex_a73_tune): Likewise.
897 (arm_v7m_tune): Likewise.
898 (arm_cortex_m7_tune): Likewise.
899 (arm_v6m_tune): Likewise.
900 (arm_fa726te_tune): Likewise.
901 (arm_mem_costs): Use table lookup to calculate cost of addressing
904 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
906 * config/arm/arm.c (arm_mem_costs): New function.
907 (arm_rtx_costs_internal): Use arm_mem_costs.
909 2017-11-23 Mark Wielaard <mark@klomp.org>
911 * dwarf2out.c (init_sections_and_labels): Use generation to create
912 unique ranges_section_label and ranges_base_label. Return generation.
913 (output_rnglists): Add generation argument. Use generation to create
914 unique ranges labels.
915 (dwarf2out_finish): Get generation from init_sections_and_labels
916 and pass generation to output_rnglists.
918 2017-11-23 Mike Stump <mikestump@comcast.net>
919 Eric Botcazou <ebotcazou@adacore.com>
921 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
922 * cfgloop.h (struct loop): Add unroll field.
923 * function.h (struct function): Add has_unroll bitfield.
924 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
925 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
926 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
927 (pass_rtl_unroll_loops::gate): Likewise.
928 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
929 for which loop->unroll==1.
930 (decide_unroll_constant_iterations): Use note for consistency and
931 take loop->unroll into account. Return early if loop->unroll is set.
932 Fix thinko in existing test.
933 (decide_unroll_runtime_iterations): Use note for consistency and
934 take loop->unroll into account.
935 (decide_unroll_stupid): Likewise.
936 * lto-streamer-in.c (input_cfg): Read loop->unroll.
937 * lto-streamer-out.c (output_cfg): Write loop->unroll.
938 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
940 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
941 (print_loop): Print loop->unroll if set.
942 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
943 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
944 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
946 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
947 loop->unroll is set and smaller than the trip count. Otherwise bypass
948 entirely the heuristics if loop->unroll is set. Remove dead note.
949 Fix off-by-one bug in other note.
950 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
951 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
953 (tree_unroll_loops_completely): Make static.
954 (pass_complete_unroll::execute): Use correct type for variable.
955 (pass_complete_unrolli::execute): Fix formatting.
956 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
958 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
960 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
961 enable when TARGET_PREFER_AVX128 is set.
963 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
965 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
967 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
968 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
970 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
972 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
973 frequencies. Fix estimation of aggregate parameters.
975 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
977 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
980 2017-11-23 Tom de Vries <tom@codesourcery.com>
982 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
984 2017-11-23 Tom de Vries <tom@codesourcery.com>
986 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
989 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
992 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
993 sibcall_value_pcrel_fdpic): Use local variable instead of
995 (calli_tbr_rel): Add missing operand 2.
996 (call_valuei_tbr_rel): Add missing operand 3.
998 2017-11-23 Jakub Jelinek <jakub@redhat.com>
1001 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
1002 bitpos/bitsize covering the whole destination, use store_expr only if
1003 the complex mode is the same. Otherwise, use expand_normal and if
1004 it returns CONCAT, subreg each part separately instead of trying to
1005 subreg the whole result.
1007 2017-11-23 Richard Biener <rguenther@suse.de>
1009 PR tree-optimization/23094
1010 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
1011 come here from walking over backedges in the first iteration.
1012 (vn_reference_lookup_3): Skip clobbers that store the same value.
1014 2017-11-23 Richard Biener <rguenther@suse.de>
1016 PR tree-optimization/81403
1017 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
1018 a block we need a leader relative to.
1019 (phi_translate_1): For nary processing require a leader from
1020 get_representative_for given we run expression simplification
1021 using match-and-simplify. Remove previous fix.
1023 2017-11-22 Jeff Law <law@redhat.com>
1025 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
1026 Use new method allocate_value_range rather than accessing the
1027 vrp_value_range_pool data member directly.
1028 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
1029 to use extract_range_from_stmt method to avoid need for
1030 extract_range_from_assignment method.
1031 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
1032 method rather than setting values_propgated data member directly.
1033 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
1034 and values propagated data members and extract_range_from_assignment
1035 method. Reorder private data members to conform to standards.
1036 Add new methods set_lattice_propagation_complete and
1037 allocate_value_range.
1039 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
1041 PR rtl-optimization/83030
1042 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
1043 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
1044 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
1045 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
1046 CROSSING_JUMP_P flag is already set before setting it.
1047 * cfgrtl.c (fixup_partition_crossing): Likewise.
1048 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
1051 2017-11-22 Jakub Jelinek <jakub@redhat.com>
1053 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
1054 Handle the case where both arguments are using gen_const_vec_series.
1056 2017-11-22 David Malcolm <dmalcolm@redhat.com>
1059 * pretty-print.c (pp_format): Move quoting implementation to
1060 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
1061 to pass address of "quote" local.
1062 (pp_begin_quote): New function.
1063 (pp_end_quote): New function.
1064 * pretty-print.h (printer_fn): Convert penultimate param from bool
1066 (pp_begin_quote): New decl.
1067 (pp_end_quote): New decl.
1068 * tree-diagnostic.c (default_tree_printer): Convert penultimate
1069 param from bool to bool *.
1070 * tree-diagnostic.h (default_tree_printer): Likewise.
1072 2017-11-22 Jeff Law <law@redhat.com>
1074 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
1075 Initialize vr_values.
1076 (evrp_range_analyzer::try_find_new_range): Call methods attached to
1077 vr_values via vr_values class instance rather than delegators.
1078 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
1079 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
1080 (evrp_range_analyzer::push_value_range): Likewise.
1081 (evrp_range_analyzer::pop_value_range): Likewise.
1082 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
1083 most delegators. Those remaining are exposed as public interfaces.
1084 Make vr_values a pointer and private.
1085 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
1087 (evrp_range_analyzer::get_vr_value): New method.
1088 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
1089 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
1090 (class evrp_dom_walker): Attach evrp_folder class, initialize
1091 it in the ctor. Remove temporary delegators.
1092 (evrp_dom_walker::before_dom_children): Call methods in attached
1093 evrp_range_analyzer class via class instance pointer. Use
1094 free value_range_constant_singleton to remove need for
1095 op_with_constant_singleton_value delegator method. Do not
1096 create a vrp_prop class instance for every call! Narrow
1097 scope of a couple variables.
1098 (evrp_dom_walker::cleanup): Call methods in attached
1099 evrp_range_analyzer class via class instance pointer.
1100 * vr-values.h (class vr_values): Privatize many methods and
1103 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1105 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
1106 * cfgexpand.c (expand_debug_expr): Remove handling for them.
1107 * expr.c (expand_expr_real_2): Likewise.
1108 * fold-const.c (const_unop): Likewise.
1109 * optabs-tree.c (optab_for_tree_code): Likewise.
1110 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
1111 * tree-inline.c (estimate_operator_cost): Likewise.
1112 * tree-pretty-print.c (dump_generic_node): Likewise.
1113 (op_code_prio): Likewise.
1114 (op_symbol_code): Likewise.
1115 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
1116 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
1117 * internal-fn.c (direct_internal_fn_optab): New function.
1118 (direct_internal_fn_array, direct_internal_fn_supported_p
1119 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
1120 * fold-const-call.c (fold_const_reduction): New function.
1121 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
1123 * tree-vect-loop.c: Include internal-fn.h.
1124 (reduction_code_for_scalar_code): Rename to...
1125 (reduction_fn_for_scalar_code): ...this and return an internal
1127 (vect_model_reduction_cost): Take an internal_fn rather than
1129 (vect_create_epilog_for_reduction): Likewise. Build calls rather
1131 (vectorizable_reduction): Use internal functions rather than tree
1132 codes for the reduction operation. Update calls to the functions
1134 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
1135 Use calls to internal functions rather than REDUC tree codes.
1136 * config/aarch64/aarch64-simd.md: Update comment accordingly.
1138 2017-11-22 Olivier Hainque <hainque@adacore.com>
1140 * config/vxworks.c (vxworks_override_options): Pick default
1141 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
1142 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
1143 DWARF_GNAT_ENCODINGS_DEFAULT.
1144 * config/vxworksae.h: Likewise.
1146 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
1148 PR tree-optimization/83104
1149 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
1152 2017-11-22 Marek Polacek <polacek@redhat.com>
1153 H.J. Lu <hongjiu.lu@intel.com>
1154 Jason Merrill <jason@redhat.com>
1159 * calls.c (initialize_argument_information): Call
1160 warn_parameter_passing_abi target hook.
1161 (store_one_arg): Use 0 for empty record size. Don't push 0 size
1162 argument onto stack.
1163 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
1164 * common.opt: Update -fabi-version description.
1165 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
1166 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
1168 (ix86_is_empty_record): New function.
1169 (ix86_warn_parameter_passing_abi): New function.
1170 (TARGET_EMPTY_RECORD_P): Redefine.
1171 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
1172 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
1173 * doc/tm.texi: Regenerated.
1174 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
1175 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
1176 * dwarf2out.c (get_ultimate_context): Move to tree.c.
1177 * explow.c (hard_function_value): Call arg_int_size_in_bytes
1178 instead of int_size_in_bytes.
1179 * expr.c (copy_blkmode_to_reg): Likewise.
1180 * function.c (aggregate_value_p): Return 0 for empty types.
1181 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi target hook.
1182 (locate_and_pad_parm): Call arg size_in_bytes instead
1184 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
1185 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
1186 * target.def (empty_record_p, warn_parameter_passing_abi): New target
1188 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
1189 (std_gimplify_va_arg_expr): Skip empty records. Call
1190 arg_size_in_bytes instead size_in_bytes.
1191 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
1192 * tree-core.h (tree_type_common): Add empty_flag.
1193 (tree_decl_common): Update comments.
1194 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
1196 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
1197 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
1199 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
1200 * tree.c (default_is_empty_type): New function.
1201 (default_is_empty_record): New function.
1202 (arg_int_size_in_bytes): New function.
1203 (arg_size_in_bytes): New function.
1204 (get_ultimate_context): New function.
1205 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
1206 TRANSLATION_UNIT_WARN_EMPTY_P.
1207 (default_is_empty_record, arg_int_size_in_bytes,
1208 arg_size_in_bytes, get_ultimate_context): Declare.
1210 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1212 * config/arm/arm.c (cmse_clear_registers): New function.
1213 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
1214 code by call to cmse_clear_registers.
1215 (cmse_nonsecure_entry_clear_before_return): Likewise.
1217 2017-11-22 Tamar Christina <tamar.christina@arm.com>
1219 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
1220 (vdot_s32, vdotq_s32): New.
1221 (vdot_lane_u32, vdotq_lane_u32): New.
1222 (vdot_lane_s32, vdotq_lane_s32): New.
1225 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1228 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
1229 for unsigned values with fewer HWIs than the precision.
1230 (test_overflow): New function.
1231 (wide_int_cc_tests): Call it.
1233 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1235 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
1236 has the same mode class as Pmode.
1238 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1240 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
1241 padding_bits_to_clear_ptr.
1242 (cmse_nonsecure_entry_clear_before_return): Likewise.
1244 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1246 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
1247 auto_sbitap instead of integer bitfield to control register needing
1250 2017-11-22 Jakub Jelinek <jakub@redhat.com>
1252 PR tree-optimization/83044
1253 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
1254 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
1255 that need the upper bound. Subtract offset from
1256 get_addr_base_and_unit_offset only if positive and subtract it
1257 before division by eltsize rather than after it.
1260 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
1261 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
1262 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
1266 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
1267 expand_binop with *mul_widen_optab, make sure at least one of the
1268 operands doesn't have VOIDmode.
1271 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
1273 PR rtl-optimization/82044
1274 PR tree-optimization/82042
1275 * dse.c (record_store): Check for overflow.
1276 (check_mem_read_rtx): Properly check for overflow if width == -1, call
1277 add_wild_read instead of clear_rhs_from_active_local_stores on
1278 overflow and log it into dump_file.
1280 2017-11-22 Richard Biener <rguenther@suse.de>
1282 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
1283 fake edges to exit when looking for a place to insert.
1284 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
1286 (insert_into_preds_of_block): Commit edge insertion immediately,
1287 assert that doesn't require new BBs.
1288 (fini_pre): Release expressions.
1289 (pass_pre::execute): Shuffle things around a bit, if the fn
1290 is too large do not compute AVAIL either as this is really the
1293 2017-11-22 Richard Biener <rguenther@suse.de>
1295 PR tree-optimization/83089
1296 * tree-if-conv.c (pass_if_conversion::execute): If anything
1297 changed reset SCEV and free the number of iteration estimates.
1299 2017-11-21 Martin Sebor <msebor@redhat.com>
1301 PR tree-optimization/82945
1302 * calls.h (warn_nonstring_bound): Remove unused function.
1304 2017-11-21 Martin Sebor <msebor@redhat.com>
1306 PR tree-optimization/82945
1307 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
1308 * calls.h (maybe_warn_nonstring_arg): Declare new function.
1309 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
1311 (initialize_argument_information): Call maybe_warn_nonstring_arg.
1312 * calls.h (get_attr_nonstring_decl): Declare new function.
1313 * doc/extend.texi (attribute nonstring): Update.
1314 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
1315 get_attr_nonstring_decl and handle it.
1316 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
1317 detection of nul-termination.
1318 (strlen_to_stridx): Change to a pointer.
1319 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
1320 (pass_strlen::execute): Same.
1322 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
1324 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
1325 for the new option -mprefer-vector-width=[none|128|256|512].
1326 * config/i386/i386.c (ix86_target_string): remove old style options
1327 -mprefer-avx256 and make -mprefer-avx128 as alias.
1328 (ix86_option_override_internal): Apply defaults for the
1329 -mprefer-vector-width=[128|256] option.
1330 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
1331 Implement macros to work with -mprefer-vector-width=.
1332 * config/i386/i386.opt: Implemented option
1333 -mprefer-vector-width=[none|128|256|512].
1334 * doc/invoke.texi: Documentation for
1335 -mprefer-vector-width=[none|128|256|512].
1337 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
1339 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
1341 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
1343 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
1345 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
1347 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
1349 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
1350 * expr.c (expand_expr_real_2): Likewise.
1351 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
1352 fold_binary_loc): Likewise.
1353 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
1354 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
1355 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
1356 MINUS_EXPR transformations.
1357 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
1358 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
1359 * tree-inline.c (estimate_operator_cost): Likewise.
1360 * tree-pretty-print.c (dump_generic_node, op_code_prio,
1361 op_symbol_code): Likewise.
1362 * tree-vect-stmts.c (vectorizable_operation): Likewise.
1363 * vr-values.c (extract_range_from_binary_expr): Likewise.
1364 * varasm.c (initializer_constant_valid_p_1): Likewise.
1365 * tree.def: New tree code POINTER_DIFF_EXPR.
1367 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
1369 * config/i386/i386.md (*bswap<mode>2_movbe): Add
1370 integer suffix to movbe mnemonic.
1371 (*bswaphi2_movbe): Ditto.
1372 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
1374 2017-11-21 Jakub Jelinek <jakub@redhat.com>
1377 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
1378 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
1379 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
1381 2017-11-21 Martin Liska <mliska@suse.cz>
1383 * tree-inline.c (expand_call_inline): Remove not needed
1386 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
1387 Jakub Jelinek <jakub@redhat.com>
1390 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
1391 Remove static keyword from f variable.
1393 2017-11-21 Jakub Jelinek <jakub@redhat.com>
1395 PR tree-optimization/83086
1396 * gimple-ssa-store-merging.c
1397 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
1398 rather than n.base_addr.
1400 2017-11-21 Martin Liska <mliska@suse.cz>
1402 PR rtl-optimization/82044
1403 PR tree-optimization/82042
1404 * dse.c (check_mem_read_rtx): Check for overflow.
1406 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
1408 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
1411 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
1413 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
1414 and its corresponding call arg location note.
1416 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
1417 Andrew Burgess <andrew.burgess@embecosm.com>
1419 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
1421 (arc_return_slot_offset): Likewise.
1422 (arc_eh_return_address_location): New declaration.
1423 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
1424 (MUST_SAVE_REGISTER): Add exception handler case.
1425 (MUST_SAVE_RETURN_ADDR): Likewise.
1426 (arc_frame_pointer_required): Likewise.
1427 (arc_frame_pointer_needed): New function.
1428 (arc_compute_frame_size): Changed.
1429 (arc_expand_prologue): Likewise.
1430 (arc_expand_epilogue): Likewise.
1431 (arc_initial_elimination_offset): Likewise.
1432 (arc_return_slot_offset): Delete.
1433 (arc_eh_return_address_location): New function.
1434 (arc_builtin_setjmp_frame_value): Likewise.
1435 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
1436 (EH_RETURN_STACKADJ_RTX): Define.
1437 (EH_RETURN_HANDLER_RTX): Likewise.
1438 * config/arc/arc.md (eh_return): Delete.
1440 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
1442 * print-tree.h (debug_vec_tree): Remove prototype.
1443 * gdbinit.in (pvt): Remove macro.
1445 2017-11-21 Jakub Jelinek <jakub@redhat.com>
1447 PR tree-optimization/83047
1448 * gimple-ssa-store-merging.c
1449 (imm_store_chain_info::output_merged_store): If the loads with the
1450 same vuse are in different basic blocks, for load_gsi pick a load
1451 location that is dominated by the other loads.
1454 * config/i386/i386.c (ix86_memmodel_check): Start
1455 -Winvalid-memory-model diagnostics with lowercase letter.
1458 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
1459 set in_cold_section_p to the partition containing loc_list->first.
1460 When seeing loc_list->last_before_switch node, update secname and
1461 perform range_across_switch second partition handling only after that.
1464 * run-rtl-passes.c: Include debug.h.
1465 (run_rtl_passes): Call debug_hooks->assembly_start.
1466 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
1470 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
1471 OPTAB_DIRECT in calls to expand_simple_binop.
1473 2017-11-20 David Malcolm <dmalcolm@redhat.com>
1476 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
1478 2017-11-20 David Malcolm <dmalcolm@redhat.com>
1481 * spellcheck.h (best_match::blithely_get_best_candidate): New
1484 2017-11-20 Jakub Jelinek <jakub@redhat.com>
1486 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
1487 with lower case letter.
1489 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
1491 * config/i386/i386.md (bswaphi2): New expander.
1492 (*bswaphi2_movbe): New insn pattern.
1493 (bswaphi -> rorhi pepehole2): New peephole pattern.
1495 2017-11-20 Jeff Law <law@redhat.com>
1497 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
1498 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
1499 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
1500 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
1501 gimple-ssa-evrp-analyze.h.
1503 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
1504 set BB_VISITED here.
1505 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
1507 2017-11-20 Bin Cheng <bin.cheng@arm.com>
1509 * tree-predcom.c: Add general comment on Store-Store chains.
1510 (split_data_refs_to_components): Postpone clearing eliminate_store_p
1512 (get_chain_last_ref_at): Rename into...
1513 (get_chain_last_write_at): ...this.
1514 (get_chain_last_write_before_load): New function.
1515 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
1516 CT_STORE_STORE when write reference is added.
1517 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
1518 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
1519 (initialize_root_vars_store_elim_1): Ditto.
1520 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
1521 definition is created.
1522 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
1523 chain by replacing it with dominant stored value.
1525 2017-11-20 Bin Cheng <bin.cheng@arm.com>
1527 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
1529 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
1531 * vr-values.c (extract_range_from_binary_expr): Use a full range
1534 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
1536 * config/arm/arm.md (R4_REGNUM): Define constant.
1537 (nonsecure_call_internal): Remove r4 clobber.
1538 (nonsecure_call_value_internal): Likewise.
1539 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
1540 clobber and resequence match_operands.
1541 (nonsecure_call_value_reg_thumb1_v5): Likewise.
1542 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
1543 (nonsecure_call_value_reg_thumb2): Likewise.
1545 2017-11-20 Jakub Jelinek <jakub@redhat.com>
1547 PR tree-optimization/78821
1548 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
1549 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
1550 to the address of the base rather than the base itself.
1551 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
1552 (find_bswap_or_nop_finalize): New function.
1553 (find_bswap_or_nop): Use it.
1554 (bswap_replace): Return a tree rather than bool, change first
1555 argument from gimple * to gimple_stmt_iterator, allow inserting
1556 into an empty sequence, allow ins_stmt to be NULL - then emit
1557 all stmts into gsi. Fix up MEM_REF address gimplification.
1558 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
1559 (struct store_immediate_info): Add N and INS_STMT non-static
1561 (store_immediate_info::store_immediate_info): Initialize them
1562 from newly added ctor args.
1563 (merged_store_group::apply_stores): Formatting fixes. Sort by
1565 (stmts_may_clobber_ref_p): For stores call also
1566 refs_anti_dependent_p.
1567 (gather_bswap_load_refs): New function.
1568 (imm_store_chain_info::try_coalesce_bswap): New method.
1569 (imm_store_chain_info::coalesce_immediate_stores): Use it.
1570 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
1571 (imm_store_chain_info::output_merged_store): Fail if number of
1572 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
1573 and NOP_EXPR rhs_code.
1574 (pass_store_merging::process_store): Compute n and ins_stmt, if
1575 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
1576 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
1578 (pass_store_merging::execute): Calculate dominators.
1580 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
1581 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
1582 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
1583 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
1584 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
1585 class pass_optimize_bswap, bswap_replace,
1586 pass_optimize_bswap::execute): Moved to ...
1587 * gimple-ssa-store-merging.c: ... this file.
1588 Include optabs-tree.h.
1589 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
1590 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
1591 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
1592 anonymous namespace, remove static keywords.
1593 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
1594 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
1596 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
1599 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
1601 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
1603 * vec.h (debug_helper): New function.
1604 (DEFINE_DEBUG_VEC): New macro.
1605 * hash-set.h (debug_helper): New function.
1606 (DEFINE_DEBUG_HASH_SET): New macro.
1607 * cfg.c (debug_slim (edge)): New function.
1608 Call DEFINE_DEBUG_VEC for edges.
1609 Call DEFINE_DEBUG_HASH_SET for edges.
1610 * cfghooks.c (debug_slim (basic_block)): New function.
1611 Call DEFINE_DEBUG_VEC for basic blocks.
1612 Call DEFINE_DEBUG_HASH_SET for basic blocks.
1613 * print-tree.c (debug_slim): New function to handle trees.
1614 Call DEFINE_DEBUG_VEC for trees.
1615 Call DEFINE_DEBUG_HASH_SET for trees.
1616 (debug (vec<tree, va_gc>) &): Remove.
1617 (debug (<vec<tree, va_gc>) *): Remove.
1618 * print-rtl.c (debug_slim): New function to handle const_rtx.
1619 Call DEFINE_DEBUG_VEC for rtx_def.
1620 Call DEFINE_DEBUG_VEC for rtx_insn.
1621 Call DEFINE_DEBUG_HASH_SET for rtx_def.
1622 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
1623 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
1624 (debug (vec<rtx_insn *> *ptr): Remove.
1625 (debug_insn_vector): Remove.
1626 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
1628 2017-11-20 Tom de Vries <tom@codesourcery.com>
1630 PR rtl-optimization/82020
1631 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
1632 IF_THEN_ELSE condition.
1634 2017-11-19 Jeff Law <law@redhat.com>
1636 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1637 of degenerates resulting from ignoring an edge.
1639 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1642 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
1644 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1647 * profile-count.c (profile_count::to_sreal_scale): Fix return value
1648 for uninitialied counts.
1650 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1653 * tree-inline.c (estimate_num_insns): Set to 1 at least.
1655 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
1658 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
1660 2017-11-19 Tom de Vries <tom@codesourcery.com>
1662 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
1665 2017-11-19 Tom de Vries <tom@codesourcery.com>
1667 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
1670 2017-11-19 Tom de Vries <tom@codesourcery.com>
1672 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1675 2017-11-19 Tom de Vries <tom@codesourcery.com>
1677 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
1679 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
1681 2017-11-19 Tom de Vries <tom@codesourcery.com>
1683 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
1685 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
1686 semicolon after MCORE_EXPORT_NAME call.
1688 2017-11-19 Tom de Vries <tom@codesourcery.com>
1691 * vmsdbgout.c (vmsdbgout_early_finish): New function.
1692 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
1694 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
1696 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
1697 (cgraph_edge::clone): Cleanup updating of profile.
1698 * ipa-cp.c (update_profiling_info): Likewise.
1699 * ipa-inline-transform.c (inline_transform): Likewise.
1700 * ipa-inline.c (inline_small_functions): Add missing space to dump.
1701 * ipa-split.c (execute_split_functions): Do not split when function
1703 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
1704 * profile-count.c (profile_count::dump): Add global0.
1705 (profile_count::to_cgraph_frequency): Do not ICE when entry is
1707 (profile_count::to_sreal_scale): Likewise.
1708 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
1709 (profile_count::combine_with_ipa_count): New function.
1710 * profile-count.h (profile_guessed_global0adjusted): New.
1711 (profile_count::adjusted_zero): New.
1712 (profile_count::global0adjusted): New.
1713 (profile_count::combine_with_ipa_count): New.
1714 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
1715 correct profile of return block of split functions.
1716 (copy_cfg_body): Remove unused profile_count.
1717 (copy_body): Likewise.
1718 (expand_call_inline): Update.
1719 (tree_function_versioning): Update.
1721 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
1723 * hash-set.h (hash_set::empty): New.
1724 * tree-ssa-threadbackward.h: Delete.
1725 * tree-ssa-threadbackward.c (class thread_jumps): New.
1726 Move max_threaded_paths into class.
1727 (fsm_find_thread_path): Remove arguments that are now in class.
1728 (profitable_jump_thread_path): Rename to...
1729 (thread_jumps::profitable_jump_thread_path): ...this.
1730 (convert_and_register_jump_thread_path): Rename to...
1731 (thread_jumps::convert_and_register_current_path): ...this.
1732 (check_subpath_and_update_thread_path): Rename to...
1733 (thread_jumps::check_subpath_and_update_thread_path): ...this.
1734 (register_jump_thread_path_if_profitable): Rename to...
1735 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
1736 (handle_phi): Rename to...
1737 (thread_jumps::handle_phi): ...this.
1738 (handle_assignment): Rename to...
1739 (thread_jumps::handle_assignment): ...this.
1740 (fsm_find_control_statement_thread_paths): Rename to...
1741 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
1742 (find_jump_threads_backwards): Rename to...
1743 (thread_jumps::find_jump_threads_backwards): ...this.
1744 Initialize path local data.
1745 (pass_thread_jumps::execute): Call find_jump_threads_backwards
1746 from within thread_jumps class.
1747 (pass_early_thread_jumps::execute): Same.
1749 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1751 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
1753 2017-11-17 Steve Ellcey <sellcey@cavium.com>
1755 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
1756 to canonical location.
1758 2017-11-17 Steve Ellcey <sellcey@cavium.com>
1761 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
1763 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
1765 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
1767 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
1768 rcrt1.o%s/grcrt1.o%s for -static-pie.
1770 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1772 * i386.c (ix86_multiplication_cost, ix86_division_cost,
1773 ix86_shift_rotate_cost): Break out from ...
1774 (ix86_rtx_costs): ... here.
1775 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
1778 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1780 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
1781 when entry block was promoted unlikely.
1782 (estimate_bb_frequencies): Increase frequency scale.
1783 * profile-count.h (profile_count): Export precision info.
1785 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1787 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
1788 disturbin profile of entry block.
1790 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1792 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
1795 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
1797 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
1798 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
1800 2017-11-17 Jeff Law <law@redhat.com>
1802 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
1803 from evrp_dom_walker class. Various methods moved into new class.
1804 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
1805 (evrp_range_analyzer::enter): New method.
1806 (evrp_range_analyzer::leave): New method.
1807 (evrp_dom_walker): Remove delegators no longer needed by this class.
1808 Replace vr_values data member with evrp_range_analyzer
1810 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
1811 method extracted from evrp_dom_walker::before_dom_children.
1812 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
1813 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
1815 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
1816 Add private copy constructor and move assignment operators.
1817 Privatize methods and class data where trivially possible.
1818 (evrp_dom_walker::cleanup): New function, extracted from
1819 execute_early_vrp. Simplify access to class data.
1821 * vr-values.h (get_output_for_vrp): Prototype.
1822 * vr-values.c (get_output_for_vrp): New function extracted from
1823 vrp_visit_assignment_or_call and extract_range_from_stmt.
1824 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
1826 2017-11-17 Luis Machado <luis.machado@linaro.org>
1828 * config/aarch64/aarch64.c
1829 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
1830 (qdf24xx_tunings) <autoprefetcher_model>: Set to
1831 tune_params::AUTOPREFETCHER_WEAK.
1833 2017-11-17 Tamar Christina <tamar.christina@arm.com>
1836 * config/arm/arm.c (arm_valid_target_attribute_rec):
1837 Parse "arch=" and "+<ext>".
1838 (arm_valid_target_attribute_tree): Re-init global options.
1839 (arm_option_override): Make non-static.
1840 (arm_options_perform_arch_sanity_checks): Make errors fatal.
1841 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
1842 (__ARM_FEATURE_CRC32): Support undef.
1843 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
1844 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
1846 2017-11-17 David Malcolm <dmalcolm@redhat.com>
1848 * gdbinit.in (break-on-diagnostic): New command.
1850 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1852 * config/i386/i386.c (ix86_expand_epilogue): Change simple
1853 return to indirect jump for EH return if control-flow
1854 protection is enabled. Change explicit 'false' argument in
1855 pro_epilogue_adjust_stack with a value of
1857 * config/i386/i386.md (simple_return_indirect_internal):
1858 Remove SImode restriction to support 64-bit.
1860 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
1862 * combine.c (added_notes_insn): New.
1863 (try_combine): Handle added_notes_insn like added_links_insn.
1864 Rewrite return value code.
1865 (distribute_notes): Set added_notes_insn to the earliest insn we added
1868 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
1870 PR rtl-optimization/82621
1871 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
1872 dest of one of those SETs is unused.
1874 2017-11-17 Richard Biener <rguenther@suse.de>
1877 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
1878 * tree-pretty-print.c (dump_generic_node): Handle
1879 annot_expr_parallel_kind.
1880 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
1881 * gimplify.c (gimple_boolify): Likewise.
1883 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1885 * config.gcc (extra_headers): Add cet.h for x86 targets.
1886 * config/i386/cet.h: New file.
1887 * doc/install.texi: Add --enable-cet/--disable-cet.
1889 2017-11-17 Richard Biener <rguenther@suse.de>
1891 PR tree-optimization/83017
1892 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
1893 (gen_parallel_loop): Properly count iterations.
1894 (parallelize_loops): Handle loop->can_be_parallel independent
1895 of flag_loop_parallelize_all. Make static profitability test match
1897 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
1898 * invoke.texi (parloops-min-per-thread): Document.
1900 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
1902 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
1903 upstreaming review comments.
1905 2017-11-17 Tamar Christina <tamar.christina@arm.com>
1907 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
1908 with fast unaligned access.
1909 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
1911 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
1913 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
1914 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
1916 2017-11-17 Richard Biener <rguenther@suse.de>
1918 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
1919 folding of references.
1921 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
1924 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
1925 of replacing call to strncmp with corresponding call to strcmp when
1928 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
1930 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
1931 option prefer-avx256 for skylake-avx512 configuration.
1932 * config/i386/i386.c (ix86_option_override_internal): Ditto.
1933 (get_builtin_code_for_version): Ditto.
1935 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
1936 Monk Chiang <sh.chiang04@gmail.com>
1938 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
1939 (FIXED_REGISTERS): Reserve more register numbers.
1940 (CALL_USED_REGISTERS): Likewise.
1941 (REG_ALLOC_ORDER): Likewise.
1942 (REG_CLASS_CONTENTS): Likewise.
1943 (REGISTER_NAMES): Likewise.
1945 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
1946 Kito Cheng <kito.cheng@gmail.com>
1948 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
1950 * config/nds32/iterators.md: Add vector mode iterators and attributes.
1952 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
1954 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
1955 parameter list for vec_splats.
1957 2017-11-16 Joseph Myers <joseph@codesourcery.com>
1959 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
1961 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
1963 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
1966 * doc/install.texi: Document incompatibility of Perl >=5.6.26
1967 with the required version of automake 1.11.6.
1969 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
1971 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
1973 (DU_C2_3_power9): Correct reservation combinations.
1974 (FP_DIV_power9, VEC_DIV_power9): New.
1975 (power9-alu): Split out rotate/shift...
1976 (power9-rot): ...to here, correct dispatch resource.
1977 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
1979 (power9-fp): Correct latency.
1980 (power9-sdiv): Add div/sqrt resource.
1981 (power9-ddiv): Correct latency, add div/sqrt resource.
1982 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
1983 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
1985 (power9-qpdiv, power9-qpmul): Adjust resource usage.
1987 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1989 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
1990 switch statement mapping KF built-ins to TF built-ins if we don't
1991 have the proper ISA 3.0 assembler support.
1993 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
1995 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
1996 (lower_emutls_function_body): Do not compute it.
1998 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2000 * ipa-split.c (split_bb_info): Turn time to sreal.
2001 (split_point): Likewise.
2002 (dump_split_point): Likewise.
2003 (fine_split_points): Likewise.
2004 (execute_split_functions): Only zero split_bbs; turn time to sreals.
2006 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2008 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
2011 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2013 * predict.c (combine_predictions_for_bb): Preserve zero predicted
2015 (expensive_function_p): Remove useless assert.
2016 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
2018 2017-11-16 Martin Sebor <msebor@redhat.com>
2020 PR tree-optimization/82588
2021 PR tree-optimization/82583
2022 * tree-vrp.c (check_array_ref): Handle flexible array members,
2023 string literals, and inner indices.
2024 (search_for_addr_array): Add detail to diagnostics.
2026 2017-11-16 Nathan Sidwell <nathan@acm.org>
2030 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
2031 (SET_DECL_ASSEMBLER_NAME): Forward to
2032 overwrite_decl_assembler_name.
2033 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
2034 (overwrite_decl_assembler_name): Declare.
2035 * tree.c (overwrite_decl_assembler_name): New.
2036 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
2037 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
2038 (LANG_HOOKS_INITIALIZER): Add it.
2039 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
2040 * langhooks.c (lhd_set_decl_assembler_name): Use
2041 SET_DECL_ASSEMBLER_NAME.
2042 (lhd_overwrite_decl_assembler_name): Default implementation.
2044 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
2045 Jackson Woodruff <jackson.woodruff@arm.com>
2047 PR tree-optimization/71026
2048 * match.pd: Canonicalize constant multiplies in division.
2050 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
2052 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
2053 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
2054 Remove OPT_fomit_frame_pointer entry.
2055 * common/config/arc/arc-common.c: Likewise.
2056 * common/config/arm/arm-common.c: Likewise.
2057 * common/config/avr/avr-common.c: Likewise.
2058 * common/config/c6x/c6x-common.c: Likewise.
2059 * common/config/cr16/cr16-common.c: Likewise.
2060 * common/config/cris/cris-common.c: Likewise.
2061 * common/config/epiphany/epiphany-common.c: Likewise.
2062 * common/config/fr30/fr30-common.c: Likewise.
2063 * common/config/frv/frv-common.c: Likewise.
2064 * common/config/ia64/ia64-common.c: Likewise.
2065 * common/config/iq2000/iq2000-common.c: Likewise.
2066 * common/config/lm32/lm32-common.c: Likewise.
2067 * common/config/m32r/m32r-common.c: Likewise.
2068 * common/config/mcore/mcore-common.c: Likewise.
2069 * common/config/microblaze/microblaze-common.c: Likewise.
2070 * common/config/mips/mips-common.c: Likewise.
2071 * common/config/mmix/mmix-common.c: Likewise.
2072 * common/config/mn10300/mn10300-common.c: Likewise.
2073 * common/config/nios2/nios2-common.c: Likewise.
2074 * common/config/pa/pa-common.c: Likewise.
2075 * common/config/pdp11/pdp11-common.c: Likewise.
2076 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
2077 * common/config/riscv/riscv-common.c: Likewise.
2078 * common/config/rs6000/rs6000-common.c: Likewise.
2079 * common/config/rx/rx-common.c: Likewise.
2080 * common/config/s390/s390-common.c: Likewise.
2081 * common/config/sh/sh-common.c: Likewise.
2082 * common/config/sparc/sparc-common.c: Likewise.
2083 * common/config/tilegx/tilegx-common.c: Likewise.
2084 * common/config/tilepro/tilepro-common.c: Likewise.
2085 * common/config/v850/v850-common.c: Likewise.
2086 * common/config/visium/visium-common.c: Likewise.
2087 * common/config/xstormy16/xstormy16-common.c: Likewise.
2088 * common/config/xtensa/xtensa-common.c: Likewise.
2089 * invoke.texi (-fomit-frame-pointer): Update documentation.
2091 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2093 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
2095 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2097 * cfg.c (scale_bbs_frequencies_int,
2098 cale_bbs_frequencies_gcov_type): Remove.
2099 * cfg.h (scale_bbs_frequencies_int,
2100 cale_bbs_frequencies_gcov_type): Remove.
2102 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2104 * tree-ssa-loop-manip.c
2105 (scale_dominated_blocks_in_loop): Update to profile counts.
2106 (tree_transform_and_unroll_loop): Likewise.
2108 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2110 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
2111 scale_bbs_frequencies_int.
2113 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2115 * final.c (compute_alignments): Use counts rather than frequencies.
2117 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2119 * cfgloopanal.c: Include sreal.h
2120 (average_num_loop_insns): Use counts and sreal for accounting.
2122 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2124 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
2127 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2128 Kito Cheng <kito.cheng@gmail.com>
2130 * config/nds32/constraints.md: Provide more constraints.
2131 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
2132 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
2133 support constraints usage.
2135 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2137 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
2139 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2140 Kito Cheng <kito.cheng@gmail.com>
2142 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
2143 * config/nds32/nds32.opt: Refine the layout.
2144 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
2145 TARGET_EXT_STRING): Support new options.
2146 * config/nds32/nds32.h: Likewise.
2147 * config/nds32/nds32.md: Likewise.
2148 * config/nds32/nds32-predicates.c: Likewise.
2149 * config/nds32/constraints.md: Likewise.
2150 * common/config/nds32/nds32-common.c: Likewise.
2152 2017-11-16 Julia Koval <julia.koval@intel.com>
2155 * config/i386/gfniintrin.h: Add sse check.
2156 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
2158 2017-11-16 Julia Koval <julia.koval@intel.com>
2160 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
2161 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
2162 (ix86_handle_option): Handle -mavx512vbmi2.
2163 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
2164 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
2165 * config/i386/i386-c.c (__AVX512VBMI2__): New.
2166 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
2167 (ix86_valid_target_attribute_inner_p): Ditto.
2168 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
2169 * config/i386/i386.opt (mavx512vbmi2): New option.
2170 * doc/invoke.texi: Add new option.
2172 2017-11-16 Julia Koval <julia.koval@intel.com>
2174 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
2175 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
2176 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
2177 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
2178 _mm512_gf2p8mul_epi8): New intrinsics.
2179 * config/i386/i386-builtin-types.def
2180 (V64QI_FTYPE_V64QI_V64QI): New type.
2181 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
2182 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
2183 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
2184 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
2185 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
2186 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2188 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
2190 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
2191 explicitly as a stream of bytes.
2193 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2195 * config/rs6000/altivec.h (vec_xst_be): New #define.
2196 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
2197 and externalize from *altivec_vperm_<mode>_internal.
2198 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
2200 (XL_BE_V8HI): Likewise.
2201 (XL_BE_V4SI): Likewise.
2202 (XL_BE_V4SI): Likewise.
2203 (XL_BE_V2DI): Likewise.
2204 (XL_BE_V4SF): Likewise.
2205 (XL_BE_V2DF): Likewise.
2206 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
2207 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
2208 all array entries with these keys: VSX_BUILTIN_VEC_XL,
2209 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
2210 VSX_BUILTIN_VEC_XST_BE.
2211 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
2212 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
2214 (altivec_init_builtins): Replace conditional calls to def_builtin
2215 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
2216 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
2217 with unconditional calls. Remove calls to def_builtin for
2218 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
2219 __builtin_vec_xst_be.
2220 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
2221 to define_expand, and add alternate RTL generation for P8.
2222 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
2223 vsx_ld_elemrev_v8hi.
2224 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
2225 add alternate RTL generation for P8.
2226 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
2227 vsx_ld_elemrev_v16qi.
2228 (vsx_st_elemrev_v8hi): Convert define_insn
2229 to define_expand, and add alternate RTL generation for P8.
2230 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
2231 vsx_st_elemrev_v8hi.
2232 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
2233 add alternate RTL generation for P8.
2234 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
2235 vsx_st_elemrev_v16qi.
2237 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
2240 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
2241 TARGET_AVX512ER check.
2242 (ix86_option_override_internal): Set MASK_VZEROUPPER if
2243 neither -mzeroupper nor -mno-zeroupper is used and
2244 TARGET_EMIT_VZEROUPPER is set.
2245 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
2246 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
2248 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
2250 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2251 folding of vector compares.
2252 (fold_build_vec_cmp): New helper function.
2253 (fold_compare_helper): New helper function.
2254 (builtin_function_type): Add compare builtins to the list of functions
2255 having unsigned arguments. Cosmetic updates to comment indentation.
2256 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
2257 the not+eq combination.
2259 2017-11-15 Bin Cheng <bin.cheng@arm.com>
2261 PR tree-optimization/82726
2262 PR tree-optimization/70754
2263 * tree-predcom.c (order_drefs_by_pos): New function.
2264 (combine_chains): Move code setting has_max_use_after to...
2265 (try_combine_chains): ...here. New parameter. Sort combined chains
2266 according to position information.
2267 (tree_predictive_commoning_loop): Update call to above function.
2268 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
2270 2017-11-15 Bin Cheng <bin.cheng@arm.com>
2272 PR tree-optimization/82726
2274 2017-01-23 Bin Cheng <bin.cheng@arm.com>
2276 PR tree-optimization/70754
2277 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
2278 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
2279 combined stmt before it if not NULL.
2280 (combine_chains): Process refs reversely and compute dominance point
2284 2017-02-23 Bin Cheng <bin.cheng@arm.com>
2286 PR tree-optimization/79663
2287 * tree-predcom.c (combine_chains): Process refs in reverse order
2288 only for ZERO length chains, and add explaining comment.
2290 2017-11-15 Tamar Christina <tamar.christina@arm.com>
2292 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
2293 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
2294 * doc/invoke.texi (ARM Options): Add armv8.3-a.
2296 2017-11-15 Tamar Christina <tamar.christina@arm.com>
2298 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
2300 2017-11-15 Martin Liska <mliska@suse.cz>
2302 * tree-cfg.c (pass_warn_function_return::execute):
2303 Compare warn_return_type for greater than zero.
2305 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
2309 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
2310 to return true on Xeon and not on Xeon Phi.
2311 (ix86_check_avx256_register): Changed to ...
2312 (ix86_check_avx_upper_register): ... this. Add extra check for
2313 VALID_AVX512F_REG_OR_XI_MODE.
2314 (ix86_avx_u128_mode_needed): Changed
2315 ix86_check_avx256_register to ix86_check_avx_upper_register.
2316 (ix86_check_avx256_stores): Changed to ...
2317 (ix86_check_avx_upper_stores): ... this. Changed
2318 ix86_check_avx256_register to ix86_check_avx_upper_register.
2319 (ix86_avx_u128_mode_after): Changed
2320 avx_reg256_found to avx_upper_reg_found. Changed
2321 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
2322 (ix86_avx_u128_mode_entry): Changed
2323 ix86_check_avx256_register to ix86_check_avx_upper_register.
2324 (ix86_avx_u128_mode_exit): Ditto.
2325 * config/i386/i386.h: (host_detect_local_cpu): New define.
2327 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2329 * config/arm/xgene1.md (xgene1): Split into automatons
2330 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
2331 (xgene1_f_load): Adjust reservations and/or types.
2332 (xgene1_f_store): Likewise.
2333 (xgene1_load_pair): Likewise.
2334 (xgene1_store_pair): Likewise.
2335 (xgene1_fp_load1): Likewise.
2336 (xgene1_load1): Likewise.
2337 (xgene1_store1): Likewise.
2338 (xgene1_move): Likewise.
2339 (xgene1_alu): Likewise.
2340 (xgene1_simd): Likewise.
2341 (xgene1_bfm): Likewise.
2342 (xgene1_neon_load1): Likewise.
2343 (xgene1_neon_store1): Likewise.
2344 (xgene1_neon_logic): Likewise.
2345 (xgene1_neon_st1): Likewise.
2346 (xgene1_neon_ld1r): Likewise.
2347 (xgene1_alu_cond): Added.
2348 (xgene1_shift_reg): Likwise.
2349 (xgene1_bfx): Likewise.
2350 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
2352 2017-11-15 Jakub Jelinek <jakub@redhat.com>
2355 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
2357 (can_widen_mult_without_libcall): New function.
2358 (expand_mul_overflow): If only checking unsigned mul overflow,
2359 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
2360 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
2361 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
2362 (expand_DIVMOD): Formatting fix.
2363 * expmed.h (expand_mult): Add NO_LIBCALL argument.
2364 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
2365 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
2367 PR tree-optimization/82977
2368 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
2369 constructed temporary to strlen_to_stridx.put.
2371 2017-11-15 Martin Liska <mliska@suse.cz>
2373 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
2374 * configure: Regenerate.
2376 2017-11-15 Martin Liska <mliska@suse.cz>
2379 * config/sh/sh-mem.cc: Use proper probability for
2382 2017-11-14 Jeff Law <law@redhat.com>
2384 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
2385 the red zone for stack_clash_protection_final_dynamic_probe targets
2386 when the total dynamic stack size is zero bytes.
2388 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
2389 blocks is post order.
2391 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
2393 * dumpfile.h (TDF_COMPARE_DEBUG): New.
2394 * final.c (rest_of_clean_state): Set it for the
2395 -fcompare-debug dump.
2396 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
2397 class when TDF_COMPARE_DEBUG is set.
2399 * dwarf2out.c (gen_producer_string): Discard
2402 2017-11-15 Joseph Myers <joseph@codesourcery.com>
2405 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
2406 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
2407 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
2408 (__TGMATH_REAL_2_3): Remove macros.
2409 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
2410 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
2412 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
2414 (remquo): Define using __TGMATH_REAL_3.
2416 2017-11-14 Jeff Law <law@redhat.com>
2418 * vr-values.c: New file with contents extracted from tree-vrp.c.
2419 * Makefile.in (OBJS): Add vr-values.o
2420 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
2421 (set_value_range, set_and_canonicalize_value_range): Likewise.
2422 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
2423 (value_range_constant_singleton, symbolic_range_p): Likewise.
2424 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
2425 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
2426 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
2427 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
2428 (find_case_label_range, find_case_label_index): Likewise.
2429 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
2430 (range_int_cst_singleton_p, value_inside_range): Likewise.
2431 (get_single_symbol): Likewise.
2432 (switch_update): Move structure definition here.
2433 (to_remove_edges, to_update_switch_stmts): Provide externs.
2434 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
2435 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
2436 (vrp_val_is_min, set_value_range): Likewise.
2437 (set_and_canonicalize_value_range, copy_value_range): Likewise.
2438 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
2439 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
2440 (range_is_nonnull, range_int_cst_p): Likewwise.
2441 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
2442 (get_single_symbol, operand_less_p): Likewise
2443 (compare_values_warnv, compare_values): Likewise.
2444 (value_inside_range, value_range_constant_singleton): Likewise.
2445 (zero_nonzero_bitgs_from_vr): Likewise.
2446 (extract_range_from_binary_expr_1): Likewise.
2447 (overflow_comparison_p): Likewise.
2448 (to_remove_edges, to_update_switch_stmts): Likewise.
2449 (find_case_label-index, find_case_label_range): Likewise.
2450 (switch_update, set_value_range_to_nonnegative): Remove.
2451 (set_value_range_to_truthvalue): Likewise.
2452 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
2453 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
2454 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
2455 (find_case_label_ranges, test_for_singularity): Likewise.
2456 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
2457 (x_vr_values): Move to its remaining use site.
2459 2017-11-10 Jeff Law <law@redhat.com>
2461 * vr-values.h (VR_INITIALIZER): Move #define here.
2462 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
2463 * Makefile.in (OBJS): Add tree-evrp.o
2464 * tree-vrp.h (assert_info): Move structure definition here.
2465 (set_value_range_to_varying): Prototype.
2466 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
2467 (infer_value_range, register_edge_assert_for): Likewise.
2468 (stmt_interesting_for_vrp): Likewise.
2469 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
2470 (set_value_range_to_varying): No longer static.
2471 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
2472 (infer_value_range, register_edge_assert_for): Likewise.
2474 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
2476 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
2477 generate the XXBRD instruction.
2479 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
2480 (rs6000_builtin_type_compatible): Treat _Float128 and long double
2481 as being compatible if -mabi=ieeelongdouble.
2482 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
2483 to setup float128 built-ins with hardware support.
2484 (BU_FLOAT128_HW_2): Likewise.
2485 (BU_FLOAT128_HW_3): Likewise.
2486 (BU_FLOAT128_HW_VSX_1): Likewise.
2487 (BU_FLOAT128_HW_VSX_2): Likewise.
2488 (scalar_extract_expq): Change float128 built-in functions to
2489 accommodate having both KFmode and TFmode functions. Use the
2490 KFmode variant as the default.
2491 (scalar_extract_sigq): Likewise.
2492 (scalar_test_neg_qp): Likewise.
2493 (scalar_insert_exp_q): Likewise.
2494 (scalar_insert_exp_qp): Likewise.
2495 (scalar_test_data_class_qp): Likewise.
2496 (sqrtf128_round_to_odd): Delete processing the round to odd
2497 built-in functions as special built-in functions, and define them
2498 as float128 built-ins. Use the KFmode variant as the default.
2499 (truncf128_round_to_odd): Likewise.
2500 (addf128_round_to_odd): Likewise.
2501 (subf128_round_to_odd): Likewise.
2502 (mulf128_round_to_odd): Likewise.
2503 (divf128_round_to_odd): Likewise.
2504 (fmaf128_round_to_odd): Likewise.
2505 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
2506 support for KFmode and TFmode xststdcqp calls.
2507 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
2508 point, switch the built-in handlers for the get/set float128
2509 exponent, get float128 mantissa, float128 test built-ins, and the
2510 float128 round to odd built-in functions. Eliminate creating the
2511 float128 round to odd built-in functions as special built-ins.
2512 (rs6000_init_builtins): Eliminate special creation of the float128
2513 round to odd built-in functions.
2514 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
2515 function insns to support both TFmode and KFmode variants.
2516 (xsxsigqp_<mode>): Likewise.
2517 (xsiexpqpf_<mode>): Likewise.
2518 (xsiexpqp_<mode>): Likewise.
2519 (xststdcqp_<mode>): Likewise.
2520 (xststdcnegqp_<mode>): Likewise.
2521 (xststdcqp_<mode>): Likewise.
2523 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2525 * tree-ssa-threadupdate.c (compute_path_counts): Remove
2526 unused path_in_freq_ptr parameter.
2527 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
2529 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2531 * ipa-inline.c (edge_badness): Dump sreal frequency.
2532 (compute_inlined_call_time): Match natural implementaiton ...
2533 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
2534 forgotten division by CGRAPH_FREQ_BASE.
2536 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2538 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
2539 Solaris 11. Update comment.
2540 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
2542 * config/sol2.h (STARTFILE_SPEC): Likewise.
2543 * configure: Regenerate.
2545 2017-11-14 Carl Love <cel@us.ibm.com>
2547 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
2548 le_ and be_ prefixes to swap* variables. Remove
2549 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
2552 2017-11-14 Jason Merrill <jason@redhat.com>
2554 Support GTY((cache)) on hash_map.
2555 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
2556 (ggc_cache_remove): Override it instead of ggc_mx.
2557 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
2558 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
2559 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
2560 (simple_cache_map_traits): Override maybe_mx.
2561 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
2562 (hash_map): Friend gt_cleare_cache.
2563 (gt_cleare_cache): New.
2564 * tree.h (tree_cache_traits): New hash_map traits class.
2565 (tree_cache_map): New typedef.
2567 2017-11-14 Richard Biener <rguenther@suse.de>
2569 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
2570 paramter and handling.
2571 (cleanup_control_flow_bb): Likewise.
2572 (cleanup_control_flow_pre): New helper performing a DFS walk
2573 to call cleanup_control_flow_bb in PRE order.
2574 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
2575 via cleanup_control_flow_pre.
2577 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
2579 * config/aarch64/aarch64-simd.md
2580 (aarch64_simd_bsl<mode>_internal): Remove DImode.
2581 (*aarch64_simd_bsl<mode>_alt): Likewise.
2582 (aarch64_simd_bsldi_internal): New.
2583 (aarch64_simd_bsldi_alt): Likewise.
2585 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2587 * tracer.c (better_p): Do not compare frequencies.
2588 * reg-stack.c (better_edge): Likewise.
2589 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
2592 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2594 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
2595 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
2596 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
2597 * lto-streamer-in.c (input_function): Use update_max_bb_count.
2598 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
2599 * predict.c (maybe_hot_frequency_p): Inline to ...
2600 (maybe_hot_count_p): ... here; rewrite to counts.
2601 (counts_to_freqs): Rename to ...
2602 (update_max_bb_count): ... this one.
2603 (expensive_function_p): Use counts.
2604 (estimate_bb_frequencies): Update.
2605 (rebuild_frequencies): Update.
2606 * predict.h (counts_to_freqs): Rename to ...
2607 (update_max_bb_count): ... this one.
2608 * profile.c (compute_branch_probabilities): Add debug info
2609 * tree-inline.c (expand_call_inline): Update debug info.
2610 (optimize_inline_calls): Use update_max_bb_count..
2611 (tree_function_versioning): Use update_max_bb_count..
2612 * value-prof.c (gimple_value_profile_transformations):
2613 Do not use update_max_bb_count.
2615 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2617 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
2618 always use frequencies.
2620 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2622 * bb-reorder.c: Remove frequencies from comments.
2623 (better_edge_p): Use profile counts.
2624 (find_traces): Dump profile counts.
2625 (rotate_loop): Use profile counts.
2626 (find_traces_1_round): Likewise.
2627 (connect_better_edge_p): Use counts instead of probabilities for
2629 (copy_bb_p): Drop early check for non-0 frequency.
2630 (sanitize_hot_paths): Update comments.
2632 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2634 * ipa-split.c (struct split_point): Add count.
2635 (consider_split): Do not compute incoming frequency; compute incoming
2636 count and store it to split_point.
2637 (split_function): Set count of the call to split part correctly.
2639 2017-11-13 Carl Love <cel@us.ibm.com>
2641 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
2643 2017-11-13 Tom Tromey <tom@tromey.com>
2645 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
2647 2017-11-13 Carl Love <cel@us.ibm.com>
2649 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
2650 Add support for builtins:
2651 unsigned int vec_first_{,miss}_match_{,or_eos}index,
2652 vector {un,}signed {char,int,short},
2653 vector {un,}signed {char,int,short}) arguments.
2654 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
2655 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
2656 Add BU_P9V_AV_2 expansions for the builtins.
2657 * config/rs6000/altivec.h (vec_first_match_index,
2658 vec_first_mismatch_index, vec_first_match_or_eos_index,
2659 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
2660 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
2661 new extern declaration.
2662 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
2663 * config/rs6000/vsx.md (first_match_index_<mode>,
2664 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
2665 first_mismatch_or_eos_index_<mode>): Add define expand.
2666 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
2667 * doc/extend.texi: Update the built-in documenation file for the new
2670 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
2672 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
2673 into the min/max operations for _Float<N> and _Float<N>X types.
2675 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
2678 * dwarf2out.c (do_eh_frame): New static variable.
2679 (dwarf2out_begin_prologue): Set it.
2680 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
2682 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2684 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
2686 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
2688 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
2691 * debug.h (dwarf2out_do_eh_frame): Declare.
2692 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
2693 (dwarf2out_do_frame): Use it.
2694 (dwarf2out_do_cfi_asm): Likewise.
2695 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
2696 (dwarf2out_assembly_start): Likewise.
2697 (dwarf2out_begin_prologue): Fix comment.
2698 * toplev.c (compile_file): Always call dwarf2out_frame_finish
2699 if the target needs either debug or unwind DWARF2 info.
2700 * lto-opts.c (lto_write_options): Do not save -fexceptions,
2701 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
2702 -ftrapping-math, -ftrapv and -fwrapv.
2704 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
2706 * cgraph.c (cgraph_edge::sreal_frequency): New function.
2707 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
2708 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
2709 (estimate_edge_size_and_time): Likewise.
2710 (ipa_merge_fn_summary_after_inlining): Likewise.
2711 * ipa-inline.c (cgraph_freq_base_rec): Remove.
2712 (compute_uninlined_call_time): Use sreal_frequency.
2713 (compute_inlined_call_time): Likewise.
2714 (ipa_inline): Do not initialize cgraph_freq_base_rec.
2715 * profile-count.c: Include sreal.h.
2716 (profile_count::to_sreal_scale): New.
2717 * profile-count.h: Forward declare sreal.
2718 (profile_count::to_sreal_scale): Declare.
2720 2017-11-13 Nathan Sidwell <nathan@acm.org>
2722 * diagnostic.c (maybe_line_and_column): New.
2723 (diagnostic_get_location_text): Use it.
2724 (diagnostic_report_current_module): Likewise.
2725 (test_diagnostic_get_location_text): Add tests.
2727 2017-11-13 Luis Machado <luis.machado@linaro.org>
2729 * doc/md.texi (Specifying processor pipeline description): Fix
2730 incorrect latency for the div instruction example.
2732 2017-11-13 Jakub Jelinek <jakub@redhat.com>
2734 PR tree-optimization/78821
2735 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
2736 that bit_not_p is the same.
2737 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
2738 (split_group): Count precisely bit_not_p bits in each statement.
2739 (invert_op): New function.
2740 (imm_store_chain_info::output_merged_store): Use invert_op to
2741 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
2742 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
2744 2017-11-13 Martin Liska <mliska@suse.cz>
2746 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
2747 (struct source_info): Likewise.
2748 (add_branch_counts): Likewise.
2749 (add_line_counts): Likewise.
2750 (function_summary): Likewise.
2751 (output_intermediate_line): Likewise.
2752 (generate_results): Likewise.
2754 2017-11-13 Martin Liska <mliska@suse.cz>
2756 * gcov.c (struct block_info): Remove typedef for block_t.
2757 (struct line_info): Likewise.
2758 (line_info::has_block): Likewise.
2759 (EXIT_BLOCK): Likewise.
2760 (unblock): Likewise.
2761 (circuit): Likewise.
2762 (get_cycles_count): Likewise.
2763 (process_file): Likewise.
2764 (read_graph_file): Likewise.
2765 (solve_flow_graph): Likewise.
2766 (find_exception_blocks): Likewise.
2767 (add_line_counts): Likewise.
2768 (accumulate_line_info): Likewise.
2769 (output_line_details): Likewise.
2771 2017-11-13 Martin Liska <mliska@suse.cz>
2773 * gcov.c (struct arc_info): Remove typedef for arc_t.
2774 (struct line_info): Likewise.
2775 (add_branch_counts): Likewise.
2776 (output_branch_count): Likewise.
2777 (function_info::~function_info): Likewise.
2778 (circuit): Likewise.
2779 (output_intermediate_line): Likewise.
2780 (read_graph_file): Likewise.
2781 (solve_flow_graph): Likewise.
2782 (find_exception_blocks): Likewise.
2783 (add_line_counts): Likewise.
2784 (accumulate_line_info): Likewise.
2785 (output_line_details): Likewise.
2786 (output_function_details): Likewise.
2788 2017-11-13 Martin Liska <mliska@suse.cz>
2790 * gcov.c (struct function_info): Remove typedef for function_t.
2791 (struct source_info): Likewise.
2792 (source_info::get_functions_at_location): Likewise.
2793 (solve_flow_graph): Likewise.
2794 (find_exception_blocks): Likewise.
2795 (add_line_counts): Likewise.
2796 (output_intermediate_file): Likewise.
2797 (process_file): Likewise.
2798 (generate_results): Likewise.
2799 (release_structures): Likewise.
2800 (read_graph_file): Likewise.
2801 (read_count_file): Likewise.
2802 (accumulate_line_counts): Likewise.
2803 (output_lines): Likewise.
2805 2017-11-13 Martin Liska <mliska@suse.cz>
2807 * gcov.c (function_info::function_info): Remove num_counts
2808 and add vector<gcov_type>.
2809 (function_info::~function_info): Use the vector.
2810 (process_file): Likewise.
2811 (read_graph_file): Likewise.
2812 (read_count_file): Likewise.
2813 (solve_flow_graph): Likewise.
2815 2017-11-13 Martin Liska <mliska@suse.cz>
2817 * gcov.c (function_info::is_artificial): New function.
2818 (process_file): Erase all artificial early.
2819 (generate_results): Skip as all artificial are already
2822 2017-11-13 Martin Liska <mliska@suse.cz>
2824 * gcov.c (read_graph_file): Store to global vector of functions.
2825 (read_count_file): Iterate the vector.
2826 (process_file): Likewise.
2827 (generate_results): Likewise.
2828 (release_structures): Likewise.
2830 2017-11-13 Jakub Jelinek <jakub@redhat.com>
2832 PR tree-optimization/82954
2833 * gimple-ssa-store-merging.c
2834 (imm_store_chain_info::coalesce_immediate_stores): If
2835 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
2837 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
2839 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
2840 Upddate call to ENDIAN_LANE_N.
2841 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
2842 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
2843 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
2844 and use aarch64_endian_lane_rtx.
2845 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
2847 2017-11-12 Tom de Vries <tom@codesourcery.com>
2849 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
2851 2017-11-12 Tom de Vries <tom@codesourcery.com>
2853 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
2855 2017-11-12 Tom de Vries <tom@codesourcery.com>
2857 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
2859 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
2860 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
2861 * defaults.h (ASM_OUTPUT_LABELREF): Same.
2863 2017-11-11 Martin Sebor <msebor@redhat.com>
2866 * doc/extend.texi (attribute nonstring): Remove spurious argument.
2869 * prefic.c (translate_name): Replace strncpy with memcpy to
2870 avoid -Wstringop-truncation.
2872 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
2874 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
2876 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
2878 * predict.c (maybe_hot_frequency_p): Do not use cfun.
2880 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
2882 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
2885 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
2887 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
2888 enable generating XXBRH if the value is in a vector register.
2889 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
2890 value is in a vector register.
2891 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
2892 register to register bswap64's instead of doing the GPR sequence
2893 used on previous machines.
2894 (bswapdi2_xxbrd): New insn.
2895 (bswapdi2_reg): Disallow on ISA 3.0.
2896 (register to register bswap64 splitter): Do not split the insn on
2897 ISA 3.0 systems that use XXBRD.
2899 2017-11-10 Martin Sebor <msebor@redhat.com>
2902 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
2904 (find_subframework_file): Same.
2906 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
2908 * auto-profile.c (afdo_indirect_call): Drop frequency.
2909 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
2910 (cgraph_node::create_edge): Drop frequency argument.
2911 (cgraph_node::create_indirect_edge): Drop frequency argument.
2912 (cgraph_edge::make_speculative): Drop frequency arguments.
2913 (cgraph_edge::resolve_speculation): Do not update frequencies
2914 (cgraph_edge::dump_edge_flags): Do not dump frequency.
2915 (cgraph_node::dump): Check consistency in IPA mode.
2916 (cgraph_edge::maybe_hot_p): Use IPA counter.
2917 (cgraph_edge::verify_count_and_frequency): Rename to ...
2918 (cgraph_edge::verify_count): ... this one; drop frequency checking.
2919 (cgraph_node::verify_node): Update.
2920 * cgraph.h (struct cgraph_edge): Drop frequency.
2921 (cgraph_edge::frequency): New function.
2922 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
2924 (cgraph_edge::rebuild_edges): Likewise.
2925 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
2926 (duplicate_thunk_for_node): Do not pass frequency.
2927 (cgraph_node::create_clone): Scale only counts.
2928 (cgraph_node::create_virtual_clone): Do not pass frequency.
2929 (cgraph_node::create_edge_including_clones): Do not pass frequency.
2930 (cgraph_node::create_version_clone): Do not pass frequency.
2931 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
2932 (cgraph_node::expand_thunk): Do not pass frequency.
2933 (cgraph_node::create_wrapper): Do not pass frequency.
2934 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
2936 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
2937 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
2938 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
2939 (gather_caller_stats): Use frequency function.
2940 (ipcp_cloning_candidate_p): Use frequency function.
2941 (ipcp_propagate_stage): Use frequency function.
2942 (get_info_about_necessary_edges): Use frequency function.
2943 (update_profiling_info): Update only IPA profile.
2944 (update_specialized_profile): Use frequency functoin.
2945 (perhaps_add_new_callers): Update only IPA profile.
2946 * ipa-devirt.c (ipa_devirt): Use IPA profile.
2947 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
2948 (dump_ipa_call_summary): Use frequency function.
2949 (estimate_edge_size_and_time): Use frequency function.
2950 (ipa_merge_fn_summary_after_inlining): Use frequency function.
2951 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
2952 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
2953 (update_noncloned_counts): ... ths one; scale counts only.
2954 (clone_inlined_nodes): Do not scale frequency.
2955 (inline_call): Do not pass frequency.
2956 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
2957 (compute_inlined_call_time): Use IPA profile.
2958 (want_inline_small_function_p): Use IPA profile.
2959 (want_inline_self_recursive_call_p): Use IPA profile.
2960 (edge_badness): Use IPA profile.
2961 (lookup_recursive_calls): Use IPA profile.
2962 (recursive_inlining): Do not pass frequency.
2963 (resolve_noninline_speculation): Do not update frequency.
2964 (inline_small_functions): Collect max of IPA profile.
2965 (dump_overall_stats): Dump IPA porfile.
2966 (dump_inline_stats): Dump IPA porfile.
2967 (ipa_inline): Collect IPA stats.
2968 * ipa-inline.h (clone_inlined_nodes): Update prototype.
2969 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
2970 (ipa_propagate_frequency): Use frequency function.
2971 (ipa_profile): Cleanup.
2972 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
2973 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
2974 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
2975 (input_node): Do not stream frequency.
2976 (input_edge): Do not stream frequency.
2977 (merge_profile_summaries): Scale only IPA profiles.
2978 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
2979 * predict.c (drop_profile): Do not recompute frequency.
2980 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
2981 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
2982 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
2983 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
2984 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
2985 * tree-inline.c (copy_bb): Do not scale frequency.
2986 (expand_call_inline): Do not scale frequency.
2987 (tree_function_versioning): Do not scale frequency.
2988 * ubsan.c (ubsan_create_edge): Do not pass frequency.
2990 2017-11-10 Julia Koval <julia.koval@intel.com>
2992 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
2993 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
2994 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
2995 (_mm256_maskz_gf2p8affine_epi64_epi8)
2996 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
2997 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
2998 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
2999 (__builtin_ia32_vgf2p8affineqb_v32qi)
3000 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
3001 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
3003 2017-11-10 Tamar Christina <tamar.christina@arm.com>
3007 (arm_option_override): Refactor.
3008 (arm_option_reconfigure_globals): New.
3009 (arm_options_perform_arch_sanity_checks): New.
3010 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
3012 (arm_options_perform_arch_sanity_checks): Likewise
3014 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
3016 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
3017 (power9-qpmul): New.
3018 * rs6000/rs6000.md ("type" attr): Add qmul.
3019 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
3020 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
3021 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
3023 2017-11-10 Martin Sebor <msebor@redhat.com>
3026 * builtins.c (compute_objsize): Handle arrays that
3027 compute_builtin_object_size likes to fail for. Make extern.
3028 * builtins.h (compute_objsize): Declare.
3029 (check_strncpy_sizes): New function.
3030 (expand_builtin_strncpy): Call check_strncpy_sizes.
3031 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
3032 -Wstringop-truncation.
3033 (gimple_fold_builtin_strncat): Same.
3034 * gimple.c (gimple_build_call_from_tree): Set call location.
3035 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
3036 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
3037 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
3038 (handle_builtin_strlen): Use strlen_to_stridx.
3039 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
3041 Use strlen_to_stridx.
3042 (pass_strlen::execute): Release strlen_to_stridx.
3043 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
3044 (-Wstringop-truncation): Document new option.
3046 2017-11-10 Martin Liska <mliska@suse.cz>
3048 PR gcov-profile/82702
3049 * gcov.c (main): Handle intermediate files in a different
3051 (get_gcov_intermediate_filename): New function.
3052 (output_gcov_file): Remove support of intermediate files.
3053 (generate_results): Allocate intermediate file.
3054 (release_structures): Clean-up properly fn_end.
3055 (output_intermediate_file): Start iterating with line 1.
3057 2017-11-10 Jakub Jelinek <jakub@redhat.com>
3059 PR tree-optimization/82929
3060 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
3061 ops_swapped_p non-static data member.
3062 (store_immediate_info::store_immediate_info): Clear it.
3063 (imm_store_chain_info::coalesce_immediate_stores): If swapping
3064 ops set ops_swapped_p.
3065 (count_multiple_uses): Handle ops_swapped_p.
3067 2017-11-10 Martin Liska <mliska@suse.cz>
3069 * coverage.c (coverage_init): Stream information about
3070 support of has_unexecuted_blocks.
3071 * doc/gcov.texi: Document that.
3072 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
3073 * gcov.c (read_graph_file): Likewise.
3074 (output_line_beginning): Fix a small issue with
3077 2017-11-10 Bin Cheng <bin.cheng@arm.com>
3079 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
3080 reference of trivial component.
3082 2017-11-10 Jakub Jelinek <jakub@redhat.com>
3085 * gimple-ssa-store-merging.c
3086 (pass_store_merging::terminate_all_aliasing_chains): For
3087 gimple_store_p stmts also call refs_output_dependent_p.
3089 PR rtl-optimization/82913
3090 * compare-elim.c (try_merge_compare): Punt if def_insn is not
3093 2017-11-09 Jeff Law <law@redhat.com>
3095 * vr-values.h: New file with vr_values class.
3096 * tree-vrp.c: Include vr-values.h
3097 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
3098 data objects into the vr_values class.
3099 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
3100 (get_value_range): Make it a member function within vr_values class.
3101 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
3102 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
3103 (op_with_constant_singleton_value_range): Likewise.
3104 (extract_range_for_var_from_comparison_expr): Likewise.
3105 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
3106 (extract_range_from_binary_expr): Likewise.
3107 (extract_range_from_unary_expr): Likewise.
3108 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
3109 (check_for_binary_op_overflow, extract_range_basic): Likewise.
3110 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
3111 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
3112 (compare_name_with_value, compare_names): Likewise.
3113 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
3114 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
3115 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
3116 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
3117 (extract_range_from_phi_node): Likewise.
3118 (simplify_truth_ops_using_ranges): Likewise.
3119 (simplify_div_or_mod_using_ranges): Likewise.
3120 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
3121 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
3122 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
3123 (simplify_float_conversion_using_ranges): Likewise.
3124 (simplify_internal_call_using_ranges): Likewise.
3125 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
3126 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
3127 poitner via x_vr_values for calls into gimple folder.
3128 (vrp_initialize_lattice): Make this the vr_values ctor.
3129 (vrp_free_lattice): Make this the vr_values dtor.
3130 (set_vr_value): New function.
3131 (class vrp_prop): Add vr_values data member. Add various member
3132 functions as well as member functions that delegate to vr_values.
3133 (check_array_ref): Make a member function within vrp_prop class.
3134 (search_for_addr_array, vrp_initialize): Likewise.
3135 (vrp_finalize): Likewise. Revamp to avoid direct access to
3136 vr_value, values_propagated, etc.
3137 (check_array_bounds): Extract vrp_prop class instance pointer from
3138 walk info structure. Use it to call member functions.
3139 (check_all_array_refs): Make a member function within vrp_prop class.
3140 Smuggle class instance pointer via walk info structure.
3141 (x_vr_values): New local static.
3142 (vrp_valueize): Use x_vr_values to get class instance.
3143 (vr_valueize_1): Likewise.
3144 (class vrp_folder): Add vr_values data member. Add various member
3145 functions as well as member functions that delegate to vr_values.
3146 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
3147 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
3148 class instance from vr_values. Use it to call member functions.
3149 (vrp_dom_walker): Add vr_values data member.
3150 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
3151 instance via x_vr_values.
3152 (identify_jump_threads): Accept vr_values as argument. Store
3153 it into the walker structure.
3154 (evrp_dom_walker): Add vr_values class data member. Add various
3156 (evrp_dom_walker::try_find_new_range): Use vr_values data
3157 member to access the memory allocator.
3158 (evrp_dom_walker::before_dom_children): Store vr_values class
3159 instance into the vrp_folder class.
3160 (evrp_dom_walker::push_value_range): Rework to avoid direct
3161 access to num_vr_values and vr_value.
3162 (evrp_dom_walker::pop_value_range): Likewise.
3163 (execute_early_vrp): Remove call to vrp_initialize_lattice.
3164 Use vr_values to get to dump_all_value_ranges member function.
3165 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
3166 and simplify_cond_using_ranges_2 via vrp_prop class instance.
3167 Pass vr_values class instance down to identify_jump_threads.
3168 Remove call to vrp_free_lattice.
3169 (debug_all_value_ranges): Remove.
3171 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
3172 (vrp_folder): Likewise.
3174 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
3175 Get it from the existing bitmap instead.
3176 (vrp_intersect_ranges_1): Likewise.
3178 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3180 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
3182 (store_immediate_info::store_immediate_info): Add bitnotp argument,
3183 set bit_not_p to it.
3184 (imm_store_chain_info::coalesce_immediate_stores): Break group
3185 if bit_not_p is different.
3186 (count_multiple_uses, split_group,
3187 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
3188 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
3189 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
3190 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
3191 to store_immediate_info ctor.
3193 2017-11-09 Jim Wilson <jimw@sifive.com>
3195 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
3196 (scan_prog_file): Likewise.
3198 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
3200 * bb-reorder.c (max_entry_frequency): Remove.
3201 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
3202 connect_traces, push_to_next_round_p): Remove prototypes.
3203 (find_traces_1_round): Use counts only.
3204 (push_to_next_round_p): Likewise.
3205 (find_traces): Likewise.
3206 (rotate_loop): Likewise.
3207 (find_traces_1_round): Likewise.
3208 (connect_traces): Likewise.
3209 (edge_order): Likewise.
3211 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
3213 * config/arm/arm.c (output_return_instruction): Add comments to
3214 indicate requirement for cmse_nonsecure_entry return to account
3215 for the size of clearing instruction output here.
3216 (thumb_exit): Likewise.
3217 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
3218 return in hardfloat mode.
3220 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
3222 * config/rs6000/rs6000.c (machine_function): Add a bool,
3223 "toc_is_wrapped_separately".
3224 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
3225 if it wasn't explicitly set or unset, we are optimizing for speed, and
3226 doing separate shrink-wrapping.
3227 (rs6000_get_separate_components): Enable the TOC component if
3228 saving the TOC register in the prologue.
3229 (rs6000_components_for_bb): Handle the TOC component.
3230 (rs6000_emit_prologue_components): Store the TOC register where needed.
3231 (rs6000_set_handled_components): Mark TOC as handled, if handled.
3232 (rs6000_emit_prologue): Don't save the TOC if that is already done.
3234 2017-11-09 Martin Jambor <mjambor@suse.cz>
3236 * ipa-param-manipulation.c: New file.
3237 * ipa-param-manipulation.h: Likewise.
3238 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
3239 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
3240 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
3241 (ipa_parm_adjustment): Likewise.
3242 (ipa_parm_adjustment_vec): Likewise.
3243 (ipa_get_vector_of_formal_parms): Moved declaration to
3244 ipa-param-manipulation.h.
3245 (ipa_get_vector_of_formal_parm_types): Likewise.
3246 (ipa_modify_formal_parameters): Likewise.
3247 (ipa_modify_call_arguments): Likewise.
3248 (ipa_combine_adjustments): Likewise.
3249 (ipa_dump_param_adjustments): Likewise.
3250 (ipa_modify_expr): Likewise.
3251 (ipa_get_adjustment_candidate): Likewise.
3252 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
3253 ipa-param-manipulation.c.
3254 (ipa_get_vector_of_formal_parm_types): Likewise.
3255 (ipa_modify_formal_parameters): Likewise.
3256 (ipa_modify_call_arguments): Likewise.
3257 (ipa_modify_expr): Likewise.
3258 (get_ssa_base_param): Likewise.
3259 (ipa_get_adjustment_candidate): Likewise.
3260 (index_in_adjustments_multiple_times_p): Likewise.
3261 (ipa_combine_adjustments): Likewise.
3262 (ipa_dump_param_adjustments): Likewise.
3263 * tree-sra.c: Also include ipa-param-manipulation.h
3264 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
3267 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3268 Alan Hayward <alan.hayward@arm.com>
3269 David Sherwood <david.sherwood@arm.com>
3271 * doc/sourcebuild.texi (vect_masked_store): Document.
3273 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3274 Alan Hayward <alan.hayward@arm.com>
3275 David Sherwood <david.sherwood@arm.com>
3277 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
3279 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3280 Alan Hayward <alan.hayward@arm.com>
3281 David Sherwood <david.sherwood@arm.com>
3283 * doc/sourcebuild.texi (vect_variable_length): Document.
3285 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3286 Alan Hayward <alan.hayward@arm.com>
3287 David Sherwood <david.sherwood@arm.com>
3289 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
3291 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3292 Alan Hayward <alan.hayward@arm.com>
3293 David Sherwood <david.sherwood@arm.com>
3295 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
3297 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3298 Alan Hayward <alan.hayward@arm.com>
3299 David Sherwood <david.sherwood@arm.com>
3301 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
3302 previously undocumented selectors.
3303 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
3305 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3307 * doc/rtl.texi (const_vector): Say that elements can be
3308 const_wide_ints too.
3309 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
3310 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
3311 (gen_vec_duplicate): Use it instead of CONSTANT_P.
3312 * optabs.c (expand_vector_broadcast): Likewise.
3314 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3315 Alan Hayward <alan.hayward@arm.com>
3316 David Sherwood <david.sherwood@arm.com>
3318 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
3319 scaled index even if the unscaled address was invalid.
3320 Don't increase the complexity of using a scale in that case.
3322 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3323 Alan Hayward <alan.hayward@arm.com>
3324 David Sherwood <david.sherwood@arm.com>
3326 * doc/rtl.texi: Rewrite the subreg rules so that they partition
3327 the inner register into REGMODE_NATURAL_SIZE bytes rather than
3328 UNITS_PER_WORD bytes.
3329 * emit-rtl.c (validate_subreg): Divide subregs into blocks
3330 based on REGMODE_NATURAL_SIZE of the inner mode.
3331 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
3332 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
3333 * expmed.c (lowpart_bit_field_p): Divide the value up into
3334 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
3335 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
3336 whether something is likely to occupy more than one register.
3338 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
3341 * ipa-inline-transform.c (update_noncloned_frequencies): Use
3342 profile_count::adjust_for_ipa_scaling.
3343 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
3344 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
3346 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
3348 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3350 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
3351 (split_group): Add total_orig and total_new arguments, estimate the
3352 number of statements related to the store group without store merging
3353 and with store merging.
3354 (imm_store_chain_info::output_merged_store): Adjust split_group
3355 callers, punt if estimated number of statements with store merging
3356 is not smaller than estimated number of statements without it.
3358 (handled_load): Remove has_single_use checks.
3359 (pass_store_merging::process_store): Likewise.
3361 2017-11-09 Richard Biener <rguenther@suse.de>
3363 PR tree-optimization/82902
3364 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
3366 2017-11-09 Martin Liska <mliska@suse.cz>
3369 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
3372 2017-11-09 Martin Liska <mliska@suse.cz>
3374 PR tree-optimization/82669
3375 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
3377 2017-11-09 Martin Liska <mliska@suse.cz>
3379 PR gcov-profile/48463
3380 * coverage.c (coverage_begin_function): Output also end locus
3381 of a function and information whether the function is
3383 * gcov-dump.c (tag_function): Parse and print the information.
3384 * gcov.c (INCLUDE_MAP): Add include.
3385 (INCLUDE_SET): Likewise.
3386 (struct line_info): Move earlier in the source file because
3387 of vector<line_info> in function_info structure.
3388 (line_info::line_info): Likewise.
3389 (line_info::has_block): Likewise.
3390 (struct source_info): Add new member index.
3391 (source_info::get_functions_at_location): New function.
3392 (function_info::group_line_p): New function.
3393 (output_intermediate_line): New function.
3394 (output_intermediate_file): Use the mentioned function.
3395 (struct function_start): New.
3396 (struct function_start_pair_hash): Likewise.
3397 (process_file): Add code that identifies group functions.
3398 Assign lines either to global or function scope.
3399 (generate_results): Skip artificial functions.
3400 (find_source): Assign index for each source file.
3401 (read_graph_file): Read new flag artificial and end_line.
3402 (add_line_counts): Assign it either to global of function scope.
3403 (accumulate_line_counts): Isolate core of the function to
3404 accumulate_line_info and call it for both function and global
3406 (accumulate_line_info): New function.
3407 (output_line_beginning): Fix GNU coding style.
3408 (print_source_line): New function.
3409 (output_line_details): Likewise.
3410 (output_function_details): Likewise.
3411 (output_lines): Iterate both source (global) scope and function
3413 (struct function_line_start_cmp): New class.
3414 * doc/gcov.texi: Reflect changes in documentation.
3416 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3419 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
3420 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
3421 and similarly for not instead of neg.
3423 2017-11-08 Andi Kleen <ak@linux.intel.com>
3425 * config/i386/i386.opt: Add -mforce-indirect-call.
3426 * config/i386/predicates.md: Check for flag_force_indirect_call.
3427 * doc/invoke.texi: Document -mforce-indirect-call
3429 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
3431 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
3433 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
3434 config/riscv/riscv.c (predict.h): New include.
3435 (riscv_slow_unaligned_access_p): No longer static.
3436 (riscv_block_move_straight): Add require.
3437 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
3439 2017-11-08 Jakub Jelinek <jakub@redhat.com>
3442 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
3443 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
3444 nonimmediate_operand predicate for operand 1 instead of
3447 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3449 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
3451 * config/aarch64/constraints.md (Uml): New constraint.
3452 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
3455 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3457 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
3458 of two vec_duplicates into a vec_concat.
3460 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3462 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
3463 Simplify vec_merge of vec_duplicate and vec_concat.
3464 * config/aarch64/constraints.md (Utq): New constraint.
3465 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
3468 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3470 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
3471 Simplify vec_merge of vec_duplicate and const_vector.
3472 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
3474 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
3475 mode iterator. Update predicate on operand 1 to
3476 handle non-const_vec constants. Delete constraints.
3477 (*aarch64_combinez_be<mode>): Likewise for operand 2.
3479 2017-11-08 Jakub Jelinek <jakub@redhat.com>
3481 PR tree-optimization/78821
3482 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
3484 (store_operand_info::store_operand_info): Initialize it to false.
3485 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
3486 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
3487 store in the group, and if chain_info is non-NULL, to ignore altogether
3489 (compatible_load_p): Fail if bit_not_p does not match.
3490 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
3491 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
3492 (pass_store_merging::process_store): Adjust
3493 terminate_all_aliasing_chains calls to pass NULL in all current spots,
3494 call terminate_all_aliasing_chains newly when adding a store into
3495 a chain with non-NULL chain_info.
3497 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
3499 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
3501 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
3503 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3505 (aarch64_layout_frame): Initialise emit_frame_chain.
3506 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
3507 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
3509 2017-11-08 Martin Liska <mliska@suse.cz>
3511 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
3512 of gimple_call_internal_p.
3514 2017-11-07 Tom de Vries <tom@codesourcery.com>
3516 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
3518 2017-11-07 Tom de Vries <tom@codesourcery.com>
3520 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
3523 2017-11-08 Martin Liska <mliska@suse.cz>
3526 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
3528 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
3530 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
3532 2017-11-07 Jakub Jelinek <jakub@redhat.com>
3535 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
3536 (*cmp<mode>_ccz_1): New insn with $k alternative.
3539 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
3540 RTX_COMM_COMPARE as commutative as well.
3541 (ix86_binary_operator_ok): Formatting fix.
3542 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
3543 *<code><mode>3<mask_name><round_saeonly_name>,
3544 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
3545 *<s>mul<mode>3_highpart<mask_name>,
3546 *vec_widen_umult_even_v16si<mask_name>,
3547 *vec_widen_umult_even_v8si<mask_name>,
3548 *vec_widen_umult_even_v4si<mask_name>,
3549 *vec_widen_smult_even_v16si<mask_name>,
3550 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
3551 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
3552 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
3553 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
3554 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
3555 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
3556 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
3557 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
3558 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
3559 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
3560 ix86_binary_operator_ok. Formatting fixes.
3561 (*<plusminus_insn><mode>3<mask_name><round_name>,
3562 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
3565 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
3567 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
3568 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
3569 eq and ne if TARGET_ISEL.
3570 (cmp): New code_iterator.
3571 (UNS, UNSU_, UNSIK): New code_attrs.
3572 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
3573 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
3575 ("eq<mode>3"): ... this.
3576 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
3578 ("ne<mode>3"): ... this.
3580 2017-11-07 Julia Koval <julia.koval@intel.com>
3583 * common/config/i386/i386-common.c
3584 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
3585 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
3586 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
3587 * config/i386/i386.opt: Ditto.
3588 * config/i386/i386.c (ix86_target_string): Ditto.
3589 (ix86_option_override_internal): Ditto.
3590 (ix86_init_mpx_builtins): Move MPX to args2.
3591 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
3592 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3593 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3594 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3595 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3596 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3597 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
3599 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
3602 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
3603 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
3604 (zero-extendsidi peephole2): Remove peephole.
3606 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
3609 * stor-layout.c: Include attribs.h.
3610 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
3611 explicit lookup of "aligned" attribute.
3613 2017-11-07 Andrew Waterman <andrew@sifive.com>
3615 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
3616 (riscv_expand_block_move): Likewise.
3617 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
3619 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
3620 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
3621 * config/riscv/riscv.c (riscv_block_move_straight): New function.
3622 (riscv_adjust_block_mem): Likewise.
3623 (riscv_block_move_loop): Likewise.
3624 (riscv_expand_block_move): Likewise.
3625 * config/riscv/riscv.md (movmemsi): New pattern.
3627 2017-11-07 Michael Clark <michaeljclark@mac.com>
3629 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
3630 (MUSL_DYNAMIC_LINKER): Likewise.
3632 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
3634 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
3635 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
3638 2017-11-07 Richard Biener <rguenther@suse.de>
3640 * match.pd: Fix build.
3642 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
3643 Jackson Woodruff <jackson.woodruff@arm.com>
3645 PR tree-optimization/71026
3646 * match.pd: Canonicalize negate in division.
3648 2017-11-07 Sudakshina Das <sudi.das@arm.com>
3651 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
3653 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
3655 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
3656 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
3658 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
3660 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
3661 non-scalar integral types.
3662 * match.pd (negate_expr_p): Handle MINUS_EXPR.
3663 (-(A-B), -(~A)): New transformations.
3665 2017-11-07 Tom de Vries <tom@codesourcery.com>
3667 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
3668 semicolon after "do {} while (0)".
3669 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3670 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3671 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3672 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3673 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3675 2017-11-07 Tom de Vries <tom@codesourcery.com>
3677 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
3678 after "do {} while (0)".
3679 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3680 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3681 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3682 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3683 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
3685 2017-11-07 Tom de Vries <tom@codesourcery.com>
3688 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
3690 (arm_rtx_costs_internal): Add missing semicolon after
3691 HANDLE_NARROW_SHIFT_ARITH call.
3693 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
3695 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3696 disable isel if it was not set explicitly.
3698 2017-11-06 James Bowman <james.bowman@ftdichip.com>
3700 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
3701 (add_type_attribute) likewise.
3703 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
3705 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
3706 of struct ix86_frame.
3707 (ix86_initial_elimination_offset): Likewise.
3708 (ix86_expand_split_stack_prologue): Likewise.
3710 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
3712 * tree-vrp.h (enum value_range_type): Update stale comment.
3714 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3715 Alan Hayward <alan.hayward@arm.com>
3716 David Sherwood <david.sherwood@arm.com>
3718 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
3719 (aarch64_expand_vec_perm_const): Take the number of units too.
3720 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
3721 (aarch64_expand_vec_perm_const): Likewise.
3722 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
3723 (vec_perm<mode>): Update accordingly.
3725 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3726 Alan Hayward <alan.hayward@arm.com>
3727 David Sherwood <david.sherwood@arm.com>
3729 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
3730 Take the number of units too.
3731 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
3732 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
3733 but check for a vector mode before rather than after the call.
3734 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
3735 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
3736 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
3737 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
3738 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
3739 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
3740 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
3741 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
3742 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
3743 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
3744 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
3745 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
3746 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
3747 (aarch64_sqdmull2_n<mode>): Update accordingly.
3749 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3750 Alan Hayward <alan.hayward@arm.com>
3751 David Sherwood <david.sherwood@arm.com>
3753 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
3754 the number of units too.
3755 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
3756 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
3757 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
3758 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
3759 (vec_store_lanesxi<mode>): Update accordingly.
3761 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3762 Alan Hayward <alan.hayward@arm.com>
3763 David Sherwood <david.sherwood@arm.com>
3765 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
3766 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
3767 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
3768 of units rather than the mode.
3769 * config/aarch64/iterators.md (nunits): New mode attribute.
3770 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
3771 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
3772 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
3773 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
3774 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
3775 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
3776 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
3777 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
3778 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
3779 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
3780 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3781 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
3782 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
3783 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
3784 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
3785 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
3786 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
3787 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
3788 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
3789 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
3790 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
3791 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
3792 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
3793 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3794 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
3795 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
3796 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3797 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
3798 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
3799 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
3800 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
3801 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
3802 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
3803 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
3804 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
3805 (aarch64_simd_vec_setv2di): Likewise.
3807 2017-11-06 Carl Love <cel@us.ibm.com>
3809 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
3811 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
3812 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
3814 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
3816 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
3817 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
3818 Add power 8 macro expansions.
3819 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
3820 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
3821 power 8 instructions. (VSX_XXBR): Add iterator.
3823 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
3825 * config/arm/arm.md (predicable_short_it): Change default to "no",
3826 improve documentation, remove uses that are identical to the default.
3827 (enabled_for_depr_it): Rename to enabled_for_short_it.
3828 * gcc/config/arm/arm-fixed.md (predicable_short_it):
3829 Remove default uses.
3830 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
3831 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
3832 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
3833 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
3835 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
3838 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
3839 float128 helper macros, which are no longer used after deleting
3840 the old 'q' built-in functions, and moving the round to odd
3841 built-in functions to being special built-in functions.
3842 (BU_FLOAT128_2): Likewise.
3843 (BU_FLOAT128_1_HW): Likewise.
3844 (BU_FLOAT128_2_HW): Likewise.
3845 (BU_FLOAT128_3_HW): Likewise.
3846 (FABSQ): Delete old 'q' built-in functions.
3847 (COPYSIGNQ): Likewise.
3848 (SQRTF128_ODD): Move round to odd built-in functions to be
3849 special built-in functions, so that we can handle
3850 -mabi=ieeelongdouble.
3851 (TRUNCF128_ODD): Likewise.
3852 (ADDF128_ODD): Likewise.
3853 (SUBF128_ODD): Likewise.
3854 (MULF128_ODD): Likewise.
3855 (DIVF128_ODD): Likewise.
3856 (FMAF128_ODD): Likewise.
3857 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
3858 built-in names to 'f128'.
3859 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
3860 old 'q' built-in functions, as the machine independent code for
3861 'f128' built-in functions handles this.
3862 (rs6000_expand_builtin): Add expansion for float128 round to odd
3863 functions, keying off on -mabi=ieeelongdouble of whether to use
3864 the KFmode or TFmode variant.
3865 (rs6000_init_builtins): Initialize the _Float128 round to odd
3867 * doc/extend.texi (PowerPC Built-in Functions): Document the old
3868 _Float128 'q' built-in functions are now mapped into the new
3869 'f128' built-in functions.
3871 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
3873 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
3874 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
3876 2017-11-06 David Malcolm <dmalcolm@redhat.com>
3879 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
3880 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
3881 * toplev.c: Include "ipa-fnsummary.h".
3882 (toplev::finalize): Call ipa_fnsummary_c_finalize.
3884 2017-11-06 Jakub Jelinek <jakub@redhat.com>
3886 PR tree-optimization/82838
3887 * gimple-ssa-store-merging.c
3888 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
3889 on a separate gimple_seq which is then appended to seq.
3891 2017-11-06 Jeff Law <law@redhat.com>
3894 * config/i386/i386.c (PROBE_INTERVAL): Remove.
3895 (get_probe_interval): New functions.
3896 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
3897 (ix86_adjust_stack_and_probe): Likewise.
3898 (output_adjust_stack_and_probe): Likewise.
3899 (ix86_emit_probe_stack_range): Likewise.
3900 (ix86_expand_prologue): Likewise.
3902 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3904 PR tree-optimization/82816
3905 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
3906 if the modes of the two types are the same.
3907 (convert_plusminus_to_widen): Likewise.
3909 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3911 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
3913 (usadv16qi): New define_expand.
3914 (usadv8hi): New define_expand.
3916 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
3919 * ipa-inline-transform.c (update_noncloned_frequencies): Always
3921 (inline_transform): Likewise.
3922 * predict.c (counts_to_freqs): Remove useless conditional.
3923 * profile-count.h (profile_count::apply_scale): Move sanity check.
3924 * tree-inline.c (copy_bb): Always scale.
3925 (copy_cfg_body): Likewise.
3927 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
3930 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
3932 (*cmp_ite0): Add enabled_for_depr_it attribute.
3933 (*cmp_ite1): Likewise.
3935 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
3937 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
3940 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
3942 * tree-vrp.c (vrp_int_const_binop): Return true on success and
3943 return the value by pointer.
3944 (extract_range_from_multiplicative_op_1): Update accordingly.
3945 Return as soon as an operation fails.
3947 2017-11-05 Tom de Vries <tom@codesourcery.com>
3950 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
3951 (DEF_SANITIZER_BUILTIN): ... here.
3952 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
3953 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
3955 2017-11-05 Tom de Vries <tom@codesourcery.com>
3958 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
3960 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
3961 ASM_OUTPUT_BEFORE_CASE_LABEL call.
3962 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
3964 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
3965 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
3966 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
3968 2017-11-05 Tom de Vries <tom@codesourcery.com>
3971 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
3974 2017-11-04 Michael Clark <michaeljclark@mac.com>
3976 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
3977 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
3979 (*addsi3_extended): Likewise.
3980 (*addsi3_extended2): Likewise.
3981 (<optab>si3): Likewise.
3982 (<optab>di3): Likewise.
3983 (<optab><mode>3): Likewise.
3984 (<*optabe>si3_internal): Likewise.
3985 (zero_extendqi<SUPERQI:mode>2): Likewise.
3986 (*add<mode>hi3): Likewise.
3987 (*xor<mode>hi3): Likewise.
3988 (<optab>di3): Likewise.
3989 (*<optab>si3_extend): Likewise.
3990 (*sge<u>_<X:mode><GPR:mode>): Likewise.
3991 (*slt<u>_<X:mode><GPR:mode>): Likewise.
3992 (*sle<u>_<X:mode><GPR:mode>): Likewise.
3994 2017-11-04 Andrew Waterman <andrew@sifive.com>
3996 * config/riscv/riscv.c (riscv_option_override): Conditionally set
3997 TARGET_STRICT_ALIGN based upon -mtune argument.
3999 2017-11-04 Andrew Waterman <andrew@sifive.com>
4001 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
4003 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
4005 * config/i386/i386.c (choose_basereg): Use optional scratch
4006 register and add assertion.
4007 (x86_emit_outlined_ms2sysv_save): Use scratch register when
4008 needed, and don't allocate stack.
4009 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
4010 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
4011 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
4013 2017-11-03 Jeff Law <law@redhat.com>
4015 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
4016 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
4017 to probe at the start of a noreturn function.
4019 2017-11-03 Jakub Jelinek <jakub@redhat.com>
4021 PR tree-optimization/78821
4022 * gimple-ssa-store-merging.c: Update the file comment.
4023 (MAX_STORE_ALIAS_CHECKS): Define.
4024 (struct store_operand_info): New type.
4025 (store_operand_info::store_operand_info): New constructor.
4026 (struct store_immediate_info): Add rhs_code and ops data members.
4027 (store_immediate_info::store_immediate_info): Add rhscode, op0r
4028 and op1r arguments to the ctor, initialize corresponding data members.
4029 (struct merged_store_group): Add load_align_base and load_align
4031 (merged_store_group::merged_store_group): Initialize them.
4032 (merged_store_group::do_merge): Update them.
4033 (merged_store_group::apply_stores): Pick the constant for
4034 encode_tree_to_bitpos from one of the two operands, or skip
4035 encode_tree_to_bitpos if neither operand is a constant.
4036 (class pass_store_merging): Add process_store method decl. Remove
4037 bool argument from terminate_all_aliasing_chains method decl.
4038 (pass_store_merging::terminate_all_aliasing_chains): Remove
4039 var_offset_p argument and corresponding handling.
4040 (stmts_may_clobber_ref_p): New function.
4041 (compatible_load_p): New function.
4042 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
4043 if there is overlap and rhs_code is not INTEGER_CST. For
4044 non-overlapping stores terminate group if rhs is not mergeable.
4045 (get_alias_type_for_stmts): Change first argument from
4046 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
4047 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
4048 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
4050 (get_location_for_stmts): Change first argument from
4051 auto_vec<gimple *> & to vec<gimple *> &.
4052 (struct split_store): Remove orig_stmts data member, add orig_stores.
4053 (split_store::split_store): Create orig_stores rather than orig_stmts.
4054 (find_constituent_stmts): Renamed to ...
4055 (find_constituent_stores): ... this. Change second argument from
4056 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
4057 to info structures rather than the statements.
4058 (split_group): Rename ALLOW_UNALIGNED argument to
4059 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
4060 it. Adjust find_constituent_stores caller.
4061 (imm_store_chain_info::output_merged_store): Handle rhs_code other
4062 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
4063 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
4064 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
4065 (mem_valid_for_store_merging): New function.
4066 (handled_load): New function.
4067 (pass_store_merging::process_store): New method.
4068 (pass_store_merging::execute): Use process_store method. Adjust
4069 terminate_all_aliasing_chains caller.
4071 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
4073 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
4074 Return true for more constants, symbols and label references.
4075 (aarch64_valid_floating_const): Remove unused function.
4077 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
4080 * config/aarch64/aarch64.c (aarch64_layout_frame):
4081 Undo forcing of LR at bottom of frame.
4083 2017-11-03 Jeff Law <law@redhat.com>
4086 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
4087 for int_registers_saved.
4089 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
4090 extracted from tree-ssa-dom.c.
4091 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
4092 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
4093 (record_equivalences_from_incoming_edge): Add additional argument
4094 to single_pred_edge_ignoring_loop_edges call.
4095 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
4096 (uncprop_dom_walker::before_dom_children): Add additional argument
4097 to single_pred_edge_ignoring_loop_edges call.
4098 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
4099 single_pred_edge_ignoring_loop_edges rather than open coding.
4100 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
4102 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
4104 * match.pd (-(-A)): Rewrite.
4106 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
4108 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
4109 (rs6000_emit_int_cmove): New declaration.
4110 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
4111 (rs6000_emit_sISEL): Delete.
4112 (rs6000_emit_int_cmove): Make non-static.
4113 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
4114 instead of rs6000_emit_sISEL.
4116 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
4118 * asan.c (create_cond_insert_point): Maintain profile.
4119 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
4121 * basic-block.h (struct basic_block_def): Remove frequency.
4122 (EDGE_FREQUENCY): Use to_frequency
4123 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
4125 (find_traces): Update to use to_frequency.
4126 (find_traces_1_round): Likewise; use only IPA counts.
4127 (bb_to_key): Likewise.
4128 (connect_traces): Use IPA counts only.
4129 (copy_bb_p): Update to use to_frequency.
4130 (fix_up_crossing_landing_pad): Likewise.
4131 (sanitize_hot_paths): Likewise.
4132 * bt-load.c (basic_block_freq): Likewise.
4133 * cfg.c (init_flow): Set count_max to uninitialized.
4134 (check_bb_profile): Remove frequencies; check counts.
4135 (dump_bb_info): Do not dump frequencies.
4136 (update_bb_profile_for_threading): Update counts only.
4137 (scale_bbs_frequencies_int): Likewise.
4138 (MAX_SAFE_MULTIPLIER): Remove.
4139 (scale_bbs_frequencies_gcov_type): Update counts only.
4140 (scale_bbs_frequencies_profile_count): Update counts only.
4141 (scale_bbs_frequencies): Update counts only.
4142 * cfg.h (struct control_flow_graph): Add count-max.
4143 (update_bb_profile_for_threading): Update prototype.
4144 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
4145 (find_many_sub_basic_blocks): Likewise.
4146 * cfgcleanup.c (try_forward_edges): Likewise.
4147 (try_crossjump_to_edge): Likewise.
4148 * cfgexpand.c (expand_gimple_cond): Likewise.
4149 (expand_gimple_tailcall): Likewise.
4150 (construct_init_block): Likewise.
4151 (construct_exit_block): Likewise.
4152 * cfghooks.c (verify_flow_info): Check consistency of counts.
4153 (dump_bb_for_graph): Do not dump frequencies.
4154 (split_block_1): Do not update frequencies.
4155 (split_edge): Do not update frequencies.
4156 (make_forwarder_block): Do not update frequencies.
4157 (duplicate_block): Do not update frequencies.
4158 (account_profile_record): Do not update frequencies.
4159 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
4160 for global heuristics.
4161 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
4162 (expected_loop_iterations_unbounded): Use counts only.
4163 * cfgloopmanip.c (scale_loop_profile): Simplify.
4164 (create_empty_loop_on_edge): Simplify
4166 (duplicate_loop_to_header_edge): Simplify
4167 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
4168 (update_br_prob_note): Take care of removing note when profile
4170 (relink_block_chain): Do not dump frequency.
4171 (rtl_account_profile_record): Use to_frequency.
4172 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
4173 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
4174 (cgraph_update_edges_for_call_stmt_node): Likewise.
4175 (cgraph_edge::verify_count_and_frequency): Update.
4176 (cgraph_node::verify_node): Temporarily disable frequency verification.
4177 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
4178 to_cgraph_frequency.
4179 (cgraph_edge::rebuild_edges): Convert to ipa counts.
4180 * cgraphunit.c (init_lowered_empty_function): Do not initialize
4182 (cgraph_node::expand_thunk): Update profile.
4183 * except.c (dw2_build_landing_pads): Do not update frequency.
4184 * final.c (compute_alignments): Use to_frequency.
4185 (dump_basic_block_info): Do not dump frequency.
4186 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
4187 (dump_gimple_bb_header): Do not dump frequency.
4188 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
4190 * gimple-streamer-in.c (input_bb): Do not stream frequency.
4191 * gimple-streamer-out.c (output_bb): Do not stream frequency.
4192 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
4193 (init_before_recovery): Do not update frequency.
4194 (sched_create_recovery_edges): Do not update frequency.
4195 * hsa-gen.c (convert_switch_statements): Do not update frequency.
4196 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
4197 (ipa_cp_c_finalize): Set max_count to uninitialized.
4198 * ipa-fnsummary.c (get_minimal_bb): Use counts.
4199 (param_change_prob): Use counts.
4200 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
4202 * ipa-split.c (consider_split): Use to_frequency.
4203 (split_function): Use to_frequency.
4204 * ira-build.c (loop_compare_func): Likewise.
4205 (mark_loops_for_removal): Likewise.
4206 (mark_all_loops_for_removal): Likewise.
4207 * loop-doloop.c (doloop_modify): Do not update frequency.
4208 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
4210 * lto-streamer-in.c (input_function): Update count_max.
4211 * omp-expand.c (expand_omp_taskreg): Update count_max.
4212 * omp-simd-clone.c (simd_clone_adjust): Update profile.
4213 * predict.c (maybe_hot_frequency_p): Use to_frequency.
4214 (maybe_hot_count_p): Use ipa counts only.
4215 (maybe_hot_bb_p): Simplify.
4216 (maybe_hot_edge_p): Simplify.
4217 (probably_never_executed): Do not take frequency argument.
4218 (probably_never_executed_bb_p): Do not pass frequency.
4219 (probably_never_executed_edge_p): Likewise.
4220 (combine_predictions_for_bb): Check that profile is nonzero.
4221 (propagate_freq): Do not set frequency.
4222 (drop_profile): Simplify.
4223 (counts_to_freqs): Simplify.
4224 (expensive_function_p): Use to_frequency.
4225 (propagate_unlikely_bbs_forward): Simplify.
4226 (determine_unlikely_bbs): Simplify.
4227 (estimate_bb_frequencies): Add hack to silence graphite issues.
4228 (compute_function_frequency): Use ipa counts.
4229 (pass_profile::execute): Update.
4230 (rebuild_frequencies): Use counts only.
4231 (force_edge_cold): Use counts only.
4232 * profile-count.c (profile_count::dump): Dump new count types.
4233 (profile_count::differs_from_p): Check compatiblity.
4234 (profile_count::to_frequency): New function.
4235 (profile_count::to_cgraph_frequency): New function.
4236 * profile-count.h (struct function): Declare.
4237 (enum profile_quality): Add profile_guessed_local and
4238 profile_guessed_global0.
4239 (class profile_proability): Decrease number of bits to 29;
4240 update from_reg_br_prob_note and to_reg_br_prob_note.
4241 (class profile_count: Update comment; decrease number of bits
4242 to 61. Check compatibility.
4243 (profile_count::compatible_p): New private member function.
4244 (profile_count::ipa_p): New member function.
4245 (profile_count::operator<): Handle global zero correctly.
4246 (profile_count::operator>): Handle global zero correctly.
4247 (profile_count::operator<=): Handle global zero correctly.
4248 (profile_count::operator>=): Handle global zero correctly.
4249 (profile_count::nonzero_p): New member function.
4250 (profile_count::force_nonzero): New member function.
4251 (profile_count::max): New member function.
4252 (profile_count::apply_scale): Handle IPA scalling.
4253 (profile_count::guessed_local): New member function.
4254 (profile_count::global0): New member function.
4255 (profile_count::ipa): New member function.
4256 (profile_count::to_frequency): Declare.
4257 (profile_count::to_cgraph_frequency): Declare.
4258 * profile.c (OVERLAP_BASE): Delete.
4259 (compute_frequency_overlap): Delete.
4260 (compute_branch_probabilities): Do not use compute_frequency_overlap.
4261 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
4262 * sched-ebb.c (rank): Use counts only.
4263 * shrink-wrap.c (handle_simple_exit): Use counts only.
4264 (try_shrink_wrapping): Use counts only.
4265 (place_prologue_for_one_component): Use counts only.
4266 * tracer.c (find_best_predecessor): Use to_frequency.
4267 (find_trace): Use to_frequency.
4268 (tail_duplicate): Use to_frequency.
4269 * trans-mem.c (expand_transaction): Do not update frequency.
4270 * tree-call-cdce.c: Do not update frequency.
4271 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
4272 (gimple_merge_blocks): Likewise.
4273 (gimple_split_edge): Likewise.
4274 (gimple_duplicate_sese_region): Likewise.
4275 (gimple_duplicate_sese_tail): Likewise.
4276 (move_sese_region_to_fn): Likewise.
4277 (gimple_account_profile_record): Likewise.
4278 (insert_cond_bb): Likewise.
4279 * tree-complex.c (expand_complex_div_wide): Likewise.
4280 * tree-eh.c (lower_resx): Update profile.
4281 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
4283 (initialize_cfun): Do not initialize frequencies
4284 (freqs_to_counts): Delete.
4285 (copy_cfg_body): Ignore count parameter.
4286 (copy_body): Update.
4287 (expand_call_inline): Update count_max.
4288 (optimize_inline_calls): Update count_max.
4289 (tree_function_versioning): Update count_max.
4290 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
4291 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
4293 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
4294 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
4295 (try_peel_loop): Likewise.
4296 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
4298 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
4299 (tree_transform_and_unroll_loop): Do not use frequencies
4300 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
4301 Use reliable prediction only.
4302 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
4303 * tree-ssa-sink.c (select_best_block): Use to_frequency.
4304 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
4305 probability scaling.
4306 * tree-ssa-threadupdate.c (create_block_for_threading): Do
4307 not update frequency
4308 (any_remaining_duplicated_blocks): Likewise.
4309 (update_profile): Likewise.
4310 (estimated_freqs_path): Delete.
4311 (freqs_to_counts_path): Delete.
4312 (clear_counts_path): Delete.
4313 (ssa_fix_duplicate_block_edges): Likewise.
4314 (duplicate_thread_path): Likewise.
4315 * tree-switch-conversion.c (gen_inbound_check): Use counts.
4316 * tree-tailcall.c (decrease_profile): Do not update frequency.
4317 (eliminate_tail_call): Likewise.
4318 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
4319 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
4320 (optimize_mask_stores): Likewise.
4321 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
4322 * ubsan.c (ubsan_expand_null_ifn): Update profile.
4323 (ubsan_expand_ptr_ifn): Update profile.
4324 * value-prof.c (gimple_ic): Simplify.
4325 * value-prof.h (gimple_ic): Update prototype.
4326 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
4327 * ipa-inline.c (compute_uninlined_call_time): Be sure that
4329 (want_inline_self_recursive_call_p): Likewise.
4330 (resolve_noninline_speculation): Only cummulate defined counts.
4331 (inline_small_functions): Use nonzero_p.
4332 (ipa_inline): Do not access freed node.
4334 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
4336 * config/aarch64/aarch64.c (aarch64_override_options_internal):
4337 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
4339 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
4341 * config/riscv/riscv.c (riscv_legitimize_move): Handle
4342 non-legitimate address.
4344 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
4346 * config/rs6000/rs6000.md (*lt0_disi): Delete.
4347 (*lt0_<mode>di, *lt0_<mode>si): New.
4349 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
4351 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
4352 TARGET_PAIRED_FLOAT.
4354 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
4355 Jim Wilson <jim.wilson@linaro.org>
4357 * config/aarch64/aarch64-cores.def (saphira): New CPU.
4358 * config/aarch64/aarch64-tune.md: Regenerated.
4359 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
4360 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
4362 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
4364 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
4365 (arc_expand_prologue): Restore blink for millicode.
4366 * config/arc/linux.h (LINK_EH_SPEC): Defined.
4368 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
4371 * config/i386/i386.c (ix86_vector_duplicate_value): Use
4372 gen_vec_duplicate after forcing the scalar into a register.
4374 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
4376 * combine (try_combine): Print the insns input to try_combine to the
4379 2017-11-02 Steve Ellcey <sellcey@cavium.com>
4382 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
4383 Remove second argument from aarch64_process_target_attr call.
4384 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
4386 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
4388 (aarch64_handle_attr_arch): Remove second argument.
4389 (aarch64_handle_attr_cpu): Ditto.
4390 (aarch64_handle_attr_tune): Ditto.
4391 (aarch64_handle_attr_isa_flags): Ditto.
4392 (aarch64_process_one_target_attr): Ditto.
4393 (aarch64_process_target_attr): Ditto.
4394 (aarch64_option_valid_attribute_p): Remove second argument.
4395 on aarch64_process_target_attr call.
4397 2017-11-02 David Malcolm <dmalcolm@redhat.com>
4399 * diagnostic.c: Include "selftest-diagnostic.h".
4400 (selftest::assert_location_text): New function.
4401 (selftest::test_diagnostic_get_location_text): New function.
4402 (selftest::diagnostic_c_tests): Call it.
4404 2017-11-02 David Malcolm <dmalcolm@redhat.com>
4406 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
4407 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
4408 (class selftest::test_diagnostic_context): Move to...
4409 * selftest-diagnostic.c: New file.
4410 * selftest-diagnostic.h: New file.
4412 2017-11-02 James Bowman <james.bowman@ftdichip.com>
4414 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
4415 offset range for FT32B.
4416 * config/ft32/ft32.h: option "mcompress" enables relaxation.
4417 * config/ft32/ft32.md: Add TARGET_NOPM.
4418 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
4419 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
4421 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
4423 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
4425 2017-11-02 Jeff Law <law@redhat.com>
4427 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
4428 virtual keyword on FINAL OVERRIDE members.
4430 * tree-ssa-propagate.h (ssa_propagation_engine): Group
4431 virtuals together. Add virtual destructor.
4432 (substitute_and_fold_engine): Similarly.
4434 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
4436 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
4438 2017-11-02 Richard Biener <rguenther@suse.de>
4440 PR tree-optimization/82795
4441 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
4443 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4445 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
4447 (gcc_GAS_CHECK_FEATURE): Remove.
4448 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
4449 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
4450 * configure: Regenerate.
4452 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
4454 * config/arc/arc.c (hwloop_optimize): Account for empty
4457 2017-11-02 Richard Biener <rguenther@suse.de>
4460 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
4461 Truncate ARRAY_REF index and element size.
4463 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
4465 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
4467 2017-11-01 Jeff Law <law@redhat.com>
4469 * tree-ssa-ccp.c (ccp_folder): New class derived from
4470 substitute_and_fold_engine.
4471 (ccp_folder::get_value): New member function.
4472 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
4473 (ccp_fold_stmt): Remove prototype.
4474 (ccp_finalize): Call substitute_and_fold from the ccp_class.
4475 * tree-ssa-copy.c (copy_folder): New class derived from
4476 substitute_and_fold_engine.
4477 (copy_folder::get_value): Renamed from get_value.
4478 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
4479 * tree-vrp.c (vrp_folder): New class derived from
4480 substitute_and_fold_engine.
4481 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
4482 (vrp_folder::get_value): New member function.
4483 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
4484 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
4485 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
4486 provide a class interface to folder/substitute routines.
4487 (ssa_prop_fold_stmt_fn): Remove typedef.
4488 (ssa_prop_get_value_fn): Likewise.
4489 (subsitute_and_fold): Remove prototype.
4490 (replace_uses_in): Likewise.
4491 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
4492 Renamed from replace_uses_in. Call the virtual member function
4493 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
4494 (substitute_and_fold_dom_walker): Remove initialization of
4495 data member entries for calbacks. Add substitute_and_fold_engine
4496 member and initialize it.
4497 (substitute_and_fold_dom_walker::before_dom_children0: Use the
4498 member functions for get_value, replace_phi_args_in c
4499 replace_uses_in, and fold_stmt calls.
4500 (substitute_and_fold_engine::substitute_and_fold): Renamed from
4501 substitute_and_fold. Remove assert. Update ctor call.
4503 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
4504 (ssa_prop_visit_phi_fn): Likewise.
4505 (class ssa_propagation_engine): New class to provide an interface
4507 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
4509 (ssa_prop_visit_phi): Likewise.
4510 (ssa_propagation_engine::simulate_stmt): Moved into class.
4511 Call visit_phi/visit_stmt from the class rather than via
4512 file scoped static variables.
4513 (ssa_propagation_engine::simulate_block): Moved into class.
4514 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
4515 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
4516 set file scoped statics for the visit_stmt/visit_phi callbacks.
4517 * tree-complex.c (complex_propagate): New class derived from
4518 ssa_propagation_engine.
4519 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
4520 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
4521 (tree_lower_complex): Call ssa_propagate via the complex_propagate
4523 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
4524 ssa_propagation_engine.
4525 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
4526 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
4527 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
4528 * tree-ssa-copy.c (copy_prop): New class derived from
4529 ssa_propagation_engine.
4530 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
4531 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
4532 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
4533 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
4534 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
4535 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
4536 (execute_vrp): Call ssa_propagate from the vrp_prop class.
4538 2017-11-01 Jakub Jelinek <jakub@redhat.com>
4540 PR rtl-optimization/82778
4541 PR rtl-optimization/82597
4542 * compare-elim.c (struct comparison): Add in_a_setter field.
4543 (find_comparison_dom_walker::before_dom_children): Remove killed
4544 bitmap and df_simulate_find_defs call, instead walk the defs.
4545 Compute last_setter and initialize in_a_setter. Merge definitions
4546 with first initialization for a few variables.
4547 (try_validate_parallel): Use insn_invalid_p instead of
4548 recog_memoized. Return insn rather than just the pattern.
4549 (try_merge_compare): Fix up comment. Don't uselessly test if
4550 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
4552 (execute_compare_elim_after_reload): Remove df_chain_add_problem
4555 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4556 Alan Hayward <alan.hayward@arm.com>
4557 David Sherwood <david.sherwood@arm.com>
4559 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
4560 aarch64_hard_regno_nregs to get the number of registers
4563 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4564 Alan Hayward <alan.hayward@arm.com>
4565 David Sherwood <david.sherwood@arm.com>
4567 * config/aarch64/constraints.md (Upl): Rename to...
4569 * config/aarch64/aarch64.md
4570 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
4573 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4574 Alan Hayward <alan.hayward@arm.com>
4575 David Sherwood <david.sherwood@arm.com>
4577 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
4578 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
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 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
4586 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
4587 (aarch64_evpc_dup): Generate rtl direcly, rather than using
4589 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
4590 of a single element.
4591 * config/aarch64/iterators.md: Add a comment above the permute
4592 unspecs to say that they are generated directly by
4593 aarch64_expand_vec_perm_const.
4594 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
4596 2017-11-01 Nathan Sidwell <nathan@acm.org>
4598 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
4600 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
4602 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
4603 and medany code models, and describe what they do.
4605 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4607 Revert accidental duplicate:
4609 * combine.c (can_change_dest_mode): Reject changes in
4610 REGMODE_NATURAL_SIZE.
4612 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
4614 PR rtl-optimization/64682
4615 PR rtl-optimization/69567
4616 PR rtl-optimization/69737
4617 PR rtl-optimization/82683
4618 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
4619 register mentioned in the note, drop the note, unless it came from I3,
4620 in which case it should go to I3 again.
4622 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4624 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
4625 and return false if not.
4626 (clear_bytes_written_by, live_bytes_read): Update accordingly.
4628 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4630 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
4631 range is known to be empty.
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 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
4638 and CONST_VECTOR_NUNITS instead of computing the number of units from
4639 the byte sizes of the vector and element.
4640 (simplify_binary_operation_1): Likewise.
4641 (simplify_const_binary_operation): Likewise.
4642 (simplify_ternary_operation): Likewise.
4644 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4645 Alan Hayward <alan.hayward@arm.com>
4646 David Sherwood <david.sherwood@arm.com>
4648 * var-tracking.c (INT_MEM_OFFSET): Replace with...
4649 (int_mem_offset): ...this new function.
4650 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
4651 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
4652 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
4655 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4656 Alan Hayward <alan.hayward@arm.com>
4657 David Sherwood <david.sherwood@arm.com>
4659 * lower-subreg.c (interesting_mode_p): New function.
4660 (compute_costs, find_decomposable_subregs, decompose_register)
4661 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
4662 (resolve_clobber, dump_choices): Use it.
4664 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4665 Alan Hayward <alan.hayward@arm.com>
4666 David Sherwood <david.sherwood@arm.com>
4668 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
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 * alias.c (find_base_value, find_base_term): Only process integer
4675 truncations. Check the precision rather than the size.
4677 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4678 Alan Hayward <alan.hayward@arm.com>
4679 David Sherwood <david.sherwood@arm.com>
4681 * machmode.h (is_narrower_int_mode): New function
4682 * optabs.c (expand_float, expand_fix): Use it.
4683 * dwarf2out.c (rotate_loc_descriptor): Likewise.
4685 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4686 Alan Hayward <alan.hayward@arm.com>
4687 David Sherwood <david.sherwood@arm.com>
4689 * rtl.h (narrower_subreg_mode): New function.
4690 * ira-color.c (update_costs_from_allocno): Use it.
4692 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4693 Alan Hayward <alan.hayward@arm.com>
4694 David Sherwood <david.sherwood@arm.com>
4696 * optabs-query.h (convert_optab_p): New function, split out from...
4697 (convert_optab_handler): ...here.
4698 (widening_optab_handler): Delete.
4699 (find_widening_optab_handler): Remove permit_non_widening parameter.
4700 (find_widening_optab_handler_and_mode): Likewise. Provide an
4701 override that operates on mode class wrappers.
4702 * optabs-query.c (widening_optab_handler): Delete.
4703 (find_widening_optab_handler_and_mode): Remove permit_non_widening
4704 parameter. Assert that the two modes are the same class and that
4705 the "from" mode is narrower than the "to" mode. Use
4706 convert_optab_handler instead of widening_optab_handler.
4707 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
4708 instead of widening_optab_handler.
4709 * expr.c (expand_expr_real_2): Update calls to
4710 find_widening_optab_handler.
4711 * optabs.c (expand_widen_pattern_expr): Likewise.
4712 (expand_binop_directly): Take the insn_code as a parameter.
4713 (expand_binop): Only call find_widening_optab_handler for
4714 conversion optabs; use optab_handler otherwise. Update calls
4715 to find_widening_optab_handler and expand_binop_directly.
4716 Use convert_optab_handler instead of widening_optab_handler.
4717 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
4718 find_widening_optab_handler and use scalar_mode rather than
4720 (convert_plusminus_to_widen): Likewise.
4722 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4723 Alan Hayward <alan.hayward@arm.com>
4724 David Sherwood <david.sherwood@arm.com>
4726 * machmode.h (fixed_size_mode): New class.
4727 * rtl.h (get_pool_mode): Return fixed_size_mode.
4728 * gengtype.c (main): Add fixed_size_mode.
4729 * target.def (get_raw_result_mode): Return a fixed_size_mode.
4730 (get_raw_arg_mode): Likewise.
4731 * doc/tm.texi: Regenerate.
4732 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
4733 * targhooks.c (default_get_reg_raw_mode): Likewise.
4734 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
4735 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
4736 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
4737 (msp430_get_raw_result_mode): Likewise.
4738 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
4739 * dbxout.c (dbxout_parms): Require fixed-size modes.
4740 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
4741 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
4742 * omp-low.c (lower_oacc_reductions): Likewise.
4743 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
4744 (simplify_subreg): Update accordingly.
4745 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
4746 (force_const_mem): Update accordingly. Return NULL_RTX for modes
4747 that aren't fixed-size.
4748 (get_pool_mode): Return a fixed_size_mode.
4749 (output_constant_pool_2): Take a fixed_size_mode.
4751 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4752 Alan Hayward <alan.hayward@arm.com>
4753 David Sherwood <david.sherwood@arm.com>
4755 * doc/rtl.texi (vec_series): Document.
4756 (const): Say that the operand can be a vec_series.
4757 * rtl.def (VEC_SERIES): New rtx code.
4758 * rtl.h (const_vec_series_p_1): Declare.
4759 (const_vec_series_p): New function.
4760 * emit-rtl.h (gen_const_vec_series): Declare.
4761 (gen_vec_series): Likewise.
4762 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
4763 (gen_vec_series): Likewise.
4764 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
4765 * simplify-rtx.c (simplify_unary_operation): Handle negations
4767 (simplify_binary_operation_series): New function.
4768 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
4769 (test_vector_ops_series): New function.
4770 (test_vector_ops): Call it.
4771 * config/powerpcspe/altivec.md (altivec_lvsl): Use
4772 gen_const_vec_series.
4773 (altivec_lvsr): Likewise.
4774 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
4776 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4777 Alan Hayward <alan.hayward@arm.com>
4778 David Sherwood <david.sherwood@arm.com>
4780 * doc/rtl.texi (const): Update description of address constants.
4781 Say that vector constants are allowed too.
4782 * common.md (E, F): Use CONSTANT_P instead of checking for
4784 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
4785 checking for CONST_VECTOR.
4786 * expmed.c (make_tree): Use build_vector_from_val for a CONST
4788 * expr.c (expand_expr_real_2): Check for vector modes instead
4789 of checking for CONST_VECTOR.
4790 * rtl.h (const_vec_p): New function.
4791 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
4792 (unwrap_const_vec_duplicate): Handle them here too.
4794 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4795 David Malcolm <dmalcolm@redhat.com>
4796 Alan Hayward <alan.hayward@arm.com>
4797 David Sherwood <david.sherwood@arm.com>
4799 * rtl.h (vec_duplicate_p): New function.
4800 * selftest-rtl.c (assert_rtx_eq_at): New function.
4801 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
4802 (assert_rtx_eq_at): Declare.
4803 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
4804 * selftest-run-tests.c (selftest::run_tests): Call it.
4805 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
4806 (simplify_unary_operation_1): Recursively handle vector duplicates.
4807 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
4809 (simplify_subreg): Handle subregs of vector duplicates.
4810 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
4811 (selftest::simplify_rtx_c_tests): New functions.
4813 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4814 Alan Hayward <alan.hayward@arm.com>
4815 David Sherwood <david.sherwood@arm.com>
4817 * emit-rtl.h (gen_const_vec_duplicate): Declare.
4818 (gen_vec_duplicate): Likewise.
4819 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
4821 (gen_const_vector): ...here.
4822 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
4823 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
4824 whose elements are all equal.
4825 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
4826 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
4827 (simplify_relational_operation): Likewise.
4828 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
4830 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
4831 (aarch64_expand_vector_init): Likewise.
4832 * config/arm/arm.c (neon_vdup_constant): Likewise.
4833 (neon_expand_vector_init): Likewise.
4834 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
4835 (arm_block_set_unaligned_vect): Likewise.
4836 (arm_block_set_aligned_vect): Likewise.
4837 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
4838 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
4839 (expand_vec_perm_even_odd_pack): Likewise.
4840 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
4841 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
4842 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
4843 gen_const_vec_duplicate.
4844 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
4845 * config/mips/mips.c (mips_gen_const_int_vector): Use
4846 gen_const_vec_duplicate.
4847 (mips_expand_vector_init): Use CONST0_RTX.
4848 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
4849 (define_split): Use gen_const_vec_duplicate.
4850 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
4851 (define_split): Use gen_const_vec_duplicate.
4852 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
4853 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
4854 * config/spu/spu.c (spu_const): Likewise.
4856 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
4857 Alan Hayward <alan.hayward@arm.com>
4858 David Sherwood <david.sherwood@arm.com>
4860 * combine.c (can_change_dest_mode): Reject changes in
4861 REGMODE_NATURAL_SIZE.
4863 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
4865 * configure.ac (--enable-libssp): New.
4866 (gcc_cv_libc_provides_ssp): Check for explicit setting before
4867 trying to determine target-specific default. Adjust indentation.
4868 * configure: Regenerated.
4869 * doc/install.texi (Configuration): Expand --disable-libssp
4872 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
4874 config/i386/i386.c (ix86_expand_epilogue): Correct stack
4877 2017-10-31 Martin Jambor <mjambor@suse.cz>
4880 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
4882 2017-10-31 David Malcolm <dmalcolm@redhat.com>
4884 * auto-profile.c (autofdo_source_profile::read): Use
4885 UNKNOWN_LOCATION rather than 0.
4886 * diagnostic-core.h (warning_at_rich_loc): Rename to...
4887 (warning_at): ...this overload.
4888 (warning_at_rich_loc_n): Rename to...
4889 (warning_n): ...this overload.
4890 (error_at_rich_loc): Rename to...
4891 (error_at): ...this overload.
4892 (pedwarn_at_rich_loc): Rename to...
4893 (pedwarn): ...this overload.
4894 (permerror_at_rich_loc): Rename to...
4895 (permerror): ...this overload.
4896 (inform_at_rich_loc): Rename to...
4897 (inform): ...this overload.
4898 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
4899 (diagnostic_n_impl_richloc): Rename to...
4900 (diagnostic_n_impl): ...this rich_location *-based decl.
4901 (inform_at_rich_loc): Rename to...
4902 (inform): ...this, and add an assertion.
4903 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
4904 (warning_at_rich_loc): Rename to...
4905 (warning_at): ...this, and add an assertion.
4906 (warning_at_rich_loc_n): Rename to...
4907 (warning_n): ...this, and add an assertion.
4908 (warning_n): Update location_t-based implementation for removal of
4909 location_t-based diagnostic_n_impl.
4910 (pedwarn_at_rich_loc): Rename to...
4911 (pedwarn): ...this, and add an assertion.
4912 (permerror_at_rich_loc): Rename to...
4913 (permerror): ...this, and add an assertion.
4914 (error_n): Update for removal of location_t-based diagnostic_n_impl.
4915 (error_at_rich_loc): Rename to...
4916 (error_at): ...this, and add an assertion.
4917 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
4918 (driver::do_spec_on_infiles): Likewise.
4919 * substring-locations.c (format_warning_va): Update for renaming
4920 of inform_at_rich_loc.
4922 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
4924 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
4925 _Float<N>X built-in functions so that the variant without the
4926 "__builtin_" prefix is only enabled for the GNU C and Objective C
4927 languages when they are in non-strict ANSI/ISO mode.
4928 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
4929 * target.def (floatn_builtin_p): Add a target hook to control
4930 whether _Float<N> and _Float<N>X built-in functions without the
4931 "__builtin_" prefix are enabled, and return true for C and
4932 Objective C in the default hook. Include langhooks.h in
4934 * targhooks.h (default_floatn_builtin_p): Likewise.
4935 * targhooks.c (default_floatn_builtin_p): Likewise.
4936 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
4937 floatn_builtin_p target hook.
4938 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
4940 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
4941 Eric Botcazou <ebotcazou@adacore.com>
4943 PR rtl-optimization/81803
4944 * lra-constraints.c (curr_insn_transform): Also reload the whole
4945 register for a strict subreg no wider than a word if this is for
4946 a WORD_REGISTER_OPERATIONS target.
4948 2017-10-31 Jason Merrill <jason@redhat.com>
4950 * gdbinit.in: Skip over inlines from timevar.h.
4952 2017-10-31 Martin Liska <mliska@suse.cz>
4954 * doc/gcov.texi: Document new option.
4955 * gcov.c (print_usage): Likewise print it.
4956 (process_args): Support the argument.
4957 (format_count): New function.
4958 (format_gcov): Use the function.
4960 2017-10-31 Martin Liska <mliska@suse.cz>
4962 * gcov.c (struct name_map): do not use typedef.
4963 Define operator== and operator<.
4964 (name_search): Remove.
4965 (name_sort): Remove.
4966 (main): Do not allocate names.
4967 (process_file): Add vertical space.
4968 (generate_results): Use std::find.
4969 (release_structures): Do not release memory.
4970 (find_source): Use std::find.
4972 2017-10-31 Martin Liska <mliska@suse.cz>
4974 * gcov.c (struct line_info): Remove it's typedef.
4975 (line_info::line_info): Add proper ctor.
4976 (line_info::has_block): Do not use a typedef.
4977 (struct source_info): Do not use typedef.
4978 (circuit): Likewise.
4979 (get_cycles_count): Likewise.
4980 (output_intermediate_file): Iterate via vector iterator.
4981 (add_line_counts): Use std::vector methods.
4982 (accumulate_line_counts): Likewise.
4983 (output_lines): Likewise.
4985 2017-10-31 Martin Liska <mliska@suse.cz>
4987 * gcov.c (struct source_info): Remove typedef.
4988 (source_info::source_info): Add proper ctor.
4989 (accumulate_line_counts): Use struct, not it's typedef.
4990 (output_gcov_file): Likewise.
4991 (output_lines): Likewise.
4992 (main): Do not allocate an array.
4993 (output_intermediate_file): Use size of vector container.
4994 (process_file): Resize the vector.
4995 (generate_results): Do not preallocate, use newly added vector
4997 (release_structures): Do not release sources.
4998 (find_source): Use vector methods.
4999 (add_line_counts): Do not use typedef.
5001 2017-10-31 Martin Liska <mliska@suse.cz>
5003 * doc/gcov.texi: Document that.
5004 * gcov.c (add_line_counts): Mark lines with a non-executed
5006 (output_line_beginning): Handle such lines.
5007 (output_lines): Pass new argument.
5008 (output_intermediate_file): Print it in intermediate format.
5010 2017-10-31 Martin Liska <mliska@suse.cz>
5012 * color-macros.h: New file.
5013 * diagnostic-color.c: Factor out color related to macros to
5015 * doc/gcov.texi: Document -k option.
5016 * gcov.c (INCLUDE_STRING): Include string.h.
5017 (print_usage): Add -k option.
5018 (process_args): Parse it.
5019 (pad_count_string): New function.
5020 (output_line_beginning): Likewise.
5021 (DEFAULT_LINE_START): New macro.
5022 (output_lines): Support color output.
5024 2017-10-31 Martin Liska <mliska@suse.cz>
5026 PR gcov-profile/82633
5027 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
5028 their interaction with GCOV infrastructure.
5029 * configure.ac: Add -fkeep-{inline,static}-functions to
5031 * configure: Regenerate.
5033 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
5036 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
5038 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
5041 * config/rs6000/rs6000.md (allocate_stack): Force update interval
5042 into a register if it does not fit into an immediate offset field.
5044 2017-10-31 Olivier Hainque <hainque@adacore.com>
5046 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
5048 2017-10-31 Julia Koval <julia.koval@intel.com>
5050 * config.gcc: Add gfniintrin.h.
5051 * config/i386/gfniintrin.h: New.
5052 * config/i386/i386-builtin-types.def
5053 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
5054 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
5055 __builtin_ia32_vgf2p8affineinvqb_v32qi,
5056 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
5057 __builtin_ia32_vgf2p8affineinvqb_v16qi,
5058 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
5059 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
5060 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
5061 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
5062 V64QI_FTYPE_V64QI_V64QI_INT): New types.
5063 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5064 * config/i386/immintrin.h: Include gfniintrin.h.
5065 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
5067 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
5069 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
5071 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
5073 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
5074 (arm_ashldi3_1bit): Remove pattern.
5075 (ashrdi3): Remove shift by 1 expansion.
5076 (arm_ashrdi3_1bit): Remove pattern.
5077 (lshrdi3): Remove shift by 1 expansion.
5078 (arm_lshrdi3_1bit): Remove pattern.
5079 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
5080 cost of ashldi3 by 1.
5081 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
5082 (<shift>di3_neon): Likewise.
5084 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5086 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
5087 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
5088 and (*aarch64_simd_mov<VQ:mode>).
5089 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
5090 pattern alternative.
5091 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
5092 attributes to match pattern alternative.
5094 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
5096 * config.gcc (powerpc*-*-*): Add emmintrin.h.
5097 * config/rs6000/emmintrin.h: New file.
5098 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
5100 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
5102 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
5103 * (movdi_vfp_cortexa8): Remove pattern.
5105 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5107 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
5109 (Specific, alpha*-dec-osf5.1): Remove.
5110 (Specific, mips-sgi-irix5): Remove.
5111 (Specific, mips-sgi-irix6): Remove.
5113 2017-10-30 Jakub Jelinek <jakub@redhat.com>
5116 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
5117 arguments to clear_bit_region_be.
5119 2017-10-30 Jim Wilson <wilson@tuliptree.org>
5121 * gimplify.c: Include memmodel.h.
5123 2017-10-30 Martin Jambor <mjambor@suse.cz>
5125 * omp-grid.c (grid_attempt_target_gridification): Also insert a
5126 condition whether loop should be executed at all.
5128 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
5130 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
5131 gimple folding of vec_madd() intrinsics.
5132 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
5133 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
5134 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
5136 2017-10-30 Richard Biener <rguenther@suse.de>
5138 PR tree-optimization/82762
5140 2017-10-23 Richard Biener <rguenther@suse.de>
5142 PR tree-optimization/82129
5144 2017-08-01 Richard Biener <rguenther@suse.de>
5146 PR tree-optimization/81181
5147 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5148 (compute_antic): ... end of iteration here.
5150 2017-10-30 Joseph Myers <joseph@codesourcery.com>
5152 * doc/invoke.texi (C Dialect Options): Document -std=c17,
5153 -std=iso9899:2017 and -std=gnu17.
5154 * doc/standards.texi (C Language): Document C17 support.
5155 * doc/cpp.texi (Overview): Mention -std=c17.
5156 (Standard Predefined Macros): Document C11 and C17 values of
5157 __STDC_VERSION__. Do not refer to C99 support as incomplete.
5158 * doc/extend.texi (Inline): Do not list individual options for
5159 standards newer than C99.
5160 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
5162 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
5165 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
5167 * asan.c (asan_finish_file): Align asan globals array by shadow
5170 2017-10-30 Jakub Jelinek <jakub@redhat.com>
5173 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
5174 (struct store_immediate_info): Add bitregion_start and bitregion_end
5176 (store_immediate_info::store_immediate_info): Add brs and bre
5177 arguments and initialize bitregion_{start,end} from those.
5178 (struct merged_store_group): Add bitregion_start, bitregion_end,
5179 align_base and mask fields. Drop unnecessary struct keyword from
5180 struct store_immediate_info. Add do_merge method.
5181 (clear_bit_region_be): Use memset instead of loop storing zeros.
5182 (merged_store_group::do_merge): New method.
5183 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
5184 stores as long as the surrounding bitregions have no gaps.
5185 (merged_store_group::merge_overlapping): Use do_merge.
5186 (merged_store_group::apply_stores): Test that bitregion_{start,end}
5187 is byte aligned, rather than requiring that start and width are
5188 byte aligned. Drop unnecessary struct keyword from
5189 struct store_immediate_info. Allocate and populate also mask array.
5190 Make start of the arrays relative to bitregion_start rather than
5191 start and size them according to bitregion_{end,start} difference.
5192 (struct imm_store_chain_info): Drop unnecessary struct keyword from
5193 struct store_immediate_info.
5194 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
5195 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
5196 struct keyword from struct store_immediate_info.
5197 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
5198 between stores as long as the surrounding bitregions have no gaps.
5200 (struct split_store): Add orig non-static data member.
5201 (split_store::split_store): Initialize orig to false.
5202 (find_constituent_stmts): Return store_immediate_info *, non-NULL
5203 if there is exactly a single original stmt. Change stmts argument
5204 to pointer from reference, if NULL, don't push anything to it. Add
5205 first argument, use it to optimize skipping over orig stmts that
5206 are known to be before bitpos already. Simplify.
5207 (split_group): Return unsigned int count how many stores are or
5208 would be needed rather than a bool. Add allow_unaligned argument.
5209 Change split_stores argument from reference to pointer, if NULL,
5210 only do a dry run computing how many stores would be produced.
5211 Rewritten algorithm to use both alignment and misalign if
5212 !allow_unaligned and handle bitfield stores with gaps.
5213 (imm_store_chain_info::output_merged_store): Set start_byte_pos
5214 from bitregion_start instead of start. Compute allow_unaligned
5215 here, if true, do 2 split_group dry runs to compute which one
5216 produces fewer stores and prefer aligned if equal. Punt if
5217 new count is bigger or equal than original before emitting any
5218 statements, rather than during that. Remove no longer needed
5219 new_ssa_names tracking. Replace num_stmts with
5220 split_stores.length (). Use 32-bit stack allocated entries
5221 in split_stores auto_vec. Try to reuse original store lhs/rhs1
5222 if possible. Handle bitfields with gaps.
5223 (pass_store_merging::execute): Ignore bitsize == 0 stores.
5224 Compute bitregion_{start,end} for the stores and construct
5225 store_immediate_info with that. Formatting fixes.
5227 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
5230 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
5231 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
5233 2017-10-29 Jim Wilson <wilson@tuliptree.org>
5235 * gimplify.c: Include tm_p.h.
5237 * common.opt (gcoff): Re-add as ignored option.
5238 (gcoff1, gcoff2, gcoff3): Likewise.
5240 * Makefile.in (OBJS): Delete sdbout.o.
5241 (GTFILES): Delete $(srcdir)/sdbout.c.
5242 * debug.h: Delete sdb_debug_hooks.
5243 * final.c: Delete sdbout.h include.
5244 (final_scan_insn): Delete SDB_DEBUG check.
5245 (rest_of_clean_state): Likewise.
5246 * output.h: Delete sdb_begin_function_line.
5249 * toplev.c: Delete sdbout.h include.
5250 (process_options): Delete SDB_DEBUG check.
5251 * tree-core.h (tree_type_common): Delete pointer field of
5253 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
5254 TYPE_SYMTAB_POINTER.
5255 * tree.h (TYPE_SYMTAB_POINTER): Delete.
5256 (TYPE_SYMTAB_IS_POINTER): Delete.
5257 (TYPE_SYMTAB_IS_DIE): Renumber.
5258 * xcoffout.c: Refer to former sdbout.c file.
5259 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
5261 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
5262 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
5263 (Debugging Options): Delete -gcoff.
5264 (-gstabs): Delete SDB reference.
5266 (-gcoff@var{level}): Delete.
5267 * doc/passes.texi (Debugging information output): Delete SDB and
5268 sdbout.c references.
5269 * doc/tm.texi: Regenerate.
5270 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
5271 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
5273 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
5274 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
5275 (SDB_DEBUGGING_INFO): Delete.
5276 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
5277 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
5278 * target.def (output_source_filename): Delete COFF reference.
5280 * common.opt (gcoff): Delete.
5281 (gxcoff+): Update Negative chain.
5282 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
5284 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
5285 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
5286 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
5288 (expand_function_start): Change sdb reference to past tense.
5289 (expand_function_end): Change sdb reference to past tense.
5290 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
5291 * opts.c (debug_type_names): Delete coff entry.
5292 (common_handle_option): Delete OPT_gcoff case.
5293 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
5295 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
5296 * config/cris/cris.h: Delete SDB reference in comment.
5297 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
5298 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
5299 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
5300 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
5302 (ix86_expand_prologue): Likewise.
5303 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
5304 * config/ia64/ia64.h: Likewise.
5305 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
5306 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
5308 * config/mmix/mmix.h: Likewise.
5309 * config/nds32/nds32.c: Likewise.
5310 * config/stormy/storym16.h: Likewise.
5311 * config/visium/visium.h: Likewise.
5312 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
5314 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
5316 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
5317 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
5318 FRAME_POINTER_REGNUM point at high end of local var area.
5320 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
5322 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
5323 Move comment around. Do not reset best_edge for a copiable
5324 destination if the copy would cause a partition change.
5325 (better_edge_p): Remove redundant check.
5327 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
5329 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
5331 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5333 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
5334 for math functions that have _Float<N> and _Float<N>X variants.
5335 (mathfn_built_in_2): Add support for math functions that have
5336 _Float<N> and _Float<N>X variants.
5337 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
5338 (expand_builtin_mathfn_ternary): Add support for fma with
5339 _Float<N> and _Float<N>X variants.
5340 (expand_builtin): Likewise.
5341 (fold_builtin_3): Likewise.
5342 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
5343 create math function _Float<N> and _Float<N>X variants as external
5345 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
5346 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
5347 the __builtin_ prefix and if not strict ansi, without the prefix.
5348 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
5349 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
5350 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
5351 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
5352 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
5353 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
5354 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
5355 function signatures for fma _Float<N> and _Float<N>X variants.
5356 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
5357 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
5358 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
5359 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
5360 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
5361 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
5362 * gencfn-macros.c (print_case_cfn): Add support for math functions
5363 that have _Float<N> and _Float<N>X variants.
5364 (print_define_operator_list): Likewise.
5365 (fltfn_suffixes): Likewise.
5367 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
5368 for math functions that have _Float<N> and _Float<N>X variants.
5369 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
5370 and _Float<N>X variants.
5371 (COPYSIGN): Likewise.
5374 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
5375 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
5377 (integer_valued_read_call_p): Likewise.
5378 * fold-const-call.c (fold_const_call_ss): Likewise.
5379 (fold_const_call_sss): Add support for copysign, fmin, and fmax
5380 _Float<N> and _Float<N>X variants.
5381 (fold_const_call_ssss): Add support for fma _Float<N> and
5382 _Float<N>X variants.
5383 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
5384 support for copysign and fma _Float<N> and _Float<N>X variants.
5385 (backprop::process_builtin_call_use): Likewise.
5386 * tree-call-cdce.c (can_test_argument_range); Add support for
5387 sqrt _Float<N> and _Float<N>X variants.
5388 (edom_only_function): Likewise.
5389 (get_no_error_domain): Likewise.
5390 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
5391 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
5392 copysign _Float<N> and _Float<N>X variants.
5393 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
5394 handled by machine independent code.
5395 (FMAF128): Likewise.
5396 * doc/cpp.texi (Common Predefined Macros): Document defining
5397 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
5398 fma _Float<N> and _Float<N>X variants.
5400 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
5403 * config/i386/i386-modes.def (CCFPU): Remove definition.
5404 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
5405 (ix86_cc_modes_compatible): Ditto.
5406 (ix86_expand_carry_flag_compare): Ditto.
5407 (ix86_expand_int_movcc): Ditto.
5408 (ix86_expand_int_addcc): Ditto.
5409 (ix86_reverse_condition): Ditto.
5410 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
5411 Return true/false for unordered/ordered fp comparisons.
5412 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
5413 (ix86_prepare_fp_compare_args): Update for rename.
5414 (ix86_expand_fp_compare): Update for rename. Generate unordered
5415 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
5416 (ix86_expand_sse_compare_and_jump): Ditto.
5417 * config/i386/predicates.md (fcmov_comparison_operator):
5418 Remove CCFPU mode handling.
5419 (ix86_comparison_operator): Ditto.
5420 (ix86_carry_flag_operator): Ditto.
5421 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
5422 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
5423 (*cmpu<mode>_cc_i387): Ditto.
5424 (FPCMP): Remove mode iterator.
5425 (unord): Remove mode attribute.
5426 (unord_subst): New define_subst transformation
5427 (unord): New define_subst attribute.
5429 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
5430 (*cmpi<unord>xf_i387): Ditto.
5431 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
5432 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
5433 using unord_subst transformation.
5434 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
5435 (round_saeonly): Also handle CCFP mode.
5436 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
5437 Remove UNSPEC_SAHF unspec handling.
5439 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
5441 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
5443 2017-10-27 Jeff Law <law@redhat.com>
5445 * gimple-ssa-sprintf.c: Include domwalk.h.
5446 (class sprintf_dom_walker): New class, derived from dom_walker.
5447 (sprintf_dom_walker::before_dom_children): New function.
5448 (struct call_info): Moved into sprintf_dom_walker class
5449 (compute_formath_length, handle_gimple_call): Likewise.
5450 (sprintf_length::execute): Call the dominator walker rather
5451 than walking the statements.
5453 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
5454 gimple statement locations.
5455 (check_array_bounds): Corresponding changes. Get the statement's
5456 location directly from wi->stmt.
5458 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
5461 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
5463 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
5465 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
5466 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
5468 2017-10-27 Jakub Jelinek <jakub@redhat.com>
5471 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
5472 * config/i386/i386.c (maybe_get_pool_constant): Removed.
5473 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
5474 maybe_get_pool_constant.
5475 * config/i386/predicates.md (zero_extended_scalar_load_operand):
5478 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5480 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
5481 2.26 caveat. Update gas and gld versions.
5482 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
5485 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5487 * cgraph.h (set_malloc_flag): Declare.
5488 * cgraph.c (set_malloc_flag_1): New function.
5489 (set_malloc_flag): Likewise.
5490 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
5491 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
5493 (read_ipa_call_summary): Add support for reading is_return_callee.
5494 (write_ipa_call_summary): Stream is_return_callee.
5495 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
5496 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
5497 ipa-prop.h, ipa-fnsummary.h.
5498 (pure_const_names): Change to static.
5499 (malloc_state_e): Define.
5500 (malloc_state_names): Define.
5501 (funct_state_d): Add field malloc_state.
5502 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
5503 (check_retval_uses): New function.
5504 (malloc_candidate_p): Likewise.
5505 (analyze_function): Add support for malloc attribute.
5506 (pure_const_write_summary): Stream malloc_state.
5507 (pure_const_read_summary): Add support for reading malloc_state.
5508 (dump_malloc_lattice): New function.
5509 (propagate_malloc): New function.
5510 (warn_function_malloc): New function.
5511 (ipa_pure_const::execute): Call propagate_malloc and
5512 ipa_free_fn_summary.
5513 (pass_local_pure_const::execute): Add support for malloc attribute.
5514 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
5515 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
5517 2017-10-27 Martin Liska <mliska@suse.cz>
5519 PR gcov-profile/82457
5520 * doc/invoke.texi: Document that one needs a non-strict ISO mode
5521 for fork-like functions to be properly instrumented.
5523 2017-10-27 Richard Biener <rguenther@suse.de>
5526 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
5527 info when we redirected EH.
5529 2017-10-26 Michael Collison <michael.collison@arm.com>
5531 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
5533 (<optab>_trunchf<GPI:mode>2: New pattern.
5534 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
5535 * config/aarch64/iterators.md (wv): New mode attribute.
5536 (vf, VF): New mode attributes.
5537 (vgp, VGP): New mode attributes.
5538 (s): Update attribute with SImode and DImode prefixes.
5540 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
5542 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
5543 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
5544 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
5545 (nios2_option_overide): Initialize it. Don't allow R0-relative
5546 addressing with PIC.
5547 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
5548 (nios2_symbolic_constant_p): Likewise.
5549 (nios2_legitimate_address_p): Likewise.
5550 (nios2_r0rel_section_name_p): New.
5551 (nios2_symbol_ref_in_r0rel_data_p): New.
5552 (nios2_emit_move_sequence): Handle r0rel_constant_p.
5553 (r0rel_constant_p): New.
5554 (nios2_print_operand_address): Handle r0rel_constant_p.
5555 (nios2_cdx_narrow_form_p): Likewise.
5556 * config/nios2/nios2.opt (mr0rel-sec=): New option.
5557 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
5558 (Nios II Options): Document -mr0rel-sec.
5560 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
5562 * config/nios2/nios2.c: Include xregex.h.
5563 (nios2_gprel_sec_regex): New.
5564 (nios2_option_overide): Initialize it. Don't allow GP-relative
5565 addressing with PIC.
5566 (nios2_small_section_name_p): Check for regex match.
5567 * config/nios2/nios2.opt (mgprel-sec=): New option.
5568 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
5569 (Nios II Options): Document -mgprel-sec.
5571 2017-10-26 Jim Wilson <wilson@tuliptree.org>
5573 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
5575 2017-10-26 Tom de Vries <tom@codesourcery.com>
5577 PR tree-optimization/82707
5578 * gimple.c (gimple_copy): Fix unsharing of
5579 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
5581 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
5583 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
5584 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
5585 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
5586 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
5587 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
5588 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
5589 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
5590 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
5591 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
5592 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
5593 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
5594 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
5595 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
5596 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
5597 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
5598 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
5599 _mm512_mask_cmpunord_ps_mask): New intrinsics.
5601 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5603 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
5605 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
5606 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
5607 warning to rs6000.c. Remove the Undocumented flag, since it has
5609 (-mabi=ibmlongdouble): Likewise.
5610 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
5611 already set, set the default format for long double.
5612 (rs6000_debug_reg_global): Print whether long double is IBM or
5614 (rs6000_option_override_internal): Rework setting long double
5615 format. Only warn if the user is changing the long double default
5616 and they did not use -Wno-psabi.
5617 * doc/invoke.texi (PowerPC options): Update the documentation for
5618 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
5620 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5621 Alan Hayward <alan.hayward@arm.com>
5622 David Sherwood <david.sherwood@arm.com>
5624 * rtl.h (wider_subreg_mode): New function.
5625 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
5626 rather than an unsigned int *.
5627 * ira-color.c (regno_max_ref_width): Replace with...
5628 (regno_max_ref_mode): ...this new variable.
5629 (coalesced_pseudo_reg_slot_compare): Update accordingly.
5630 Use wider_subreg_mode.
5631 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
5632 rather than an unsigned int *.
5633 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
5634 (process_alt_operands): Likewise.
5635 (invariant_p): Likewise.
5636 * lra-spills.c (assign_mem_slot): Likewise.
5637 (add_pseudo_to_slot): Likewise.
5638 * lra.c (collect_non_operand_hard_regs): Likewise.
5639 (add_regs_to_insn_regno_info): Likewise.
5640 * reload1.c (regno_max_ref_width): Replace with...
5641 (regno_max_ref_mode): ...this new variable.
5642 (reload): Update accordingly. Update call to
5643 ira_sort_regnos_for_alter_reg.
5644 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
5645 (init_eliminable_invariants): Update to use regno_max_ref_mode.
5646 (scan_paradoxical_subregs): Likewise.
5648 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
5650 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
5651 (aarch64_frame): Add emit_frame_chain boolean.
5652 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
5653 Move eh_return case to aarch64_layout_frame.
5654 (aarch64_layout_frame): Initialize emit_frame_chain.
5655 (aarch64_expand_prologue): Use emit_frame_chain.
5657 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
5659 * config/aarch64/aarch64.c (aarch64_layout_frame):
5660 Ensure LR is always stored at the bottom of the callee-saves.
5661 Remove rarely used frame layout which saves callee-saves at top of
5662 frame, so the store of LR can be used as a valid probe in all cases.
5664 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
5666 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
5667 Improve unaligned TImode/TFmode base/offset split.
5669 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5670 Alan Hayward <alan.hayward@arm.com>
5671 David Sherwood <david.sherwood@arm.com>
5673 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
5674 * combine.c (find_single_use_1): Likewise.
5675 (expand_field_assignment): Likewise.
5676 (move_deaths): Likewise.
5677 * lra-constraints.c (simplify_operand_subreg): Likewise.
5678 (curr_insn_transform): Likewise.
5679 * lra.c (collect_non_operand_hard_regs): Likewise.
5680 (add_regs_to_insn_regno_info): Likewise.
5681 * rtlanal.c (reg_referenced_p): Likewise.
5682 (covers_regno_no_parallel_p): Likewise.
5684 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5686 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
5687 Don't print any bits outside the precision of the value.
5688 * wide-int.cc (test_printing): Add some new tests.
5690 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5692 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
5693 supports -xbrace_comment option.
5694 * configure: Regenerate.
5695 * config.in: Regenerate.
5696 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
5697 (ASM_CPU_SPEC): Use it.
5699 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
5701 * target.def (static_rtx_alignment): New hook.
5702 * targhooks.h (default_static_rtx_alignment): Declare.
5703 * targhooks.c (default_static_rtx_alignment): New function.
5704 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
5705 * doc/tm.texi: Regenerate.
5706 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
5707 instead of targetm.constant_alignment. Remove call to
5709 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5710 (cris_preferred_mininum_alignment): New function, split out from...
5711 (cris_constant_alignment): ...here.
5712 (cris_static_rtx_alignment): New function.
5713 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
5715 (ix86_constant_alignment): ...here.
5716 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5717 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5718 (mmix_static_rtx_alignment): New function.
5719 * config/spu/spu.c (spu_static_rtx_alignment): New function.
5720 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
5722 2017-10-26 Tamar Christina <tamar.christina@arm.com>
5725 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
5726 Add flag_trapping_math and flag_fp_int_builtin_inexact.
5728 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
5730 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
5731 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
5732 mark as a sign-extending load.
5733 (local_pic_load_u): Define.
5735 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
5738 * fold-const.c (operand_equal_for_comparison_p): Also return true
5739 if ARG0 is a simple variant of ARG1 with narrower precision.
5740 (fold_ternary_loc): Always pass unstripped operands to the predicate.
5742 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
5744 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
5746 * i386.h (processor_costs): Add gather_static, gather_per_elt,
5747 scatter_static, scatter_per_elt.
5748 * x86-tune-costs.h: Add new cost entries.
5750 2017-10-25 Richard Biener <rguenther@suse.de>
5752 * tree-ssa-sccvn.h (vn_eliminate): Declare.
5753 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
5754 class pass_fre): Move to ...
5755 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
5756 class pass_fre): ... here and adjust for statistics.
5758 2017-10-25 Jakub Jelinek <jakub@redhat.com>
5761 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
5762 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
5763 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
5764 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
5767 2017-10-25 Richard Biener <rguenther@suse.de>
5769 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
5770 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
5771 eliminate_push_avail, eliminate_insert): Move inside...
5772 (class eliminate_dom_walker): ... this class in preparation
5774 (fini_eliminate): Remove by merging with ...
5775 (eliminate): ... this function. Adjust for class changes.
5776 (pass_pre::execute): Remove fini_eliminate call.
5777 (pass_fre::execute): Likewise.
5779 2017-10-24 Jakub Jelinek <jakub@redhat.com>
5782 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
5783 (VPERMI2, VPERMI2I): New mode iterators.
5784 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
5785 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
5787 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
5788 mode iterator. Remove 3 old define_insn patterns.
5789 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
5790 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
5791 VPERMI2 mode iterator, remove the other two expanders.
5792 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
5793 to use VPERMI2 mode iterator, add another alternative for vpermi2*
5794 instructions, remove the other two patterns.
5795 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
5796 mode iterator, remove the other two patterns.
5797 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
5798 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
5799 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
5800 and adjust argument order accordingly.
5801 (ix86_expand_vec_perm): Adjust caller.
5802 (expand_vec_perm_1): Likewise.
5803 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
5804 (expand_vec_perm_vpermt2_vpshub2): ... this.
5805 (ix86_expand_vec_perm_const_1): Adjust caller.
5806 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
5809 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
5810 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
5811 (vec_shl_<mode>): Remove unused expander.
5812 (avx512bw_<shift_insn><mode>3): New define_insn.
5813 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
5814 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
5816 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
5819 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
5822 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
5824 PR rtl-optimization/82396
5825 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
5826 (autopref_multipass_init): Simplify initialization.
5827 (autopref_rank_data): Simplify sort order.
5828 * gcc/sched-int.h (autopref_multipass_data_): Remove
5829 multi_mem_insn_p, min_offset and max_offset.
5831 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
5834 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
5835 Check special value of flag_omit_frame_pointer.
5836 (aarch64_can_eliminate): Likewise.
5837 (aarch64_override_options_after_change_1): Simplify handling of
5838 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
5840 2017-10-24 Richard Biener <rguenther@suse.de>
5842 PR tree-optimization/82697
5843 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
5844 zero for conditional load and unconditional store.
5846 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
5848 * doc/install.texi: Document bootstrap-cet.
5850 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
5853 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
5854 ENDBR instruction at function entrance if function is only
5857 2017-10-24 Jakub Jelinek <jakub@redhat.com>
5860 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
5861 patterns to better describe from which operation the CF is computed.
5862 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
5863 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
5864 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
5865 is 0, use _0 suffixed expanders instead of emitting a comparison
5868 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
5870 * config/i386/i386.md(*movsf_internal, *movdf_internal):
5871 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
5873 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
5876 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
5877 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
5878 * loop-iv.c (iv_get_reaching_def): Likewise.
5879 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
5880 variable is promoted and the partition contains undefined values.
5882 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5884 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
5886 (nios2_address_cost): Define.
5887 (nios2_legitimize_address): Recognize (exp + constant) directly.
5888 (TARGET_ADDRESS_COST): Define.
5890 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5892 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
5893 (nios2_symbolic_memory_operand_p): Declare.
5894 (nios2_split_large_constant): Declare.
5895 (nios2_split_symbolic_memory_operand): Declare.
5896 * config/nios2/nios2.c: Adjust includes.
5897 (nios2_symbolic_constant_allowed): New.
5898 (nios2_symbolic_constant_p): New.
5899 (nios2_plus_symbolic_constant_p): New.
5900 (nios2_valid_addr_expr_p): Recognize addresses involving
5902 (nios2_legitimate_address_p): Likewise, also LO_SUM.
5903 (nios2_symbolic_memory_operand_p): New.
5904 (nios2_large_constant_p): New.
5905 (nios2_split_large_constant): New.
5906 (nios2_split_plus_large_constant): New.
5907 (nios2_split_symbolic_memory_operand): New.
5908 (nios2_legitimize_address): Code refactoring. Handle addresses
5909 involving symbolic constants.
5910 (nios2_emit_move_sequence): Likewise.
5911 (nios2_print_operand): Improve error output.
5912 (nios2_print_operand_address): Handle LO_SUM.
5913 (nios2_cdx_narrow_form_p): Likewise.
5914 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
5915 operands involving symbolic constants.
5916 (movhi_internal, movsi_internal): Likewise.
5917 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
5918 (extendhisi2, extendqi<mode>2): Likewise.
5920 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5922 * tree-pass.h (PROP_rtl_split_insns): Define.
5923 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
5925 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
5927 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
5929 2017-10-23 Jakub Jelinek <jakub@redhat.com>
5932 * target.def (const_not_ok_for_debug_p): Default to
5933 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
5934 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
5935 * targhooks.c (default_const_not_ok_for_debug_p): New function.
5936 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
5937 which targetm.const_not_ok_for_debug_p returned true.
5938 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
5940 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
5942 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
5943 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
5944 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
5946 (ix86_const_not_ok_for_debug_p): New function.
5947 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
5948 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
5950 2017-10-23 David Malcolm <dmalcolm@redhat.com>
5953 * system.h: Conditionally include "unique-ptr.h" if
5954 INCLUDE_UNIQUE_PTR is defined.
5955 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
5956 of defining INCLUDE_UNIQUE_PTR before including "system.h".
5958 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
5960 * config/rl78/rl78.md: New define_expand "subdi3".
5962 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
5965 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
5966 DF_REF_INSN if DF_REF_INSN_INFO is false.
5968 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
5970 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
5971 xmm_move instead of sse_move.
5972 (sse_store_index): New function.
5973 (ix86_register_move_cost): Be more sensible about mismatch stall;
5974 model AVX moves correctly; make difference between sse->integer and
5976 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
5977 moves; make difference between SSE and AVX.
5978 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
5979 and zmm_move. Increase size of sse load and store tables;
5980 add unaligned load and store tables; add ssemmx_to_integer.
5981 * x86-tune-costs.h: Update all entries according to real
5982 move latencies from Agner Fog's manual and chip documentation.
5984 2017-10-23 Jakub Jelinek <jakub@redhat.com>
5987 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
5988 * config/i386/constraints.md (Wf): New constraint.
5989 * config/i386/i386.md (UNSPEC_SBB): New unspec.
5990 (cmp<dwi>_doubleword): Removed.
5991 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
5992 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
5993 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
5994 expand with cmp<dwi>_doubleword. For LTU and GEU use
5995 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
5997 * common.opt (gcolumn-info): Enable by default.
5998 * doc/invoke.texi (gcolumn-info): Document new default.
6000 2017-10-23 Richard Biener <rguenther@suse.de>
6002 PR tree-optimization/82672
6003 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
6004 Fold the stmt if we propagated into it.
6006 2017-10-23 Richard Biener <rguenther@suse.de>
6008 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
6009 (bitmap_remove_expr_from_set): ... this. All callers call this
6010 for non-constant values.
6011 (bitmap_set_subtract): Rename to...
6012 (bitmap_set_subtract_expressions): ... this. Adjust and
6014 (bitmap_set_contains_value): Remove superfluous check.
6015 (bitmap_set_replace_value): Inline into single caller ...
6016 (bitmap_value_replace_in_set): ... here and simplify.
6017 (dependent_clean): Merge into ...
6018 (clean): ... this using an overload. Adjust.
6019 (prune_clobbered_mems): Adjust.
6020 (compute_antic_aux): Likewise.
6021 (compute_partial_antic_aux): Likewise.
6023 2017-10-23 Richard Biener <rguenther@suse.de>
6025 PR tree-optimization/82129
6027 2017-08-01 Richard Biener <rguenther@suse.de>
6029 PR tree-optimization/81181
6030 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6031 (compute_antic): ... end of iteration here.
6033 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6035 * target.def (starting_frame_offset): New hook.
6036 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
6037 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
6038 * doc/tm.texi.in: Regenerate.
6039 * hooks.h (hook_hwi_void_0): Declare.
6040 * hooks.c (hook_hwi_void_0): New function.
6041 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
6042 STARTING_FRAME_OFFSET.
6043 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
6044 * reload1.c (reload): Likewise.
6045 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
6046 instead of STARTING_FRAME_OFFSET.
6047 * function.c (try_fit_stack_local): Likewise.
6048 (assign_stack_local_1): Likewise
6049 (instantiate_virtual_regs): Likewise.
6050 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
6051 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
6052 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
6053 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
6054 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
6055 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
6056 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
6057 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
6058 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
6059 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
6060 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
6061 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
6062 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
6063 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
6064 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
6065 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
6066 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
6067 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
6068 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
6069 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
6070 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
6071 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
6072 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
6073 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
6074 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
6075 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
6076 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
6077 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
6078 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
6079 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
6080 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
6081 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
6082 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
6083 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
6084 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
6085 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
6086 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
6087 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
6088 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
6089 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
6090 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
6091 * config/avr/avr.c (avr_starting_frame_offset): Make static and
6092 return a HOST_WIDE_INT.
6093 (avr_builtin_setjmp_frame_value): Use it instead of
6094 STARTING_FRAME_OFFSET.
6095 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6096 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
6097 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
6099 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6100 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
6101 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
6102 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6103 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
6104 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
6105 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6106 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
6107 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
6108 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6109 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
6110 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
6112 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6113 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
6114 * config/mips/mips.c (mips_compute_frame_info): Refer to
6115 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
6116 (mips_starting_frame_offset): New function.
6117 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6118 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
6119 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
6120 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
6121 and return a HOST_WIDE_INT.
6122 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6123 (mmix_initial_elimination_offset): Refer to
6124 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
6125 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
6126 * config/pa/pa.c (pa_starting_frame_offset): New function.
6127 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
6128 (pa_expand_prologue): Likewise.
6129 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6130 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
6131 !FRAME_GROWS_DOWNWARD handling to...
6132 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6133 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
6134 !FRAME_GROWS_DOWNWARD handling to...
6135 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6136 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
6137 !FRAME_GROWS_DOWNWARD handling to...
6138 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6139 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
6141 (rs6000_starting_frame_offset): New function.
6142 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
6143 !FRAME_GROWS_DOWNWARD handling to...
6144 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6145 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
6146 !FRAME_GROWS_DOWNWARD handling to...
6147 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6148 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
6149 !FRAME_GROWS_DOWNWARD handling to...
6150 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6151 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
6152 (rs6000_starting_frame_offset): New function.
6153 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
6154 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
6155 * config/vax/vax.c (vax_starting_frame_offset): New function.
6156 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
6157 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6158 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
6159 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
6160 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6161 * system.h (STARTING_FRAME_OFFSET): Poison.
6163 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6165 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
6166 SCALAR_TYPE_MODE instead of TYPE_MODE.
6168 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6169 Alan Hayward <alan.hayward@arm.com>
6170 David Sherwood <david.sherwood@arm.com>
6172 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
6174 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6175 Alan Hayward <alan.hayward@arm.com>
6176 David Sherwood <david.sherwood@arm.com>
6178 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
6180 2017-10-23 Richard Biener <rguenther@suse.de>
6182 PR tree-optimization/82129
6183 * tree-ssa-pre.c (bitmap_set_and): Remove.
6184 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
6185 canonicalizing expressions in the set to those with lowest
6186 ID rather than taking that from the first edge.
6188 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6190 * combine.c (rtx_equal_for_field_assignment_p): Use
6191 byte_lowpart_offset.
6193 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6194 Alan Hayward <alan.hayward@arm.com>
6195 David Sherwood <david.sherwood@arm.com>
6197 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
6198 to a SUBREG_PROMOTED_VAR.
6200 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6201 Alan Hayward <alan.hayward@arm.com>
6202 David Sherwood <david.sherwood@arm.com>
6204 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
6205 (expand_debug_source_expr): Likewise.
6206 * combine.c (combine_simplify_rtx): Likewise.
6207 * cse.c (fold_rtx): Likewise.
6208 * optabs.c (expand_float): Likewise.
6209 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6210 (simplify_binary_operation_1): Likewise.
6212 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6213 Alan Hayward <alan.hayward@arm.com>
6214 David Sherwood <david.sherwood@arm.com>
6216 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
6217 (record_promoted_value): Likewise.
6218 * expr.c (expand_expr_real_2): Likewise.
6219 * ree.c (update_reg_equal_equiv_notes): Likewise.
6220 (combine_set_extension): Likewise.
6221 * rtlanal.c (low_bitmask_len): Likewise.
6222 * simplify-rtx.c (neg_const_int): Likewise.
6223 (simplify_binary_operation_1): Likewise.
6225 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6226 Alan Hayward <alan.hayward@arm.com>
6227 David Sherwood <david.sherwood@arm.com>
6229 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
6230 * regcprop.c (maybe_mode_change): Likewise.
6231 * reload1.c (alter_reg): Likewise.
6233 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6235 * inchash.h (inchash::hash::add_wide_int): New function.
6236 * lto-streamer-out.c (hash_tree): Use it.
6238 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6240 * inchash.h (inchash::hash::add_wide_int): Rename to...
6241 (inchash::hash::add_hwi): ...this.
6242 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
6243 (polymorphic_call_target_hasher::hash): Likewise.
6244 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
6245 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
6246 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
6247 * lto-streamer-out.c (hash_tree): Likewise.
6248 * optc-save-gen.awk: Likewise.
6249 * tree.c (add_expr): Likewise.
6251 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
6254 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
6255 for ordered inequality comparisons even with TARGET_IEEE_FP.
6257 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
6260 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
6261 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
6262 Expand with cmp<dwi>_doubleword.
6264 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6266 * extend.texi: Add x86 specific to 'nocf_check' attribute.
6267 List CET intrinsics.
6268 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
6269 specific to -fcf-protection option.
6271 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6273 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
6274 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
6275 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
6276 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
6277 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
6278 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
6279 (extra_objs): Add cet.o for Linux/x86 targets.
6280 (tmake_file): Add i386/t-cet for Linux/x86 targets.
6281 * config/i386/cet.c: New file.
6282 * config/i386/cetintrin.h: Likewise.
6283 * config/i386/t-cet: Likewise.
6284 * config/i386/cpuid.h (bit_SHSTK): New.
6285 (bit_IBT): Likewise.
6286 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
6287 pass IBT and SHSTK bits.
6288 * config/i386/i386-builtin-types.def
6289 (VOID_FTYPE_UNSIGNED_PVOID): New.
6290 (VOID_FTYPE_UINT64_PVOID): Likewise.
6291 * config/i386/i386-builtin.def: Add CET intrinsics.
6292 * config/i386/i386-c.c (ix86_target_macros_internal): Add
6293 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
6294 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
6295 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
6297 * config/i386/i386.c (rest_of_insert_endbranch): New.
6298 (pass_data_insert_endbranch): Likewise.
6299 (pass_insert_endbranch): Likewise.
6300 (make_pass_insert_endbranch): Likewise.
6301 (ix86_notrack_prefixed_insn_p): Likewise.
6302 (ix86_target_string): Add -mibt, -mshstk flags.
6303 (ix86_option_override_internal): Add flag_cf_protection
6305 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
6306 (ix86_print_operand): Add 'notrack' prefix output.
6307 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
6308 (ix86_expand_builtin): Expand CET intrinsics.
6309 (x86_output_mi_thunk): Add 'endbranch' instruction.
6310 * config/i386/i386.h (TARGET_IBT): New.
6311 (TARGET_IBT_P): Likewise.
6312 (TARGET_SHSTK): Likewise.
6313 (TARGET_SHSTK_P): Likewise.
6314 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
6315 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
6316 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
6317 (builtin_setjmp_setup): New pattern.
6318 (builtin_longjmp): Likewise.
6319 (rdssp<mode>): Likewise.
6320 (incssp<mode>): Likewise.
6321 (saveprevssp): Likewise.
6322 (rstorssp): Likewise.
6323 (wrss<mode>): Likewise.
6324 (wruss<mode>): Likewise.
6325 (setssbsy): Likewise.
6326 (clrssbsy): Likewise.
6327 (nop_endbr): Likewise.
6328 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
6330 * config/i386/immintrin.h: Include <cetintrin.h>.
6331 * config/i386/linux-common.h
6332 (file_end_indicate_exec_stack_and_cet): New prototype.
6333 (TARGET_ASM_FILE_END): New.
6335 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
6337 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
6338 latencies instead of having separate table; make difference between
6339 integer and float costs.
6340 * i386.h (processor_costs): Remove scalar_stmt_cost,
6341 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
6342 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
6344 * x86-tune-costs.h: Remove entries which has been removed in
6345 procesor_costs from all tables; make cond_taken_branch_cost
6346 and cond_not_taken_branch_cost COST_N_INSNS based.
6348 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
6350 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
6352 2017-10-20 Jakub Jelinek <jakub@redhat.com>
6354 * config/i386/i386.md (isa): Remove fma_avx512f.
6355 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
6356 <avx512>_fmadd_<mode>_mask3<round_name>,
6357 <avx512>_fmsub_<mode>_mask<round_name>,
6358 <avx512>_fmsub_<mode>_mask3<round_name>,
6359 <avx512>_fnmadd_<mode>_mask<round_name>,
6360 <avx512>_fnmadd_<mode>_mask3<round_name>,
6361 <avx512>_fnmsub_<mode>_mask<round_name>,
6362 <avx512>_fnmsub_<mode>_mask3<round_name>,
6363 <avx512>_fmaddsub_<mode>_mask<round_name>,
6364 <avx512>_fmaddsub_<mode>_mask3<round_name>,
6365 <avx512>_fmsubadd_<mode>_mask<round_name>,
6366 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
6367 (*vec_widen_umult_even_v16si<mask_name>,
6368 *vec_widen_smult_even_v16si<mask_name>): Likewise.
6369 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
6371 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6373 * extend.texi: Add 'nocf_check' documentation.
6374 * gimple.texi: Add second parameter to
6375 gimple_build_call_from_tree.
6376 * invoke.texi: Add -fcf-protection documentation.
6377 * rtl.texi: Add REG_CALL_NOTRACK documenation.
6379 2017-10-20 Richard Biener <rguenther@suse.de>
6381 PR tree-optimization/82473
6382 * tree-vect-loop.c (vectorizable_reduction): Properly get at
6383 the largest input type.
6385 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6387 * c-attribs.c (handle_nocf_check_attribute): New function.
6388 (c_common_attribute_table): Add 'nocf_check' handling.
6389 * gimple-parser.c: Add second argument NULL to
6390 gimple_build_call_from_tree.
6391 * attrib.c (comp_type_attributes): Check nocf_check attribute.
6392 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
6394 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
6395 * common.opt: Add fcf-protection flag.
6396 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
6397 * flag-types.h: Add enum cf_protection_level.
6398 * gimple.c (gimple_build_call_from_tree): Add second parameter.
6399 Add 'nocf_check' attribute propagation to gimple call.
6400 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
6401 (gimple_build_call_from_tree): Update prototype.
6402 (gimple_call_nocf_check_p): New function.
6403 (gimple_call_set_nocf_check): Likewise.
6404 * gimplify.c: Add second argument to gimple_build_call_from_tree.
6405 * ipa-icf.c: Add nocf_check attribute in statement hash.
6406 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
6407 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
6408 * toplev.c (process_options): Add flag_cf_protection handling.
6410 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6412 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
6414 2017-10-20 Richard Biener <rguenther@suse.de>
6416 PR tree-optimization/82603
6417 * tree-if-conv.c (predicate_mem_writes): Make sure to only
6418 remove false predicated stores.
6420 2017-10-20 Richard Biener <rguenther@suse.de>
6422 * graphite-isl-ast-to-gimple.c
6423 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
6424 Remove return value and simplify, dump copied stmt after lhs
6426 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
6427 Reduce dump verbosity.
6428 (gsi_insert_earliest): Likewise.
6429 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
6430 * graphite.c (print_global_statistics): Adjust dumping.
6431 (print_graphite_scop_statistics): Likewise.
6432 (print_graphite_statistics): Do not dump loops here.
6433 (graphite_transform_loops): But here.
6435 2017-10-20 Nicolas Roche <roche@adacore.com>
6437 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
6438 * configure: Regenerate.
6440 2017-10-20 Jakub Jelinek <jakub@redhat.com>
6443 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
6444 functions when optimizing replace GIMPLE_RETURN stmts with
6445 calls to __builtin_unreachable ().
6448 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
6449 for -fsanitize=thread link of executables.
6450 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
6451 link of executables.
6454 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
6456 (<shift_insn><mode>3<mask_name>): Change the last of the 3
6457 define_insns for logical vector shifts to use VI248_AVX512BW
6458 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
6459 condition, useless isa and prefix attributes. Change the first
6460 2 of these define_insns to ...
6461 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
6462 define_insn for avx512vl.
6463 (<shift_insn><mode>3): ... and this, new define_insn without
6464 masking for non-avx512vl.
6467 * config/i386/sse.md (*andnot<mode>3,
6468 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
6469 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
6470 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
6471 not applied use empty suffix even for TARGET_AVX512VL.
6472 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
6473 is applied, supply evex,evex or evex,evex,evex instead of just
6476 2017-10-20 Julia Koval <julia.koval@intel.com>
6478 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
6479 (OPTION_MASK_ISA_GFNI_UNSET): New.
6480 (ix86_handle_option): Handle OPT_mgfni.
6481 * config/i386/cpuid.h (bit_GFNI): New.
6482 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
6483 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
6484 * config/i386/i386.c (ix86_target_string): Add -mgfni.
6485 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
6486 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
6487 * config/i386/i386.opt: Add mgfni.
6489 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
6491 * config/msp430/msp430.c (msp430_option_override): Disable
6492 -fdelete-null-pointer-checks.
6493 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
6495 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6497 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
6498 of x87 and SSE instructions.
6500 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
6502 * asan.c (create_cond_insert_point): Do not update edge count.
6503 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
6504 (afdo_propagate_circuit): Likewise.
6505 (afdo_calculate_branch_prob): Likewise.
6506 (afdo_annotate_cfg): Likewise.
6507 * basic-block.h (struct edge_def): Remove count.
6508 (edge_def::count): New accessor.
6509 * bb-reorder.c (rotate_loop): Update.
6510 (find_traces_1_round): Update.
6511 (connect_traces): Update.
6512 (sanitize_hot_paths): Update.
6513 * cfg.c (unchecked_make_edge): Update.
6514 (make_single_succ_edge): Update.
6515 (check_bb_profile): Update.
6516 (dump_edge_info): Update.
6517 (update_bb_profile_for_threading): Update.
6518 (scale_bbs_frequencies_int): Update.
6519 (scale_bbs_frequencies_gcov_type): Update.
6520 (scale_bbs_frequencies_profile_count): Update.
6521 (scale_bbs_frequencies): Update.
6522 * cfganal.c (connect_infinite_loops_to_exit): Update.
6523 * cfgbuild.c (compute_outgoing_frequencies): Update.
6524 (find_many_sub_basic_blocks): Update.
6525 * cfgcleanup.c (try_forward_edges): Update.
6526 (try_crossjump_to_edge): Update
6527 * cfgexpand.c (expand_gimple_cond): Update
6528 (expand_gimple_tailcall): Update
6529 (construct_exit_block): Update
6530 * cfghooks.c (verify_flow_info): Update
6531 (redirect_edge_succ_nodup): Update
6532 (split_edge): Update
6533 (make_forwarder_block): Update
6534 (duplicate_block): Update
6535 (account_profile_record): Update
6536 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
6537 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
6538 * cfgloopmanip.c (scale_loop_profile): Update.
6540 (lv_adjust_loop_entry_edge): Update.
6541 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
6542 (force_nonfallthru_and_redirect): Update.
6543 (purge_dead_edges): Update.
6544 (rtl_flow_call_edges_add): Update.
6545 * cgraphunit.c (init_lowered_empty_function): Update.
6546 (cgraph_node::expand_thunk): Update.
6547 * gimple-pretty-print.c (dump_probability): Update.
6548 (dump_edge_probability): Update.
6549 * gimple-ssa-isolate-paths.c (isolate_path): Update.
6550 * haifa-sched.c (sched_create_recovery_edges): Update.
6551 * hsa-gen.c (convert_switch_statements): Update.
6552 * ifcvt.c (dead_or_predicable): Update.
6553 * ipa-inline-transform.c (inline_transform): Update.
6554 * ipa-split.c (split_function): Update.
6555 * ipa-utils.c (ipa_merge_profiles): Update.
6556 * loop-doloop.c (add_test): Update.
6557 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
6558 * lto-streamer-in.c (input_cfg): Update.
6559 (input_function): Update.
6560 * lto-streamer-out.c (output_cfg): Update.
6561 * modulo-sched.c (sms_schedule): Update.
6562 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
6563 * predict.c (maybe_hot_edge_p): Update.
6564 (unlikely_executed_edge_p): Update.
6565 (probably_never_executed_edge_p): Update.
6566 (dump_prediction): Update.
6567 (drop_profile): Update.
6568 (propagate_unlikely_bbs_forward): Update.
6569 (determine_unlikely_bbs): Update.
6570 (force_edge_cold): Update.
6571 * profile.c (compute_branch_probabilities): Update.
6572 * reg-stack.c (better_edge): Update.
6573 * shrink-wrap.c (handle_simple_exit): Update.
6574 * tracer.c (better_p): Update.
6575 * trans-mem.c (expand_transaction): Update.
6576 (split_bb_make_tm_edge): Update.
6577 * tree-call-cdce.c: Update.
6578 * tree-cfg.c (gimple_find_sub_bbs): Update.
6579 (gimple_split_edge): Update.
6580 (gimple_duplicate_sese_region): Update.
6581 (gimple_duplicate_sese_tail): Update.
6582 (gimple_flow_call_edges_add): Update.
6583 (insert_cond_bb): Update.
6584 (execute_fixup_cfg): Update.
6585 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
6586 * tree-complex.c (expand_complex_div_wide): Update.
6587 * tree-eh.c (lower_resx): Update.
6588 (unsplit_eh): Update.
6589 (cleanup_empty_eh_move_lp): Update.
6590 * tree-inline.c (copy_edges_for_bb): Update.
6591 (freqs_to_counts): Update.
6592 (copy_cfg_body): Update.
6593 * tree-ssa-dce.c (remove_dead_stmt): Update.
6594 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
6595 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
6596 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
6597 (unloop_loops): Update.
6598 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
6599 * tree-ssa-loop-split.c (connect_loops): Update.
6600 (split_loop): Update.
6601 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
6602 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
6603 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
6604 * tree-ssa-reassoc.c (branch_fixup): Update.
6605 * tree-ssa-tail-merge.c (replace_block_by): Update.
6606 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
6607 (compute_path_counts): Update.
6608 (update_profile): Update.
6609 (recompute_probabilities): Update.
6610 (update_joiner_offpath_counts): Update.
6611 (estimated_freqs_path): Update.
6612 (freqs_to_counts_path): Update.
6613 (clear_counts_path): Update.
6614 (ssa_fix_duplicate_block_edges): Update.
6615 (duplicate_thread_path): Update.
6616 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
6617 (case_bit_test_cmp): Update.
6618 (collect_switch_conv_info): Update.
6619 (gen_inbound_check): Update.
6620 (do_jump_if_equal): Update.
6621 (emit_cmp_and_jump_insns): Update.
6622 * tree-tailcall.c (decrease_profile): Update.
6623 (eliminate_tail_call): Update.
6624 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
6625 (vect_do_peeling): Update.
6626 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
6627 * ubsan.c (ubsan_expand_null_ifn): Update.
6628 (ubsan_expand_ptr_ifn): Update.
6629 * value-prof.c (gimple_divmod_fixed_value): Update.
6630 (gimple_mod_pow2): Update.
6631 (gimple_mod_subtract): Update.
6632 (gimple_ic): Update.
6633 (gimple_stringop_fixed_value): Update.
6635 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
6638 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
6640 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
6642 PR rtl-optimization/82395
6643 * ira-color.c (allocno_priority_compare_func): Fix comparison step
6644 based on non_spilled_static_chain_regno_p.
6646 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
6648 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
6649 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
6650 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
6652 2017-10-19 Martin Sebor <msebor@redhat.com>
6654 PR tree-optimization/82596
6655 * tree.c (array_at_struct_end_p): Handle STRING_CST.
6657 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
6659 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
6660 (get_mem_refs_of_builtin_call): Likewise.
6661 * builtins.c (expand_builtin_apply): Adjust call to
6662 allocate_dynamic_stack_space.
6663 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
6664 the third argument to allocate_dynamic_stack_space, otherwise -1.
6665 (expand_builtin): Deal with all alloca variants.
6666 (is_inexpensive_builtin): Likewise.
6667 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
6668 * calls.c (special_function_p): Deal with all alloca variants.
6669 (initialize_argument_information): Adjust call to
6670 allocate_dynamic_stack_space.
6671 (expand_call): Likewise.
6672 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
6673 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
6674 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
6675 use it for the stack usage computation.
6676 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
6677 * function.c (gimplify_parameters): Call build_alloca_call_expr.
6678 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
6679 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
6680 (in_loop_p): Remove first argument and useless check.
6681 (pass_walloca::execute): Remove useless test and adjust call to above.
6682 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
6683 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
6684 (gimplify_call_expr): Deal with all alloca variants.
6685 * hsa-gen.c (gen_hsa_alloca): Likewise.
6686 (gen_hsa_insns_for_call): Likewise.
6687 * ipa-pure-const.c (special_builtin_state): Likewise.
6688 * tree-chkp.c (chkp_build_returned_bound): Likewise.
6689 * tree-object-size.c (alloc_object_size): Likewise.
6690 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
6691 (call_may_clobber_ref_p_1): Likewise.
6692 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
6693 (ccp_fold_stmt): Likewise.
6694 (optimize_stack_restore): Likewise.
6695 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
6696 (mark_all_reaching_defs_necessary_1): Likewise.
6697 (propagate_necessity): Likewise.
6698 (eliminate_unnecessary_stmts): Likewise.
6699 * tree.c (build_common_builtin_nodes): Build
6700 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
6701 (build_alloca_call_expr): New function.
6702 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
6703 (CASE_BUILT_IN_ALLOCA): Likewise.
6704 (build_alloca_call_expr): Declare.
6705 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
6707 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
6710 * dwarf2out.c (new_die_raw): New static inline function.
6711 (new_die): Use it to create the DIE.
6712 (add_AT_external_die_ref): Likewise.
6713 (clone_die): Likewise.
6714 (clone_as_declaration): Likewise.
6715 (dwarf2out_vms_debug_main_pointer): Likewise.
6716 (base_type_die): Likewise. Remove early return for corner cases.
6717 Do not call add_pubtype on the DIE here.
6718 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
6719 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
6720 typedefs for base types with DW_AT_endianity. Make sure a DIE with
6721 native order exists for base types, attach the DIE manually and call
6722 add_pubtype on it. Do not equate a reverse order DIE to the type.
6724 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
6726 * config/arm/arm.c (align_ok_ldrd_strd): New function.
6727 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
6729 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
6731 2017-10-19 Jakub Jelinek <jakub@redhat.com>
6733 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
6734 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
6735 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
6736 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
6737 * opts.c (sanitizer_opts): Add builtin.
6738 * ubsan.c (instrument_builtin): New function.
6739 (pass_ubsan::execute): Call it.
6740 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
6741 * doc/invoke.texi: Document -fsanitize=builtin.
6743 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
6744 builtins, store max (log2 (align), 0) into uchar field instead of
6745 align into uptr field.
6746 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
6747 store uchar 0 field instead of uptr 0 field.
6748 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
6749 instead of passing one address of struct with 2 locations pass
6750 two addresses of structs with 1 location each.
6751 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
6752 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
6753 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
6754 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
6755 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
6756 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
6757 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
6758 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
6760 2017-10-19 Martin Liska <mliska@suse.cz>
6763 * file-find.c (remove_prefix): Remove.
6764 * file-find.h (remove_prefix): Likewise.
6765 * gcc-ar.c: Remove smartness of lookup.
6767 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
6769 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
6770 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
6771 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
6773 2017-10-19 Jakub Jelinek <jakub@redhat.com>
6776 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
6777 (setcc + and to xor + setcc): New peephole2.
6779 2017-10-19 Tom de Vries <tom@codesourcery.com>
6781 * doc/sourcebuild.texi (Test Directives, Variants of
6782 dg-require-support): Add dg-require-stack-size.
6784 2017-10-19 Martin Liska <mliska@suse.cz>
6787 * gimplify.c (gimplify_decl_expr): Do not instrument variables
6788 that have a large alignment.
6789 (gimplify_target_expr): Likewise.
6791 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
6793 PR rtl-optimization/82602
6794 * ira.c (rtx_moveable_p): Return false for volatile asm.
6796 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
6799 * config/i386/i386-modes.def (CCGZ): New CC mode.
6800 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
6801 * config/i386/predicates.md (ix86_comparison_operator):
6803 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
6804 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
6805 with double-word subtraction.
6806 (put_condition_code): Handle CCGZmode.
6808 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
6810 * wide-int.cc (debug (const wide_int &)): New.
6811 (debug (const wide_int *)): New.
6812 (debug (const widest_int &)): New.
6813 (debug (const widest_int *)): New.
6815 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
6818 * lra-constraints.c (curr_insn_transform): Use non-input operand
6819 instead of output one for matched reload.
6821 2017-10-18 Bin Cheng <bin.cheng@arm.com>
6823 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
6824 (tree-ssa-loop-ivopts.h): New header file.
6825 (struct builtin_info): New fields.
6826 (classify_builtin_1): Compute and record base and offset parts for
6827 memset builtin partition by calling strip_offset.
6828 (offset_cmp, fuse_memset_builtins): New functions.
6829 (finalize_partitions): Fuse adjacent memset partitions by calling
6831 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
6832 Expose the interface.
6833 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
6835 2017-10-18 Bin Cheng <bin.cheng@arm.com>
6837 PR tree-optimization/82574
6838 * tree-loop-distribution.c (find_single_drs): New parameter. Check
6839 that data reference must be executed exactly once per iteration
6840 against the outermost loop in nest.
6841 (classify_partition): Update call to above function.
6843 2017-10-18 Richard Biener <rguenther@suse.de>
6845 PR tree-optimization/82591
6846 * graphite.c (graphite_transform_loops): Move code gen message
6848 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6849 Here. Handle scop_to_isl_ast failing.
6850 (scop_to_isl_ast): Limit the number of ISL operations.
6852 2017-10-18 Richard Biener <rguenther@suse.de>
6854 * graphite-isl-ast-to-gimple.c
6855 (translate_isl_ast_to_gimple::set_rename): Simplify.
6856 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
6857 (graphite_copy_stmts_from_block): ... here.
6858 (copy_bb_and_scalar_dependences): Simplify.
6859 (add_parameters_to_ivs_params): Canonicalize.
6860 (generate_entry_out_of_ssa_copies): Simplify.
6861 * graphite-sese-to-poly.c (extract_affine_name): Simplify
6862 by passing in ISL dimension.
6863 (parameter_index_in_region_1): Rename to ...
6864 (parameter_index_in_region): ... this.
6865 (extract_affine): Adjust assert, pass down parameter index.
6866 (add_param_constraints): Use range-info when available.
6867 (build_scop_context): Adjust.
6868 * sese.c (new_sese_info): Adjust.
6869 (free_sese_info): Likewise.
6870 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
6871 Remove unused typedefs.
6872 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
6874 2017-10-18 Martin Liska <mliska@suse.cz>
6876 * combine.c (simplify_compare_const): Add gcc_fallthrough.
6878 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
6880 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
6881 (s390_sched_init): Do not reset s390_sched_state if we entered the
6882 current basic block via a fallthru edge and all others are unlikely.
6884 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
6886 * config/s390/s390.c (NUM_SIDES): New variable.
6887 (LONGRUNNING_THRESHOLD): New variable.
6888 (LATENCY_FACTOR): New variable.
6889 (s390_sched_score): Decrease score for long-running instructions on
6891 (s390_sched_variable_issue): Perform bookkeeping for long-running
6894 2017-10-18 Richard Biener <rguenther@suse.de>
6896 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
6897 Simplify with removal of the parameter rename map.
6898 (set_rename): Likewise.
6899 (should_copy_to_new_region): Likewise.
6900 (graphite_copy_stmts_from_block): Likewise.
6901 (copy_bb_and_scalar_dependences): Remove initialization of
6902 unused copied_bb_map.
6904 (copy_internal_parameters): Likewise.
6905 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
6906 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
6907 Use INTEGRAL_TYPE_P.
6908 (parameter_index_in_region_1): Rename to ...
6909 (assign_parameter_index_in_region): ... this. Assert we have
6910 a parameter we handle.
6911 (scan_tree_for_params): Adjust.
6912 * sese.h (parameter_rename_map_t): Remove.
6913 (struct sese_info_t): Remove unused parameter_rename_map and
6914 copied_bb_map members.
6915 * sese.c (new_sese_info): Adjust.
6916 (free_sese_info): Likewise.
6918 2017-10-18 Martin Liska <mliska@suse.cz>
6921 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
6922 on an abnormal edge.
6924 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
6926 * doc/invoke.texi (ffunction-sections and fdata-sections):
6929 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
6931 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
6932 the use statement can throw internally.
6934 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
6936 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
6937 any RTX present on the RHS of a SET.
6938 * compare-elim.c (try_eliminate_compare): Restore comment.
6940 2017-10-17 Jakub Jelinek <jakub@redhat.com>
6942 * langhooks.h (struct lang_hooks): Document that tree_size langhook
6943 may be also called on tcc_type nodes.
6944 * langhooks.c (lhd_tree_size): Likewise.
6946 2017-10-17 David Malcolm <dmalcolm@redhat.com>
6948 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
6949 format_warning_at_substring.
6950 (maybe_warn): Convert source_range * param to a location_t. Pass
6951 UNKNOWN_LOCATION rather than NULL to fmtwarn.
6952 (format_directive): Remove code to extract source_ranges and
6953 source_range * in favor of just a location_t.
6954 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
6956 * substring-locations.c (format_warning_va): Convert
6957 source_range * param to a location_t.
6958 (format_warning_at_substring): Likewise.
6959 * substring-locations.h (format_warning_va): Likewise.
6960 (format_warning_at_substring): Likewise.
6962 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
6964 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
6966 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
6967 and scatter/gather ops.
6969 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
6970 vec_gather_load and vec_scatter_store.
6971 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
6972 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
6973 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
6974 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
6975 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
6976 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
6978 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
6980 * reg-stack.c (compare_for_stack_reg): Add bool argument.
6981 Detect FTST instruction and handle its register pops. Only pop
6982 second operand if can_pop_second_op is true.
6983 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
6984 set can_pop_second_op to false in the compare_for_stack_reg call.
6986 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
6987 output_fp_compare for stack register operands.
6988 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
6989 instructions here. Do not emit stack register pops here. Assert
6990 that FCOMPP pops next to top stack register. Rewrite function.
6992 2017-10-17 Nathan Sidwell <nathan@acm.org>
6995 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
6996 use DECL_ASSEMBLER_NAME_RAW.
6999 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
7002 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
7003 Wilco Dijkstra <wilco.dijkstra@arm.com>
7005 * builtins.c (expand_builtin_update_setjmp_buf): Add a
7006 converstion to Pmode from the buf_addr.
7008 2017-10-17 Richard Biener <rguenther@suse.de>
7010 * graphite-dependences.c (scop_get_reads_and_writes): Change
7011 output parameters to references.
7013 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
7015 PR 71026/tree-optimization
7016 * fold-const.c (distribute_real_division): Removed.
7017 (fold_binary_loc): Remove calls to distribute_real_divison.
7019 2017-10-17 Richard Biener <rguenther@suse.de>
7021 * graphite-scop-detection.c
7022 (scop_detection::stmt_has_simple_data_refs_p): Always use
7023 the full nest as region.
7024 (try_generate_gimple_bb): Likewise.
7025 * sese.c (scalar_evolution_in_region): Simplify now that
7026 SCEV can handle instantiation in regions.
7027 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
7028 in the non-loop part of a function if requested.
7030 2017-10-17 Richard Biener <rguenther@suse.de>
7032 PR tree-optimization/82563
7033 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
7035 (graphite_regenerate_ast_isl): Call it.
7036 * graphite-scop-detection.c (build_scops): Remove entry edge split.
7038 2017-10-17 Jakub Jelinek <jakub@redhat.com>
7040 PR tree-optimization/82549
7041 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
7042 Formatting fixes. Instead of calling make_bit_field_ref with negative
7045 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
7047 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
7048 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
7049 _mm_maskz_reduce_ss): New.
7050 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
7051 __builtin_ia32_reducess_mask): Ditto..
7052 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
7053 * config/i386/sse.md (reduces<mode>): Renamed to ...
7054 (reduces<mode><mask_scalar_name>): ... this.
7055 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
7057 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
7058 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
7060 2017-10-16 David Malcolm <dmalcolm@redhat.com>
7062 * Makefile.in (OBJS): Add unique-ptr-tests.o.
7063 * selftest-run-tests.c (selftest::run_tests): Call
7064 selftest::unique_ptr_tests_cc_tests.
7065 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
7066 * unique-ptr-tests.cc: New file.
7068 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
7071 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
7073 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
7074 (make_hard_regno_born, make_hard_regno_dead): Update
7075 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
7077 2017-10-16 Jeff Law <law@redhat.com>
7079 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
7081 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
7083 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
7085 2017-10-16 Olivier Hainque <hainque@adacore.com>
7087 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
7088 with_cpu if we were configured for an e500v2 target cpu name.
7090 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
7092 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
7093 * doc/invoke.texi: Document +nodsp as a valid extension for
7096 2017-10-16 Martin Liska <mliska@suse.cz>
7098 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
7099 (test_set_range): Likewise.
7100 (test_range_functions): Rename to ...
7101 (test_bit_in_range): ... this.
7102 (sbitmap_c_tests): Add new test.
7104 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7106 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
7108 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
7109 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
7111 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7113 * config/aarch64/aarch64-builtins.c
7114 (aarch64_types_quadopu_lane_qualifiers): New.
7115 (TYPES_QUADOPU_LANE): New.
7116 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
7117 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
7118 (aarch64_<sur>dot_laneq<vsi2qi>): New.
7119 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
7120 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
7121 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
7122 (Vdottype, DOTPROD): New.
7123 (sur): Add SDOT and UDOT.
7125 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7127 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
7128 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
7129 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
7131 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
7132 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
7133 Enable TARGET_DOTPROD.
7134 (cortex-a75.cortex-a55): Likewise.
7135 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
7137 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7139 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
7140 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
7142 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
7144 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
7145 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
7146 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
7147 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
7148 * config/arm/types.md (neon_dot, neon_dot_q): New.
7149 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
7151 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7153 * config/arm/arm.h (TARGET_DOTPROD): New.
7154 * config/arm/arm.c (arm_arch_dotprod): New.
7155 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
7156 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
7157 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
7158 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
7159 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
7160 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
7161 * doc/invoke.texi (armv8.2-a): Document dotprod
7163 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
7165 * i386.c (ix86_vec_cost): New function.
7166 (ix86_rtx_costs): Handle vector operations better.
7167 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
7168 * x86-tune-costs.h: Add new costs to all tables.
7170 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
7172 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
7174 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
7175 divsd, sqrtss and sqrtsd
7176 * x86-tune-costs.h: Add new entries to all costs.
7177 (znver1_cost): Fix to match real instruction latencies.
7179 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7180 Michael Collison <michael.collison@arm.com>
7182 * compare-elim.c: Include emit-rtl.h.
7183 (can_merge_compare_into_arith): New function.
7184 (try_validate_parallel): Likewise.
7185 (try_merge_compare): Likewise.
7186 (try_eliminate_compare): Call the above when no previous clobber
7188 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
7191 2017-10-14 Jakub Jelinek <jakub@redhat.com>
7195 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
7196 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
7200 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
7201 to be any operand_equal_p operands. For & (B - 1) require
7202 B to be power of 2. Recognize
7203 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
7205 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7208 * optabs.c (expand_memory_blockage): Fix call of
7209 targetm.have_memory_blockage.
7211 2017-10-14 Jakub Jelinek <jakub@redhat.com>
7214 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
7215 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
7216 objects to extra_objs instead of overwriting it.
7218 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7220 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
7221 Use any_fp_register_operand as operand[3] predicate. Simplify
7222 equality test for operands[2] and operands[4] memory location.
7223 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
7224 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
7225 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
7226 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
7227 any_fp_register_operand as operand[1] predicate. Simplify
7228 equality test for operands[0] and operands[3] memory location.
7229 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
7230 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
7231 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
7233 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7235 * target-insns.def: Add memory_blockage.
7236 * optabs.c (expand_memory_blockage): New function.
7237 (expand_asm_memory_barrier): Rename ...
7238 (expand_asm_memory_blockage): ... to this.
7239 (expand_mem_thread_fence): Call expand_memory_blockage
7240 instead of expand_asm_memory_barrier.
7241 (expand_mem_singnal_fence): Ditto.
7242 (expand_atomic_load): Ditto.
7243 (expand_atomic_store): Ditto.
7244 * doc/md.texi (Standard Pattern Names For Generation):
7245 Document memory_blockage instruction pattern.
7247 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
7249 * config/rl78/rl78.c (rl78_emit_libcall): New function.
7250 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
7251 * config/rl78/rl78.md: New define_expand "adddi3".
7253 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
7255 * cfghooks.c (verify_flow_info): Disable check that all probabilities
7258 2017-10-13 Jeff Law <law@redhat.com>
7260 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
7262 2017-10-13 Jakub Jelinek <jakub@redhat.com>
7265 * internal-fn.c (expand_mul_overflow): If both operands have
7266 the same highpart of -1 or 0 and the topmost bit of lowpart
7267 is different, overflow is if res <= 0 rather than res < 0.
7269 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
7271 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
7272 TARGET_P9_VECTOR code for unaligned_load case.
7274 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
7276 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
7278 2017-10-13 Nathan Sidwell <nathan@acm.org>
7280 * tree-core.h (tree_contains_struct): Make bool.
7281 * tree.c (tree_contains_struct): Likewise.
7282 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
7283 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
7284 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
7285 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
7286 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
7288 2017-10-13 Richard Biener <rguenther@suse.de>
7290 * graphite-isl-ast-to-gimple.c
7291 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
7292 parameters and dominance check.
7293 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
7294 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
7295 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
7296 Do not update SSA form here or do intermediate IL verification.
7297 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
7298 (graphite_initialize): Remove check on the number of loops in
7299 the function and inline into graphite_transform_loops.
7300 (graphite_finalize): Inline into graphite_transform_loops.
7301 (graphite_transform_loops): Perform SSA update and IL verification
7303 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
7305 2017-10-13 Richard Biener <rguenther@suse.de>
7307 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
7308 graphite_expression_type_precision): Avoid global constructor
7310 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
7311 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
7312 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
7313 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
7314 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
7315 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
7317 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
7320 * config/i386/i386.h (ix86_red_zone_size): New.
7321 * config/i386/i386.md (push peephole2s): Replace
7322 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
7324 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7325 Alan Hayward <alan.hayward@arm.com>
7326 David Sherwood <david.sherwood@arm.com>
7328 * combine.c (can_change_dest_mode): Reject changes in
7329 REGMODE_NATURAL_SIZE.
7331 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7332 Alan Hayward <alan.hayward@arm.com>
7333 David Sherwood <david.sherwood@arm.com>
7335 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
7336 (expand_debug_source_expr): Likewise.
7337 * combine.c (combine_simplify_rtx): Likewise.
7338 * cse.c (fold_rtx): Likewise.
7339 * fwprop.c (canonicalize_address): Likewise.
7340 * targhooks.c (default_shift_truncation_mask): Likewise.
7342 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7343 Alan Hayward <alan.hayward@arm.com>
7344 David Sherwood <david.sherwood@arm.com>
7346 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
7347 (widened_mode): Likewise.
7348 (expand_unop): Likewise.
7349 * ree.c (transform_ifelse): Likewise.
7350 (merge_def_and_ext): Likewise.
7351 (combine_reaching_defs): Likewise.
7352 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7354 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7355 Alan Hayward <alan.hayward@arm.com>
7356 David Sherwood <david.sherwood@arm.com>
7358 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
7359 * combine.c (gen_lowpart_for_combine): Likewise.
7360 * dwarf2out.c (rtl_for_decl_location): Likewise.
7361 * final.c (alter_subreg): Likewise.
7362 * rtlhooks.c (gen_lowpart_general): Likewise.
7363 (gen_lowpart_if_possible): Likewise.
7365 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7366 Alan Hayward <alan.hayward@arm.com>
7367 David Sherwood <david.sherwood@arm.com>
7369 * calls.c (expand_call): Use subreg_lowpart_offset.
7370 * cse.c (cse_insn): Likewise.
7371 * regcprop.c (copy_value): Likewise.
7372 (copyprop_hardreg_forward_1): Likewise.
7374 2017-10-13 Jakub Jelinek <jakub@redhat.com>
7377 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
7378 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
7379 =Q constraints to +Q and into insn condition add check
7380 that operands[0] and operands[1] are equal.
7381 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
7382 =Q constraints to +Q and into insn condition add check
7383 that operands[0] is equal to either operands[1] or operands[2].
7386 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
7387 instead of handling MINUS_EXPR twice (once for each argument),
7388 canonicalize operand order and handle just once, use rtype where
7389 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
7392 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
7393 any values of __C while still being pattern recognizable as a simple
7396 2017-10-13 Richard Biener <rguenther@suse.de>
7398 PR tree-optimization/82451
7400 2017-10-02 Richard Biener <rguenther@suse.de>
7402 PR tree-optimization/82355
7403 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
7404 a mapping for the enclosing loop but avoid generating one for
7406 (copy_bb_and_scalar_dependences): Remove premature codegen
7407 error on PHIs in blocks duplicated into multiple places.
7408 * graphite-scop-detection.c
7409 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
7410 in the region use it as loop and nest to analyze the DR in.
7411 (try_generate_gimple_bb): Likewise.
7412 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
7413 (add_loop_constraints): For blocks in a loop not in the region
7414 create a dimension with a single iteration.
7415 * sese.h (gbb_loop_at_index): Remove assert.
7417 * cfgloop.c (loop_preheader_edge): For the loop tree root
7418 return the single successor of the entry block.
7419 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7420 Reset the SCEV hashtable and niters.
7421 * graphite-scop-detection.c
7422 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
7423 assert that we only have POLYNOMIAL_CHREC that vary in loops
7424 contained in the region.
7425 (scop_detection::graphite_can_represent_expr): Adjust.
7426 (scop_detection::stmt_has_simple_data_refs_p): For loops
7427 not in the region set loop to NULL. The nest is now the
7428 entry edge to the region.
7429 (try_generate_gimple_bb): Likewise.
7430 * sese.c (scalar_evolution_in_region): Adjust for
7431 instantiate_scev change.
7432 * tree-data-ref.h (graphite_find_data_references_in_stmt):
7433 Make nest parameter the edge into the region.
7434 (create_data_ref): Likewise.
7435 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
7436 entry edge into a region and adjust instantiate_scev calls.
7437 (create_data_ref): Likewise.
7438 (graphite_find_data_references_in_stmt): Likewise.
7439 (find_data_references_in_stmt): Pass the loop preheader edge
7440 from the nest argument.
7441 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
7442 parameter the edge into the region.
7443 (instantiate_parameters): Use the loop preheader edge as entry.
7444 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
7446 (get_instantiated_value_entry): Make instantiate_below parameter
7447 the edge into the region.
7448 (instantiate_scev_name): Likewise. Adjust dominance checks,
7449 when we cannot use loop-based instantiation instantiate by
7450 walking use-def chains.
7451 (instantiate_scev_poly): Adjust.
7452 (instantiate_scev_binary): Likewise.
7453 (instantiate_scev_convert): Likewise.
7454 (instantiate_scev_not): Likewise.
7455 (instantiate_array_ref): Remove.
7456 (instantiate_scev_3): Likewise.
7457 (instantiate_scev_2): Likewise.
7458 (instantiate_scev_1): Likewise.
7459 (instantiate_scev_r): Do not blindly handle N-operand trees.
7460 Do not instantiate array-refs. Handle all constants and invariants.
7461 (instantiate_scev): Make instantiate_below parameter
7462 the edge into the region.
7463 (resolve_mixers): Use the loop preheader edge for the region
7464 parameter to instantiate_scev_r.
7465 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
7467 2017-10-13 Richard Biener <rguenther@suse.de>
7469 PR tree-optimization/82525
7470 * graphite-isl-ast-to-gimple.c
7471 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
7473 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
7474 Fail code generation when we cannot represent the isl integer.
7475 (binary_op_to_tree): Elide modulo operations that are no-ops
7476 in the type we code generate. Remove now superfluous code
7479 2017-10-13 Richard Biener <rguenther@suse.de>
7481 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
7482 (scop_detection::harmful_loop_in_region): Remove premature
7483 IV type restriction.
7484 (scop_detection::graphite_can_represent_scev): We can handle
7485 pointer IVs just fine.
7487 2017-10-13 Alan Modra <amodra@gmail.com>
7489 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
7490 "Clobbers and Scratch Registers". Add paragraph on
7491 alternative to clobbers for scratch registers and OpenBLAS
7494 2017-10-13 Alan Modra <amodra@gmail.com>
7496 * doc/extend.texi (Clobbers): Correct vax example. Delete old
7497 example of a memory input for a string of known length. Move
7498 commentary out of table. Add a number of new examples
7499 covering array memory inputs.
7501 2017-10-12 Martin Liska <mliska@suse.cz>
7503 PR tree-optimization/82493
7504 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
7505 (test_range_functions): New function.
7506 (sbitmap_c_tests): Likewise.
7507 * selftest-run-tests.c (selftest::run_tests): Run new tests.
7508 * selftest.h (sbitmap_c_tests): New function.
7510 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
7512 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
7514 * config/rs6000/amo.h: Fix spacing issue.
7516 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7519 * config/i386/i386.md (*ashl<mode>3_mask_1,
7520 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
7521 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
7524 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
7526 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
7527 (profile_probability): Set max_probability
7528 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
7530 * profile-count.c (profile_probability::differs_from_p): Do not
7531 rely on max_probaiblity == 10000
7533 2017-10-12 Jeff Law <law@redhat.com>
7535 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
7538 2017-10-12 Martin Sebor <msebor@redhat.com>
7542 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
7543 (handle_alias_pairs): Call it.
7544 * common.opt (-Wattribute-alias): New option.
7545 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
7546 * doc/invoke.texi (-Wattribute-alias): Document.
7548 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
7551 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
7553 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
7555 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
7556 (make_hard_regno_born, make_hard_regno_dead): Update
7557 bb_killed_pseudos and bb_gen_pseudos.
7559 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
7561 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
7563 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
7565 * config/alpha/alpha.c (alpha_split_conditional_move):
7566 Use std::swap instead of manually swapping.
7567 (alpha_stdarg_optimize_hook): Ditto.
7568 (alpha_canonicalize_comparison): Ditto.
7570 2017-10-12 Bin Cheng <bin.cheng@arm.com>
7572 * tree-loop-distribution.c (struct builtin_info): New struct.
7573 (struct partition): Refactor fields into struct builtin_info.
7574 (partition_free): Free struct builtin_info.
7575 (build_size_arg_loc, build_addr_arg_loc): Delete.
7576 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
7577 information from struct builtin_info.
7578 (find_single_drs): New function refactored from classify_partition.
7579 Also moved builtin validity checks to this function.
7580 (compute_access_range, alloc_builtin): New functions.
7581 (classify_builtin_st, classify_builtin_ldst): New functions.
7582 (classify_partition): Refactor code into functions find_single_drs,
7583 classify_builtin_st and classify_builtin_ldst.
7584 (distribute_loop): Don't do runtime alias check when distributing
7586 (find_seed_stmts_for_distribution): New function.
7587 (pass_loop_distribution::execute): Refactor code finding seed
7588 stmts into above function. Support distribution for the innermost
7589 two-level loop nest. Adjust dump information.
7591 2017-10-12 Bin Cheng <bin.cheng@arm.com>
7593 * tree-loop-distribution.c: Adjust the general comment.
7594 (NUM_PARTITION_THRESHOLD): New macro.
7595 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
7596 (classify_partition): Skip builtin pattern of loop nest's inner loop.
7597 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
7598 in call to build_partition_graph.
7599 (finalize_partitions): New parameter. Make loop distribution more
7600 conservative by fusing more partitions.
7601 (distribute_loop): Don't do runtime alias check in case of loop nest
7603 (find_seed_stmts_for_distribution): New function.
7604 (prepare_perfect_loop_nest): New function.
7605 (pass_loop_distribution::execute): Refactor code finding seed stmts
7606 and loop nest into above functions. Support loop nest distribution.
7607 Adjust dump information accordingly.
7609 2017-10-12 Bin Cheng <bin.cheng@arm.com>
7611 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
7612 and set PTYPE_SEQUENTIAL for merged partition.
7614 2017-10-12 Richard Biener <rguenther@suse.de>
7616 PR tree-optimization/69728
7618 2017-09-19 Richard Biener <rguenther@suse.de>
7620 PR tree-optimization/69728
7621 * graphite-sese-to-poly.c (schedule_error): New global.
7622 (add_loop_schedule): Handle empty domain by failing the
7624 (build_original_schedule): Handle schedule_error.
7626 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
7627 domain by returning an unchanged schedule.
7629 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7631 * genrecog.c (validate_pattern): For VEC_SELECT verify that
7632 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
7634 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
7636 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
7639 2017-10-12 Jakub Jelinek <jakub@redhat.com>
7642 * expr.c (store_field): Don't optimize away bitsize == 0 store
7643 from CALL_EXPR with addressable return type.
7645 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
7647 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
7648 * config/rs6000/rs6000.md (sel): Delete mode attribute.
7649 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
7650 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
7651 TARGET_ISEL instead of TARGET_ISEL<sel>.
7653 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
7655 * config/rs6000/rs6000.c
7656 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
7658 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
7660 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
7662 (reg_or_cint_operand): Fix comment.
7663 (reg_or_zero_operand): New predicate.
7664 * config/rs6000/rs6000-protos.h (output_isel): Delete.
7665 * config/rs6000/rs6000.c (output_isel): Delete.
7666 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
7667 instead of reg_or_cint_operand. Output instruction directly (not via
7669 (isel_unsigned_<mode>): Ditto.
7670 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
7671 gpc_reg_operand. Add an instruction alternative for this. Output
7672 instruction directly.
7673 (*isel_reversed_unsigned_<mode>): Ditto.
7675 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
7677 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
7678 (TARGET_CANONICALIZE_COMPARISON): Define.
7680 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
7683 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
7684 Check whether the dest is REG before adding REG_EQUIV note.
7686 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
7689 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
7691 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
7692 (make_hard_regno_born, make_hard_regno_dead): Update
7693 bb_killed_pseudos and bb_gen_pseudos.
7695 2017-10-11 Nathan Sidwell <nathan@acm.org>
7697 * incpath.h (enum incpath_kind): Name enum, prefix values.
7698 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
7699 * incpath.c (heads, tails): Use INC_MAX.
7700 (add_env_var_paths, add_standard_paths): Use incpath_kind.
7701 (merge_include_chains, split_quote_chain,
7702 register_include_chains): Update incpath_kind names.
7703 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
7704 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
7706 (add_framework_path, darwin_register_objc_includes): Likewise.
7707 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
7709 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
7711 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
7712 Do not use float_operator operator predicate.
7713 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
7714 * config/i386/predicates.md (float_operator): Remove predicate.
7716 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
7718 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
7719 (*jccxf_i387): Ditto.
7720 (*jcc<mode>_i387): Ditto.
7721 (*jccu<mode>_i387): Ditto.
7722 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
7723 (*jcc_*_i387 splitters): Remove.
7724 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
7725 * config/i386/i386.c (ix86_split_fp_branch): Remove.
7726 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
7729 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
7731 * profile-count.h (slow_safe_scale_64bit): New function.
7732 (safe_scale_64bit): New inline.
7733 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
7734 * profile-count.c: Include wide-int.h
7735 (slow_safe_scale_64bit): New.
7737 2017-10-11 Nathan Sidwell <nathan@acm.org>
7739 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
7740 HAS_DECL_ASSEMBLER_NAME_P.
7741 * gimple-expr.c (gimple_decl_printable_name: Check
7742 HAS_DECL_ASSEMBLER_NAME_P too.
7743 * ipa-utils.h (type_in_anonymous_namespace_p): Check
7744 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
7745 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
7746 * passes.c (rest_of_decl_compilation): Check
7747 HAS_DECL_ASSEMBLER_NAME_P too.
7748 * recog.c (verify_changes): Likewise.
7749 * tree-pretty-print.c (dump_decl_name): Likewise.
7750 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
7752 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
7753 (SET_DECL_ASSEMBLER_NAME): Use it.
7754 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
7755 (COPY_DECL_ASSEMBLER_NAME): Likewise.
7756 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
7758 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
7760 * config.gcc (i386, x86_64): Add extra objects.
7761 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
7762 (ix86_min_insn_size): Declare.
7763 (ix86_issue_rate): Declare.
7764 (ix86_adjust_cost): Declare.
7765 (ia32_multipass_dfa_lookahead): Declare.
7766 (ix86_macro_fusion_p): Declare.
7767 (ix86_macro_fusion_pair_p): Declare.
7768 (ix86_bd_has_dispatch): Declare.
7769 (ix86_bd_do_dispatch): Declare.
7770 (ix86_core2i7_init_hooks): Declare.
7771 (ix86_atom_sched_reorder): Declare.
7772 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
7773 (COSTS_N_BYTES): Move to x86-tune-costs.h.
7774 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
7775 (rip_relative_addr_p): Rename to ...
7776 (ix86_rip_relative_addr_p): ... this one; export.
7777 (memory_address_length): Update.
7778 (ix86_issue_rate): Move to x86-tune-sched.c.
7779 (ix86_flags_dependent): Move to x86-tune-sched.c.
7780 (ix86_agi_dependent): Move to x86-tune-sched.c.
7781 (exact_dependency_1): Move to x86-tune-sched.c.
7782 (exact_store_load_dependency): Move to x86-tune-sched.c.
7783 (ix86_adjust_cost): Move to x86-tune-sched.c.
7784 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
7785 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
7786 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
7787 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
7788 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
7789 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
7790 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
7791 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
7792 (min_insn_size): Rename to ...
7793 (ix86_min_insn_size): ... this one; export.
7794 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
7795 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
7796 (core2i7_first_cycle_multipass_backtrack): Move to
7797 x86-tune-sched-core.c.
7798 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
7799 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
7800 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
7801 (ix86_avoid_jump_mispredicts): Update.
7802 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
7803 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
7804 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
7805 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
7806 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
7807 (MAX_INSN): Move to ix86-tune-sched-bd.c.
7808 (MAX_IMM): Move to ix86-tune-sched-bd.c.
7809 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
7810 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
7811 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
7812 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
7813 (MAX_STORE): Move to ix86-tune-sched-bd.c.
7814 (BIG): Move to ix86-tune-sched-bd.c.
7815 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
7816 (enum insn_path): Move to ix86-tune-sched-bd.c.
7817 (get_mem_group): Move to ix86-tune-sched-bd.c.
7818 (is_cmp): Move to ix86-tune-sched-bd.c.
7819 (dispatch_violation): Move to ix86-tune-sched-bd.c.
7820 (is_branch): Move to ix86-tune-sched-bd.c.
7821 (is_prefetch): Move to ix86-tune-sched-bd.c.
7822 (init_window): Move to ix86-tune-sched-bd.c.
7823 (allocate_window): Move to ix86-tune-sched-bd.c.
7824 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
7825 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
7826 (process_end_window): Move to ix86-tune-sched-bd.c.
7827 (allocate_next_window): Move to ix86-tune-sched-bd.c.
7828 (find_constant): Move to ix86-tune-sched-bd.c.
7829 (get_num_immediates): Move to ix86-tune-sched-bd.c.
7830 (has_immediate): Move to ix86-tune-sched-bd.c.
7831 (get_insn_path): Move to ix86-tune-sched-bd.c.
7832 (get_insn_group): Move to ix86-tune-sched-bd.c.
7833 (count_num_restricted): Move to ix86-tune-sched-bd.c.
7834 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
7835 (add_insn_window): Move to ix86-tune-sched-bd.c.
7836 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
7837 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
7838 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
7839 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
7840 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
7841 (do_dispatch): Move to ix86-tune-sched-bd.c.
7842 (has_dispatch): Move to ix86-tune-sched-bd.c.
7843 * config/i386/t-i386: Add new object files.
7844 * config/i386/x86-tune-costs.h: New file.
7845 * config/i386/x86-tune-sched-atom.c: New file.
7846 * config/i386/x86-tune-sched-bd.c: New file.
7847 * config/i386/x86-tune-sched-core.c: New file.
7848 * config/i386/x86-tune-sched.c: New file.
7850 2017-10-11 Liu Hao <lh_mouse@126.com>
7852 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
7853 the generic version below instead.
7854 (should_colorize): Recognize Windows consoles as terminals
7856 * pretty-print.c [__MINGW32__] (write_all): New function.
7857 [__MINGW32__] (find_esc_head): Likewise.
7858 [__MINGW32__] (find_esc_terminator): Likewise.
7859 [__MINGW32__] (eat_esc_sequence): Likewise.
7860 [__MINGW32__] (mingw_ansi_fputs): New function that handles
7862 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
7865 2017-10-11 Richard Biener <rguenther@suse.de>
7867 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
7868 Properly call analyze_scalar_evolution with the loop of the stmt.
7870 2017-10-11 Richard Biener <rguenther@suse.de>
7872 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
7873 * tree-core.h (tree_base): Add chrec_var union member.
7874 * tree.h (CHREC_VAR): Remove.
7875 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
7876 * tree-chrec.h (build_polynomial_chrec): Adjust.
7877 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
7878 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
7880 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
7882 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
7883 * match.pd: ... here.
7884 ((T) X == (T) Y): Relax condition.
7886 2017-10-11 Bin Cheng <bin.cheng@arm.com>
7888 PR tree-optimization/82472
7889 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
7891 (break_alias_scc_partitions): Update postorder number.
7893 2017-10-11 Martin Liska <mliska@suse.cz>
7896 * opts.c (parse_no_sanitize_attribute): Do not use error_value
7898 * opts.h (parse_no_sanitize_attribute): Remove last argument.
7900 2017-10-11 Martin Liska <mliska@suse.cz>
7902 * print-rtl.c (print_insn): Move declaration of idbuf
7903 to same scope as name.
7905 2017-10-11 Martin Liska <mliska@suse.cz>
7910 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
7913 2017-10-11 Martin Liska <mliska@suse.cz>
7916 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
7919 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
7921 * config.gcc (default_gnu_indirect_function): Default to yes for
7922 arm*-*-linux* with glibc.
7924 2017-10-11 Richard Biener <rguenther@suse.de>
7926 * tree-scalar-evolution.c (get_scalar_evolution): Handle
7927 default-defs and types we do not want to analyze.
7928 (interpret_loop_phi): Replace unreachable code with an assert.
7929 (compute_scalar_evolution_in_loop): Remove and inline ...
7930 (analyze_scalar_evolution_1): ... here, replacing condition with
7931 what makes the intent clearer. Remove handling of cases
7932 get_scalar_evolution now handles.
7934 2017-10-10 Jim Wilson <wilson@tuliptree.org>
7936 PR rtl-optimization/81434
7937 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
7938 comment for main loop. In sched_group_found if, also add checks for
7939 pass and min_cost_group.
7941 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
7943 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
7944 (rs6000_insn_cost): New function.
7945 * config/rs6000/rs6000.md (cost): New attribute.
7947 2017-10-10 Jakub Jelinek <jakub@redhat.com>
7948 H.J. Lu <hongjiu.lu@intel.com>
7952 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
7953 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
7954 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
7955 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
7956 to OPTION_MASK_ISA_AVX512VL - builtins that have both
7957 OPTION_MASK_ISA_MMX and some other bit set require both
7958 mmx and the ISAs without the mmx bit.
7959 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
7960 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
7961 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
7962 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
7963 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
7964 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
7965 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
7966 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
7967 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
7968 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
7969 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
7970 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
7971 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
7972 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
7973 Add OPTION_MASK_ISA_MMX.
7975 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
7977 * config.gcc (armv7*-*-freebsd*): New target.
7978 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
7980 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
7982 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
7983 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
7986 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
7988 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
7989 a template parameter.
7990 (WIDE_INT_REF_FOR): Update accordingly.
7991 * tree.h (wi::int_traits <const_tree>): Delete.
7992 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
7993 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
7994 (wi::tree_to_wide_ref): New typedef.
7995 (wi::to_wide): New function.
7996 * calls.c (get_size_range): Use wi::to_wide when operating on
7998 * cgraph.c (cgraph_node::create_thunk): Likewise.
7999 * config/i386/i386.c (ix86_data_alignment): Likewise.
8000 (ix86_local_alignment): Likewise.
8001 * dbxout.c (stabstr_O): Likewise.
8002 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
8003 * expr.c (const_vector_from_tree): Likewise.
8004 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
8005 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
8006 (fold_negate_expr_1, int_const_binop_1, const_binop)
8007 (fold_convert_const_int_from_real, optimize_bit_field_compare)
8008 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
8009 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
8010 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
8011 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
8012 (fold_not_const, round_up_loc): Likewise.
8013 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
8014 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
8015 (alloca_call_type): Likewise.
8016 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
8017 * godump.c (go_output_typedef): Likewise.
8018 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
8019 * internal-fn.c (get_min_precision): Likewise.
8020 * ipa-cp.c (ipcp_store_vr_results): Likewise.
8021 * ipa-polymorphic-call.c
8022 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
8023 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
8024 (ipa_modify_call_arguments): Likewise.
8025 * match.pd: Likewise.
8026 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
8027 * print-tree.c (print_node_brief, print_node): Likewise.
8028 * stmt.c (expand_case): Likewise.
8029 * stor-layout.c (layout_type): Likewise.
8030 * tree-affine.c (tree_to_aff_combination): Likewise.
8031 * tree-cfg.c (group_case_labels_stmt): Likewise.
8032 * tree-data-ref.c (dr_analyze_indices): Likewise.
8033 (prune_runtime_alias_test_list): Likewise.
8034 * tree-dump.c (dequeue_and_dump): Likewise.
8035 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
8036 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
8037 * tree-pretty-print.c (dump_generic_node): Likewise.
8038 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
8039 (simple_iv_with_niters): Likewise.
8040 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
8041 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
8042 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
8043 * tree-ssa-loop-niter.c (split_to_var_and_offset)
8044 (refine_value_range_using_guard, number_of_iterations_ne_max)
8045 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
8046 (get_cst_init_from_scev, record_nonwrapping_iv)
8047 (scev_var_range_cant_overflow): Likewise.
8048 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
8049 * tree-ssa-pre.c (compute_avail): Likewise.
8050 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
8051 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
8052 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
8053 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
8054 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
8055 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
8056 (dump_case_nodes, try_switch_expansion): Likewise.
8057 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
8058 (vect_do_peeling): Likewise.
8059 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
8060 * tree-vect-stmts.c (vectorizable_load): Likewise.
8061 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
8062 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
8063 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
8064 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
8065 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
8066 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
8067 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
8068 (evrp_dom_walker::before_dom_children): Likewise.
8069 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
8070 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
8071 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
8072 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
8073 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
8074 (get_range_pos_neg): Likewise.
8075 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
8076 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
8077 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
8078 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
8079 * config/avr/avr.c (avr_fold_builtin): Likewise.
8080 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
8081 * config/msp430/msp430.c (msp430_attr): Likewise.
8082 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
8083 * config/powerpcspe/powerpcspe-c.c
8084 (altivec_resolve_overloaded_builtin): Likewise.
8085 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
8086 (rs6000_expand_ternop_builtin): Likewise.
8087 * config/rs6000/rs6000-c.c
8088 (altivec_resolve_overloaded_builtin): Likewise.
8089 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
8090 (rs6000_expand_ternop_builtin): Likewise.
8091 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
8093 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8095 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
8096 when copying loop nest with only one inner loop.
8098 2017-10-10 Richard Biener <rguenther@suse.de>
8100 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
8101 blocks if SCEV is active.
8102 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
8104 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
8105 (scev_initialize): Assert we are not yet initialized.
8107 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8109 * tree-loop-distribution.c (generate_loops_for_partition): Remove
8110 inner loop's exit stmt by making it always exit the loop, otherwise
8111 we would generate an infinite empty loop.
8113 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8115 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
8116 renaming variables in new preheader if it's deleted.
8118 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8120 * tree-loop-distribution.c (struct partition): Remove unused field
8121 loops of the structure.
8122 (partition_alloc, partition_free): Ditto.
8123 (build_rdg_partition_for_vertex): Ditto.
8125 2017-10-09 Jeff Law <law@redhat.com>
8127 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
8128 return type to match prototype and documentation.
8130 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8132 * config/rs6000/rs6000.c (processor_costs): Move to ...
8133 * config/rs6000/rs6000.h: ... here.
8134 (rs6000_cost): Declare.
8136 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
8138 * except.c (setjmp_fn): New global variable.
8139 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
8140 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
8141 if DONT_USE_BUILTIN_SETJMP is defined.
8143 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8145 * target.def (insn_cost): New hook.
8146 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
8147 * doc/tm.texi: Regenerate.
8148 * rtlanal.c (insn_cost): Use the new hook.
8150 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8152 * combine.c (combine_validate_cost): Compute the new insn_cost,
8153 not just pattern_cost.
8154 (try_combine): Adjust comment.
8156 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8158 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
8160 * combine.c (uid_insn_cost): Adjust comment.
8161 (combine_validate_cost): Adjust comment. Use pattern_cost instead
8163 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
8164 * dse.c (find_shift_sequence): Ditto.
8165 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
8166 (bb_valid_for_noce_process_p): Use pattern_cost.
8167 * rtl.h (insn_rtx_cost): Delete.
8168 (pattern_cost): New prototype.
8169 (insn_cost): New prototype.
8170 * rtlanal.c (insn_rtx_cost): Rename to...
8171 (pattern_cost): ... this.
8174 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
8176 * config/i386/i386.md (*jcc_2): Remove insn pattern.
8177 (*jcc<mode>_0_r_i387): Ditto.
8178 (*jccxf_r_i387): Ditto.
8179 (*jcc<mode>_r_i387): Ditto.
8180 (*jccu<mode>_r_i387): Ditto.
8181 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
8182 (*jcc): Rename from *jcc_1.
8184 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8186 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
8187 deferred rescans after the lvx/stvx recombination pre-pass.
8189 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
8191 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
8192 memory operation instruction support.
8193 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
8194 (rs6000-ibm-aix[789]*): Likewise.
8195 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
8196 Document new functions.
8198 2017-10-09 Richard Biener <rguenther@suse.de>
8200 PR tree-optimization/82397
8201 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
8202 equality only for semantically equal trees.
8204 2017-10-09 Richard Biener <rguenther@suse.de>
8206 PR tree-optimization/82449
8207 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
8208 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
8209 allow constant addresses.
8210 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
8213 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8215 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
8218 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8221 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
8224 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8227 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
8229 2017-10-09 Jakub Jelinek <jakub@redhat.com>
8232 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
8233 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
8235 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
8237 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
8238 (WI_BINARY_PREDICATE_RESULT): Likewise.
8239 (wi::binary_traits::operator_result): New type.
8240 (wi::binary_traits::predicate_result): Likewise.
8241 (generic_wide_int::operator~, unary generic_wide_int::operator-)
8242 (generic_wide_int::operator==, generic_wide_int::operator!=)
8243 (generic_wide_int::operator&, generic_wide_int::and_not)
8244 (generic_wide_int::operator|, generic_wide_int::or_not)
8245 (generic_wide_int::operator^, generic_wide_int::operator+
8246 (binary generic_wide_int::operator-, generic_wide_int::operator*):
8248 (operator~, unary operator-, operator==, operator!=, operator&)
8249 (operator|, operator^, operator+, binary operator-, operator*): New
8251 * expr.c (get_inner_reference): Use wi::bit_and_not.
8252 * fold-const.c (fold_binary_loc): Likewise.
8253 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
8254 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
8255 (bit_value_binop): Likewise.
8256 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
8257 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
8258 (extract_range_from_binary_expr_1): Likewise.
8259 (masked_increment): Likewise.
8260 (simplify_bit_ops_using_ranges): Likewise.
8262 2017-10-09 Martin Jambor <mjambor@suse.cz>
8265 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
8266 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
8267 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
8269 (hsa_fixup_mov_insn_type): New function.
8270 (hsa_op_with_type::get_in_type): Use it.
8271 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
8272 immediates in an assert.
8273 (hsa_op_with_type::extend_int_to_32bit): New method.
8274 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
8275 types. Convert to dest type if necessary.
8276 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
8277 (reg_for_gimple_ssa): Pass false as min32int to
8278 hsa_type_for_scalar_tree_type.
8279 (gen_hsa_addr): Fixup type when creating addresable temporary.
8280 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
8281 (gen_hsa_unary_operation): Extend operands and convert to dest type if
8282 necessary. Call hsa_fixup_mov_insn_type.
8283 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
8284 extend operands and convert to dest type if necessary.
8285 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
8286 to dest type if necessary.
8287 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
8288 if conversion nt necessary and size matches.
8289 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
8290 to dest type if necessary.
8291 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
8292 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
8294 (gen_hsa_clrsb): Likewise.
8295 (gen_hsa_ffs): Likewise.
8296 (gen_hsa_divmod): Extend operands and convert to dest type if
8298 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
8300 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
8302 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
8303 Remove empty default arguments. Use a brace block as output
8305 (conditional return): Ditto.
8307 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
8308 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
8310 (group_ending_nop): Ditto.
8311 (doloop_end): Ditto.
8312 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
8313 (splitters for those): Ditto.
8315 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
8317 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
8318 a conditional jump (and the compare for it) so that pc_rtx is the
8320 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
8321 for the deleted and renamed ctr<mode>_internal[234] patterns.
8322 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
8323 Delete second conditional return pattern.
8324 (ctr<mode>_internal2): Delete this second bdnz pattern.
8325 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
8326 (ctr<mode>_internal4): Delete this second bdz pattern.
8328 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
8330 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
8331 (always_initialized_rtx_for_ssa_name_p): New predicate.
8332 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
8333 (finish_out_of_ssa): Free new field of SA.
8334 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
8335 * tree-ssa-coalesce.c: Include tree-ssa.h.
8336 (get_parm_default_def_partitions): Remove extern keyword.
8337 (get_undefined_value_partitions): New function.
8338 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
8339 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
8341 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
8343 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
8345 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
8347 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
8349 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
8350 for targets that preffer 128bit.
8352 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
8354 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
8356 2017-10-08 Olivier Hainque <hainque@adacore.com>
8358 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
8359 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
8361 (thumb_set_return_address): Likewise.
8363 2017-10-08 Olivier Hainque <hainque@adacore.com>
8365 * common/config/arm/arm-common.c (arm_except_unwind_info):
8366 Handle DWARF2_UNWIND_INFO.
8368 2017-10-07 Michael Collison <michael.collison@arm.com>
8370 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
8373 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
8375 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
8376 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
8377 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
8378 defined, force the creation of a new block for a dispatch label.
8380 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
8382 * invoke.texi (Wsuggest-attribute=cold): Document.
8383 * common.opt (Wsuggest-attribute=cold): New
8384 * ipa-pure-const.c (warn_function_cold): New function.
8385 * predict.c (compute_function_frequency): Use it.
8386 * predict.h (warn_function_cold): Declare.
8388 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
8390 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
8393 2017-10-06 Martin Liska <mliska@suse.cz>
8395 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
8396 keyword for member functions.
8397 (struct sanopt_tree_couple): New struct.
8398 (struct sanopt_tree_couple_hash): New function.
8399 (struct sanopt_ctx): Add new hash_map.
8400 (has_dominating_ubsan_ptr_check): New function.
8401 (record_ubsan_ptr_check_stmt): Likewise.
8402 (maybe_optimize_ubsan_ptr_ifn): Likewise.
8403 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
8404 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
8406 2017-10-06 Sudakshina Das <sudi.das@arm.com>
8409 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
8410 aarch64_simd_valid_immediate on CONST_VECTORs.
8411 (aarch64_reg_or_bic_imm): Likewise.
8413 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
8415 PR rtl-optimization/82396
8416 * haifa-sched.c (ready_sort_real): Disable qsort checking.
8418 2017-10-06 Sebastian Pop <sebpop@gmail.com>
8420 * graphite-dependences.c (scop_get_reads): Move code to...
8421 (scop_get_must_writes): Move code to...
8422 (scop_get_may_writes): Move code to...
8423 (scop_get_reads_and_writes): ... here.
8424 (scop_get_dependences): Call scop_get_reads_and_writes.
8426 2017-10-06 Jakub Jelinek <jakub@redhat.com>
8428 PR tree-optimization/82434
8429 * fold-const.h (can_native_encode_type_p,
8430 can_native_encode_string_p): Remove.
8431 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
8432 don't encode anything, just return what would be otherwise returned.
8433 (native_encode_fixed, native_encode_complex, native_encode_vector):
8435 (native_encode_string): Likewise. Inline by hand
8436 can_native_encode_string_p.
8437 (can_native_encode_type_p): Remove.
8438 (can_native_encode_string_p): Remove.
8439 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
8440 STRING_CSTs using can_native_encode_string_p, test all
8441 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
8442 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
8443 argument from native_encode_expr.
8444 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
8445 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
8448 2017-10-06 Richard Biener <rguenther@suse.de>
8450 PR tree-optimization/82397
8451 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
8452 operand_equal_p but rely on data_ref_compare_tree for detecting
8454 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
8455 to match up with dr_group_sort_cmp.
8457 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8460 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
8462 * config/s390/s390-builtin-types.def: Regenerate.
8464 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8467 * config/s390/s390-builtin-types.def: Regenerate.
8468 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
8469 Change flag from B_VXE to B_VX.
8470 (s390_vec_min_dbl): Remove B_VXE flag.
8472 2017-10-06 Richard Biener <rguenther@suse.de>
8474 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
8475 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
8476 translate_isl_ast_to_gimple::is_valid_rename,
8477 translate_isl_ast_to_gimple::get_rename,
8478 translate_isl_ast_to_gimple::get_def_bb_for_const,
8479 translate_isl_ast_to_gimple::get_new_name,
8480 translate_isl_ast_to_gimple::collect_all_ssa_names,
8481 translate_isl_ast_to_gimple::copy_loop_phi_args,
8482 translate_isl_ast_to_gimple::collect_all_ssa_names,
8483 translate_isl_ast_to_gimple::copy_loop_phi_args,
8484 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
8485 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
8486 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
8487 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
8488 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
8489 translate_isl_ast_to_gimple::copy_cond_phi_args,
8490 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
8491 translate_isl_ast_to_gimple::edge_for_new_close_phis,
8492 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
8493 translate_isl_ast_to_gimple::rename_uses,
8494 translate_isl_ast_to_gimple::rename_all_uses): Remove.
8495 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
8496 (set_rename_for_each_def): Likewise.
8497 (graphite_copy_stmts_from_block): Handle debug stmt resetting
8498 here. Handle rewriting SCEV analyzable uses here.
8499 (copy_bb_and_scalar_dependences): Generate code for PHI
8501 (graphite_regenerate_ast_isl): Adjust.
8502 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
8503 (add_write, add_read): New functions.
8504 (build_cross_bb_scalars_def): Use it and simplify.
8505 (build_cross_bb_scalars_use): Likewise.
8506 (graphite_find_cross_bb_scalar_vars): Inline into...
8507 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
8508 simulating out-of-SSA. Compute liveout and add dependencies.
8509 (build_scops): Force an empty entry block.
8510 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
8512 (sese_build_liveouts): Declare.
8513 (sese_trivially_empty_bb_p): Likewise.
8514 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
8515 compute liveout and debug_liveout.
8516 (sese_bad_liveouts_use): Remove.
8517 (sese_reset_debug_liveouts_bb): Likewise.
8518 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
8519 (sese_build_liveouts): Build liveout and debug_liveout and store
8521 (new_sese_info): Adjust.
8522 (free_sese_info): Likewise.
8523 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
8524 do not build liveout here.
8525 (move_sese_in_condition): Adjust region entry.
8526 (scev_analyzable_p): Match up with chrec_apply requirements.
8527 (sese_trivially_empty_bb_p): New.
8528 * tree-into-ssa.c (get_reaching_def): Properly support generating
8529 default-defs for incremental rewrite of anonymous names.
8531 2017-10-06 Richard Biener <rguenther@suse.de>
8533 * graphite-sese-to-poly.c (extract_affine): For casts increasing
8534 precision do not perform modulo reduction.
8536 2017-10-06 Richard Biener <rguenther@suse.de>
8538 PR tree-optimization/82436
8539 * tree-vect-slp.c (vect_supported_load_permutation_p): More
8540 conservatively choose the vectorization factor when checking
8541 whether we can perform the required load permutation.
8542 (vect_transform_slp_perm_load): Assert when we may not fail.
8544 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
8546 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
8547 message for incompatible -msdata=* and -mcall-* options.
8549 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
8551 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
8552 rate for post-reload scheduling.
8554 2017-10-05 Tamar Christina <tamar.christina@arm.com>
8556 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
8558 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
8560 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
8561 to improve monte carlo in scimark.
8563 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
8565 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
8566 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
8567 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
8568 pentium4_cost, nocona_cost): Set reassociation width to 1.
8569 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
8570 width to 2 for fp operations and 1 otherwise.
8571 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
8573 (atom_cost): Set reassociation width to 2.
8574 (slm_cost, generic_cost): Set fp reassociation width
8575 to 2 and 1 otherwise.
8576 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
8577 (core_cost): Set fp reassociation width to 4 and vector to 2.
8578 (ix86_reassociation_width): Rewrite using cost table; special case
8579 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
8580 and TARGET_AVX128_OPTIMAL.
8581 * config/i386/i386.h (processor_costs): Add
8582 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
8583 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
8584 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
8585 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
8586 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
8587 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
8589 2017-10-05 Nathan Sidwell <nathan@acm.org>
8591 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
8593 2017-10-05 Tamar Christina <tamar.christina@arm.com>
8595 * config/arm/arm.c (arm_test_fpu_data): New.
8596 (arm_run_selftests): Call arm_test_fpu_data.
8598 2017-10-04 Nathan Sidwell <nathan@acm.org>
8600 * toplev.c (toplev::main): Remove excess parens on pretty_printer
8602 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
8604 2017-10-04 Sudakshina Das <sudi.das@arm.com>
8606 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
8607 check type for aarch64_simd_valid_immediate.
8608 (aarch64_output_simd_mov_immediate): Update prototype.
8609 (aarch64_simd_valid_immediate): Update prototype.
8610 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
8611 support for ORR-immediate.
8612 (and<mode>3): modified pattern to add support for BIC-immediate.
8613 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
8614 now checks for valid immediate for BIC and ORR based on new enum
8616 (aarch64_output_simd_mov_immediate): Function now used to output
8617 BIC/ORR imm as well based on new enum argument.
8618 * config/aarch64/constraints.md (Do): New vector immediate constraint.
8620 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
8621 (aarch64_reg_or_bic_imm): Likewise.
8623 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8625 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
8626 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
8628 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
8632 PR rtl-optimization/82396
8633 * haifa-sched.c (autopref_multipass_init): Simplify
8635 (autopref_rank_data): Simplify sort order.
8636 * sched-int.h (autopref_multipass_data_): Remove
8637 multi_mem_insn_p, min_offset and max_offset.
8639 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8641 * doc/sourcebuild.texi: Document vect_peeling_profitable.
8643 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8645 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
8648 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8650 * doc/sourcebuild.texi: Document vect_long_mult.
8652 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
8654 PR tree-optimization/82413
8655 * fold-const.c (build_range_check): Use widest_int when comparing
8656 the maximum ETYPE value with HIGH.
8658 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
8660 PR rtl-optimization/82396
8661 * haifa-sched.c (autopref_multipass_init): Simplify
8663 (autopref_rank_data): Simplify sort order.
8664 * sched-int.h (autopref_multipass_data_): Remove
8665 multi_mem_insn_p, min_offset and max_offset.
8667 2017-10-04 Jakub Jelinek <jakub@redhat.com>
8669 PR tree-optimization/82381
8670 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
8671 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
8674 PR tree-optimization/82374
8675 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
8676 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
8677 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
8678 current_function_decl to the new decl.
8680 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
8682 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
8683 helper macro for IEEE float128 hardware built-in functions.
8684 (SQRTF128_ODD): Add built-in functions with the round-to-odd
8686 (TRUNCF128_ODD): Likewise.
8687 (ADDF128_ODD): Likewise.
8688 (SUBF128_ODD): Likewise.
8689 (MULF128_ODD): Likewise.
8690 (DIVF128_ODD): Likewise.
8691 (FMAF128_ODD): Likewise.
8692 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
8693 UNSPEC_TRUNC_ROUND_TO_ODD.
8694 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
8695 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
8696 floating point round to odd instructions.
8697 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
8698 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
8699 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
8700 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
8701 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
8702 (trunc<mode>sf2_hw): Change the truncate with round to odd
8703 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
8704 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
8705 to odd hardware instructions.
8706 (sub<mode>3_odd): Likewise.
8707 (mul<mode>3_odd): Likewise.
8708 (div<mode>3_odd): Likewise.
8709 (sqrt<mode>2_odd): Likewise.
8710 (fma<mode>4_odd): Likewise.
8711 (fms<mode>4_odd): Likewise.
8712 (nfma<mode>4_odd): Likewise.
8713 (nfms<mode>4_odd): Likewise.
8714 (trunc<mode>df2_odd): Change the truncate with round to odd
8715 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
8717 * doc/extend.texi (PowerPC built-in functions): Update documentation
8718 for existing IEEE float128-bit built-in functions. Add built-in
8719 functions that generate the IEEE 128-bit floating point round to
8722 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
8724 PR rtl-optimization/77729
8725 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
8726 to (X&(C1&~C2))|C2 transformations.
8728 2017-10-03 Martin Jambor <mjambor@suse.cz>
8730 PR tree-optimization/82363
8731 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
8732 mismatch, mark lacc written regardless of racc.
8734 2017-10-03 Jakub Jelinek <jakub@redhat.com>
8736 PR tree-optimization/82381
8737 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
8738 stmt_to_insert nor wheather SSA_NAMEs are default defs.
8739 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
8740 fallthrough into reassoc_stmt_dominates_stmt_p.
8743 * combine.c (combine_instructions): Don't combine in unreachable
8746 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
8749 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
8750 function to not use the have_cpu variable. Do not set cpu_index,
8751 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
8753 (rs6000_valid_attribute_p): Remove duplicate initializations of
8754 old_optimize and func_optimize.
8755 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
8756 (rs6000_activate_target_options): Make global.
8757 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
8760 2017-10-02 Jakub Jelinek <jakub@redhat.com>
8762 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
8763 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
8764 Set *poffset to 0 and *psize and *pmax_size to -1 if
8765 *poffset + *psize overflows in HOST_WIDE_INT.
8767 PR tree-optimization/82387
8768 PR tree-optimization/82388
8769 PR tree-optimization/82389
8770 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
8771 instead of live_bytes non-NULL.
8773 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
8776 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
8779 2017-10-02 Richard Biener <rguenther@suse.de>
8781 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
8782 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
8783 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
8785 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
8787 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
8788 requested precision matches the type's.
8789 * calls.c (alloc_max_size): Calculate the new candidate size as
8790 a widest_int and use wi::to_widest when comparing it with the
8791 current candidate size.
8792 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
8793 zero rather than integer_zero_node.
8794 * match.pd: Check for a no-op conversion before using wi::add
8795 rather than after. Use tree_to_uhwi when summing small shift
8796 counts into an unsigned int.
8798 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
8799 Alan Hayward <alan.hayward@arm.com>
8800 David Sherwood <david.sherwood@arm.com>
8803 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
8804 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
8805 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
8806 POINTER_AND_FP_REGS.
8808 2017-10-02 Richard Biener <rguenther@suse.de>
8810 PR tree-optimization/82355
8811 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
8812 a mapping for the enclosing loop but avoid generating one for
8814 (copy_bb_and_scalar_dependences): Remove premature codegen
8815 error on PHIs in blocks duplicated into multiple places.
8816 * graphite-scop-detection.c
8817 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
8818 in the region use it as loop and nest to analyze the DR in.
8819 (try_generate_gimple_bb): Likewise.
8820 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
8821 (add_loop_constraints): For blocks in a loop not in the region
8822 create a dimension with a single iteration.
8823 * sese.h (gbb_loop_at_index): Remove assert.
8825 2017-10-01 Kevin Buettner <kevinb@redhat.com>
8827 * omp-expand.c (adjust_context_scope): New function.
8828 (expand_parallel_call): Call adjust_context_scope.
8830 2017-10-01 Jeff Law <law@redhat.com>
8832 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
8833 dom_opt_dom_walker class with direct access to private members.
8834 Add comments. Call test_for_singularity.
8835 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
8836 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
8837 m_dummy_cond anymore.
8838 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
8840 (pass_dominator:execute): Build the dummy_cond here and pass it
8841 to the dom_opt_dom_walker ctor.
8842 (test_for_singularity): New function.
8844 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
8845 Maya Rashish <coypu@sdf.org>
8847 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
8848 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
8849 (alpha*-*-netbsd*) Use nbsd_tm_file.
8850 (arm*-*-netbsdelf*) Likewise.
8851 (i[34567]86-*-netbsdelf*) Likewise.
8852 (x86_64-*-netbsd*) Likewise.
8853 (mips*-*-netbsd*) Likewise.
8854 (powerpc-*-netbsd*) Likewise.
8855 (sh*-*-netbsd*) Likewise.
8856 (sparc-*-netbsdelf*) Likewise.
8857 (sparc64-*-netbsd*) Likewise.
8858 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
8860 (vax-*-netbsdelf*) Likewise.
8861 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
8862 (UINT_FAST8_TYPE) Likewise.
8863 (INT_FAST16_TYPE) Check CHAR_FAST16.
8864 (UINT_FAST16_TYPE) Likewise.
8866 2017-09-30 Jakub Jelinek <jakub@redhat.com>
8869 * config/i386/i386.md
8870 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
8871 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
8872 *divmodsi4_zext_2): New define_insn_and_split.
8873 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
8874 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
8875 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
8876 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
8877 New define_insn_and_split.
8878 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
8879 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
8880 operands[1] having DImode when mode is SImode.
8882 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
8883 always SImode for DIV and MOD in REG_EQUAL notes.
8885 2017-09-29 Yury Gribov <tetra2005@gmail.com>
8888 * match.pd: Fix handling of NaNs in pattern.
8890 2017-09-29 Jeff Law <law@redhat.com>
8892 * sbitmap.c (bitmap_bit_in_range_p): New function.
8893 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
8894 * tree-ssa-dse.c (live_bytes_read): New function.
8895 (dse_classify_store): Ignore reads of dead bytes.
8897 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
8898 typos and whitespace errors.
8899 * config/i386/predicates.md (address_no_seg_operand): Likewise.
8900 * config/s390/s390.c (s390_emit_prologue): Likewise.
8902 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
8905 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
8906 with a symbol for LRA.
8908 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
8910 PR rtl-optimization/82338
8911 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
8913 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
8915 * genmodes.c (calc_wider_mode): Suppress qsort macro.
8916 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
8917 (qsort_chk): Declare.
8918 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
8919 (qsort_chk): New function.
8921 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8923 PR tree-optimization/82337
8924 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
8925 phi definition if the PHI result appears in an abnormal PHI.
8926 (find_basis_for_base_expr): Don't record a basis if the LHS of the
8927 basis appears in an abnormal PHI.
8929 2017-09-29 Richard Biener <rguenther@suse.de>
8931 * graphite-isl-ast-to-gimple.c
8932 (translate_isl_ast_to_gimple::set_codegen_error): New function.
8933 (binary_op_to_tree): Use it.
8934 (get_rename_from_scev): Likewise.
8935 (copy_loop_phi_nodes): Likewise.
8936 (copy_bb_and_scalar_dependences): Likewise.
8937 (translate_pending_phi_nodes): Likewise.
8939 2017-09-29 Jakub Jelinek <jakub@redhat.com>
8942 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
8943 for movabsq $(i32 << shift), r64.
8945 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
8947 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
8948 index when encoding %esp as %rsp to avoid 0x67 prefix.
8950 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
8952 * config/i386/i386.md (*movsf_internal, *movdf_internal):
8953 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
8955 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
8957 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
8958 Extensions with more than 16 double VFP registers.
8959 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
8960 to_clear_mask and all code related to it. Replace the remaining
8961 entry by a sbitmap and adapt code accordingly.
8963 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
8965 * brig-builtins.def: Change pure attributes to const.
8967 2017-09-28 Joseph Myers <joseph@codesourcery.com>
8969 * config.gcc (default_gnu_indirect_function): Default to yes for
8970 sparc*-*-linux* with glibc.
8972 2017-09-28 Joseph Myers <joseph@codesourcery.com>
8974 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
8975 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
8976 when creating .init_array and .fini_array sections with priority
8979 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
8982 * config/aarch64/aarch64.c
8983 (aarch64_builtin_support_vector_misalignment): Always return false
8984 when misalignment is unknown.
8986 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
8988 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
8989 this function to return false if the definition used by the swap
8990 instruction is artificial, or if the memory address from which the
8991 constant value is loaded is not represented by a base address held
8992 in a register or if the base address register is a frame or stack
8993 pointer. Additionally, return false if the base address of the
8994 loaded constant is a SYMBOL_REF but is not considered to be a
8996 (replace_swapped_load_constant): New function.
8997 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
8998 loaded constant vector with a load of a swapped constant vector.
9000 2017-09-27 Carl Love <cel@us.ibm.com>
9002 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
9003 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
9004 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
9007 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
9009 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
9010 first, always call autopref_rank_data otherwise.
9012 2017-09-27 Richard Biener <rguenther@suse.de>
9014 * graphite-scop-detection.c (find_scop_parameters): Move
9015 loop bound handling ...
9016 (gather_bbs::before_dom_children): ... here, avoiding the need
9017 to build scop_info->loop_nest.
9018 (record_loop_in_sese): Remove.
9019 * sese.h (sese_info_t::loop_nest): Remove.
9020 * sese.c (new_sese_info): Do not allocate loop_nest.
9021 (free_sese_info): Do not free loop_nest.
9023 2017-09-27 Jakub Jelinek <jakub@redhat.com>
9026 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
9027 lhs from calls if the lhs has addressable type.
9029 2017-09-27 Richard Biener <rguenther@suse.de>
9031 * graphite.h (scop::max_alias_set): New member.
9032 * graphite-scop-detection.c: Remove references to non-existing
9033 --param in comments.
9034 (build_alias_sets): Record the maximum alias set used for drs.
9035 (build_scops): Support zero as unlimited for
9036 --param graphite-max-arrays-per-scop.
9037 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
9039 (build_poly_sr_1): ... here. Compute alias set based on the
9040 maximum alias set used for drs rather than
9041 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
9043 2017-09-27 Richard Biener <rguenther@suse.de>
9045 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
9046 --param loop-block-tile-size=0 to disable tiling.
9048 2017-09-27 Richard Biener <rguenther@suse.de>
9050 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
9051 (graphite-max-nb-scop-params): Document special value zero.
9052 * domwalk.h (dom_walker::STOP): New symbolical constant.
9053 (dom_walker::dom_walker): Add optional parameter for bb to
9055 (dom_walker::~dom_walker): Declare.
9056 (dom_walker::before_dom_children): Document STOP return value.
9057 (dom_walker::m_user_bb_to_rpo): New member.
9058 (dom_walker::m_bb_to_rpo): Likewise.
9059 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
9060 mapping here if not provided by the user.
9061 (dom_walker::~dom_walker): Free bb to RPO mapping if not
9062 provided by the user.
9063 (dom_walker::STOP): Define.
9064 (dom_walker::walk): Do not compute bb to RPO mapping here.
9065 Support STOP return value from before_dom_children to stop
9067 * graphite-optimize-isl.c (optimize_isl): If the schedule
9068 is the same still generate code if -fgraphite-identity
9069 or -floop-parallelize-all are given.
9070 * graphite-scop-detection.c: Include cfganal.h.
9071 (gather_bbs::gather_bbs): Get and pass through bb to RPO
9073 (gather_bbs::before_dom_children): Return STOP for BBs
9075 (build_scops): Compute bb to RPO mapping and pass it to
9076 the domwalk. Treat --param graphite-max-nb-scop-params=0
9077 as not limiting the number of params.
9078 * graphite.c (graphite_initialize): Remove limit on the
9079 number of basic-blocks in a function.
9080 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
9081 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
9082 default value of 10.
9084 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9086 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
9087 Adjust code to eliminate needing to do the shift right 32-bits
9088 operation after XSCVDPSPN.
9090 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9092 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
9093 ((X / Y) != 0 -> X >= Y): Likewise.
9095 2017-09-26 Carl Love <cel@us.ibm.com>
9097 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
9098 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
9099 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
9100 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
9101 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
9102 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
9103 definitions and overloading.
9104 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
9105 statement for P9V_BUILTIN_XST_LEN_R.
9106 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
9107 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
9108 define_expand and define_insn for the instructions and builtins.
9109 * doc/extend.texi: Update the built-in documentation file for the new
9111 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
9112 define_insn for the instructions
9114 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
9117 * gcc/config/netbsd-protos.h: New file.
9118 * gcc/config/netbsd.c: New file.
9119 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
9120 * gcc/config/t-netbsd: New file.
9121 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
9122 (tmake_file) Add t-netbsd.
9123 (extra_objs) Add netbsd.o.
9125 2017-09-26 Janus Weil <janus@gcc.gnu.org>
9129 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
9131 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9133 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
9134 sign extension from a vector register to a GPR by doing a 32-bit
9135 direct move and then an EXTSW.
9136 (extendsi<mode>2 splitter): Likewise.
9137 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
9138 right or vector extract after doing XSCVDPSPN. Use
9139 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
9141 (movdi_from_sf_zero_ext): Likewise.
9142 (reload_gpr_from_vsxsf): Likewise.
9143 (p8_mfvsrd_4_disf): Delete, no longer used.
9144 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
9145 then needing to move the SFmode to a GPR to use the XSCVDPSP
9146 instruction instead of FRSP and XSCVDPSPN.
9147 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
9148 it is adjacent to the other XSCVSPDP insns.
9149 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
9150 SFmode to be in traditional Altivec registers.
9151 (vsx_xscvdpspn): Eliminate useless alternative constraint.
9152 (vsx_xscvspdpn): Likewise.
9153 (vsx_xscvspdpn_scalar): Likewise.
9155 2017-09-26 Martin Jambor <mjambor@suse.cz>
9157 * tree-sra.c (compare_access_positions): Put integral types first,
9158 stabilize sorting of integral types, remove conditions putting
9159 non-full-precision integers last.
9160 (sort_and_splice_var_accesses): Disable scalarization if a
9161 non-integert would be represented by a non-full-precision integer.
9163 2017-09-26 Joseph Myers <joseph@codesourcery.com>
9165 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
9166 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
9167 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
9168 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
9169 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
9170 conditionals inside the function instead of around it. Call
9171 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
9172 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
9174 2017-09-26 Richard Biener <rguenther@suse.de>
9176 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
9178 (scop_detection::build_scop_breadth): ... this. Removed.
9179 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
9180 (scop_detection::harmful_stmt_in_bb): Likewise.
9181 (scop_detection::graphite_can_represent_stmt): Likewise.
9182 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
9183 (scop_detection::can_represent_loop): Remove recursion, fold in ...
9184 (scop_detection::can_represent_loop_1): ... this. Removed.
9185 (scop_detection::harmful_loop_in_region): Simplify after inlining
9186 the above and remove more quadraticness.
9187 (build_scops): Adjust.
9188 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
9191 2017-09-26 Jakub Jelinek <jakub@redhat.com>
9194 * config/i386/i386.c (ix86_print_operand_address_as): Only test
9195 REGNO (base) == SP_REG if base is a REG.
9198 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
9199 if it is different SSA_NAME.
9200 (optimize_range_tests_cmp_bitwise): New function.
9201 (optimize_range_tests): Call it.
9203 2017-09-26 Richard Biener <rguenther@suse.de>
9205 PR tree-optimization/82321
9206 * graphite.c (canonicalize_loop_closed_ssa): Properly check
9207 for the def being inside the loop.
9209 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9211 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
9213 * config/s390/s390-builtins.def: Fix constraint on op4.
9215 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9217 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
9218 independent expanders.
9219 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
9220 ("vec_ordered", "vec_unordered"): New expanders.
9222 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9224 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
9227 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9229 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
9230 vec_unpacks_lo_v16qi.
9231 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
9233 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9235 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
9236 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
9237 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
9239 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9241 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
9243 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
9245 ("*vec_slb<mode>"): New insn pattern.
9246 ("vec_shr_<mode>"): New expander.
9247 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
9248 and force the shift count operand to V16QImode.
9249 ("vec_srb<mode>"): Set shift count mode to V16QI.
9251 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9253 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
9254 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
9255 ("vec_widen_smult_hi_<mode>"): New expander definitions.
9257 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
9260 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
9262 2017-09-26 Richard Biener <rguenther@suse.de>
9264 PR tree-optimization/82320
9265 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
9268 2017-09-25 Jeff Law <law@redhat.com>
9270 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
9271 prototype for new argument.
9272 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
9273 mostly extracted from rs6000_emit_allocate_stack.
9274 (rs6000_emit_probe_stack_range_stack_clash): New function.
9275 (rs6000_emit_allocate_stack): Call
9276 rs6000_emit_probe_stack_range_stack_clash as needed.
9277 (rs6000_emit_probe_stack_range): Add additional argument
9278 to call to gen_probe_stack_range{si,di}.
9279 (output_probe_stack_range): New.
9280 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
9281 (output_probe_stack_range_stack_clash): New.
9282 (rs6000_emit_prologue): Emit notes into dump file as requested.
9283 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
9284 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
9285 Add additional operand and pass it to output_probe_stack_range.
9287 2017-09-25 Bin Cheng <bin.cheng@arm.com>
9289 PR tree-optimization/82163
9290 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
9291 (checking_verify_loop_closed_ssa): New parameter.
9292 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
9293 (check_loop_closed_ssa_stmt): Delete.
9294 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
9295 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
9296 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
9299 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
9301 * brig-builtins.def: Treat HSAIL barrier builtins as
9302 setjmp/longjump style functions.
9304 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
9306 * target.def (constant_alignment): New hook.
9307 * defaults.h (CONSTANT_ALIGNMENT): Delete.
9308 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
9309 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
9310 * doc/tm.texi: Regenerate.
9311 * targhooks.h (default_constant_alignment): Declare.
9312 (constant_alignment_word_strings): Likewise.
9313 * targhooks.c (default_constant_alignment): New function.
9314 (constant_alignment_word_strings): Likewise.
9315 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
9316 instead of CONSTANT_ALIGNMENT.
9317 * varasm.c (align_variable, get_variable_align, build_constant_desc)
9318 (force_const_mem): Likewise.
9319 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
9320 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
9321 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
9322 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9323 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
9325 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
9326 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9327 constant_alignment_word_strings.
9328 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
9329 (CONSTANT_ALIGNMENT): Likewise.
9330 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9331 (arm_constant_alignment): New function.
9332 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
9333 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9334 constant_alignment_word_strings.
9335 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
9336 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9337 constant_alignment_word_strings.
9338 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
9339 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9340 (cris_constant_alignment): New function.
9341 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
9342 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9343 (epiphany_constant_alignment): New function.
9344 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
9345 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9346 constant_alignment_word_strings.
9347 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
9348 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9349 constant_alignment_word_strings.
9350 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
9351 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9352 constant_alignment_word_strings.
9353 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
9354 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
9355 * config/i386/i386.c (ix86_constant_alignment): Make static.
9356 Use the same interface as the target hook.
9357 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9358 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
9359 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9360 constant_alignment_word_strings.
9361 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
9362 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
9363 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9364 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
9365 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9366 constant_alignment_word_strings.
9367 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
9368 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9369 constant_alignment_word_strings.
9370 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
9371 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9372 constant_alignment_word_strings.
9373 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
9374 * config/microblaze/microblaze.c (microblaze_constant_alignment):
9376 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9377 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
9378 * config/mips/mips.c (mips_constant_alignment): New function.
9379 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9380 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
9381 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
9382 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9383 (mmix_constant_alignment): Make static. Use the same interface
9385 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
9386 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9387 constant_alignment_word_strings.
9388 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
9389 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9390 constant_alignment_word_strings.
9391 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
9392 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9393 constant_alignment_word_strings.
9394 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
9395 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9396 (rs6000_constant_alignment): New function.
9397 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
9398 * config/riscv/riscv.c (riscv_constant_alignment): New function.
9399 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9400 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
9401 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9402 (rs6000_constant_alignment): New function.
9403 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
9404 * config/s390/s390.c (s390_constant_alignment): New function.
9405 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9406 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
9407 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9408 constant_alignment_word_strings.
9409 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
9410 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9411 (sparc_constant_alignment): New function.
9412 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
9413 * config/spu/spu.c (spu_constant_alignment): New function.
9414 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9415 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
9416 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9417 constant_alignment_word_strings.
9418 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
9419 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9420 constant_alignment_word_strings.
9421 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
9422 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9423 constant_alignment_word_strings.
9424 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
9425 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9426 (visium_constant_alignment): New function.
9427 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
9428 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9429 (xtensa_constant_alignment): New function.
9430 * system.h (CONSTANT_ALIGNMENT): Poison.
9432 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
9434 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
9435 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
9436 (rs6000_builtin_valid_without_lhs): New helper function.
9437 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9438 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
9440 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
9442 * target.h (vec_perm_indices): Use unsigned short rather than
9444 (auto_vec_perm_indices): Likewise.
9445 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
9446 Use unsigned int rather than unsigned char.
9447 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
9449 2017-09-25 Richard Biener <rguenther@suse.de>
9451 * cfgloop.h (sort_sibling_loops): Declare.
9452 * cfgloop.c (sort_sibling_loops_cmp): New helper.
9453 (sort_sibling_loops): New function sorting the sibling loop list
9455 * graphite.c (graphite_transform_loops): Sort sibling loops.
9457 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
9459 * target.def (vec_perm_const_ok): Change sel parameter to
9461 * optabs-query.c (can_vec_perm_p): Update accordingly.
9462 * doc/tm.texi: Regenerate.
9463 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
9464 auto_vec_perm_indices and remove separate nelt field.
9465 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
9466 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
9467 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
9468 (aarch64_expand_vec_perm_const): Update accordingly.
9469 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
9470 to vec_perm_indices.
9471 * config/arm/arm.c (expand_vec_perm_d): Change perm to
9472 auto_vec_perm_indices and remove separate nelt field.
9473 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
9474 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
9475 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
9477 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
9478 to vec_perm_indices.
9479 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
9480 sel to vec_perm_indices.
9481 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
9482 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
9483 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
9485 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
9488 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
9491 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
9492 on the FUNCTION_DECL function context if it has a DIE that is a
9495 2017-09-25 Richard Biener <rguenther@suse.de>
9497 PR tree-optimization/82285
9498 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
9501 2017-09-25 Tom de Vries <tom@codesourcery.com>
9505 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
9508 2017-09-25 Richard Biener <rguenther@suse.de>
9510 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
9511 ISL errors other than isl_error_quota happen. Dump if the
9512 schedule is the same.
9513 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
9514 errors instead of aborting inside ISL.
9516 2017-09-25 Iain Sandoe <iain@codesourcery.com>
9519 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
9520 of libgcc_eh for m64.
9521 * config/i386/darwin64.h: Likewise.
9523 2017-09-25 Richard Biener <rguenther@suse.de>
9526 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
9527 attribute for incomplete types nor twice for complete ones.
9529 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
9532 * config/i386/i386.c (ix86_print_operand_address_as): Encode
9533 %esp as %rsp to avoid 0x67 prefix if there is no index or base
9536 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
9539 * config/i386/i386.opt (mprefer-avx256): Use
9540 ix86_target_flags variable.
9541 * config/i386/i386.c (ix86_target_string): Move
9542 -mprefer-avx256 to flag2_opts.
9544 2017-09-22 Jakub Jelinek <jakub@redhat.com>
9547 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
9548 and x != -1 | y != -1 into (x & y) != -1.
9550 2017-09-22 Steve Ellcey <sellcey@cavium.com>
9552 * config.gcc: Add new case statement to set
9553 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
9554 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
9555 s390x-*-linux* case statements. Added aarch64 to the list of
9556 supported architectures.
9558 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9560 PR tree-optimization/82289
9561 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
9562 STMT_VINFO_RELEVANT_P.
9564 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9565 Alan Hayward <alan.hayward@arm.com>
9566 David Sherwood <david.sherwood@arm.com>
9568 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
9569 for VR_RANGE only; don't allow VR_ANTI_RANGE.
9570 (extract_range_from_binary_expr_1): Don't call
9571 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
9573 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9574 Alan Hayward <alan.hayward@arm.com>
9575 David Sherwood <david.sherwood@arm.com>
9577 * target.def (preferred_vector_alignment): New hook.
9578 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
9580 * doc/tm.texi: Regenerate.
9581 * targhooks.h (default_preferred_vector_alignment): Declare.
9582 * targhooks.c (default_preferred_vector_alignment): New function.
9583 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
9585 (DR_TARGET_ALIGNMENT): New macro.
9586 (aligned_access_p): Update commentary.
9587 (vect_known_alignment_in_bytes): New function.
9588 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
9590 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
9591 Calculate the misalignment based on the target alignment rather than
9593 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
9594 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
9595 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
9596 the target alignment, rather than masking the element misalignment
9597 with the number of elements in a vector. Also use the target
9598 alignment when calculating the maximum number of peels.
9599 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
9600 instead of TYPE_ALIGN_UNIT.
9601 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
9602 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
9603 (vect_create_addr_base_for_vector_ref): Update call accordingly.
9604 (vect_create_data_ref_ptr): Likewise.
9605 (vect_setup_realignment): Realign by ANDing with
9606 -DR_TARGET_MISALIGNMENT.
9607 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
9608 the number of peels based on DR_TARGET_ALIGNMENT.
9609 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
9610 with the guaranteed alignment boundary when deciding whether
9612 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
9613 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
9614 (ensure_base_align): Remove stmt_info parameter. Get the
9615 target base alignment from DR_TARGET_ALIGNMENT.
9616 (vectorizable_store): Update call accordingly. Interpret
9617 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
9619 (vectorizable_load): Likewise.
9621 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
9622 Alan Hayward <alan.hayward@arm.com>
9623 David Sherwood <david.sherwood@arm.com>
9625 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
9626 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
9627 (vect_enhance_data_refs_alignment): Likewise.
9629 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
9631 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
9632 error. Only quit immediately if parsing is complete.
9633 (BEGIN): Initialize fatal_err and parse_done.
9634 (begin fpu, end fpu): Check number of arguments.
9635 (begin arch, end arch): Likewise.
9636 (begin cpu, end cpu): Likewise.
9637 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
9638 (optalias): Likewise.
9640 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
9642 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
9643 * config/arm/arm-isa.h: Delete. Move definitions to ...
9644 * arm-cpus.in: ... here. Use new feature and fgroup values.
9645 * config/arm/arm.c (arm_option_override): Use lower case for feature
9647 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
9648 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
9649 * config/arm/parsecpu.awk (END): Add new command 'isa'.
9651 (print_isa_bits_for): New function.
9652 (gen_isa): New function.
9653 (gen_comm_data): Use print_isa_bits_for.
9654 (define feature): New keyword.
9655 (define fgroup): New keyword.
9656 * config/arm/t-arm (TM_H): Remove.
9657 (GTM_H): Add arm-isa.h.
9658 (arm-isa.h): Add rule to generate file.
9659 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
9660 case for feature bit names.
9662 2017-09-22 Richard Biener <rguenther@suse.de>
9664 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
9666 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
9667 print of no dependency loops ...
9668 * graphite.c (graphite_transform_loops): ... here.
9669 (canonicalize_loop_closed_ssa_form): Work from inner to outer
9671 (same_close_phi_node, remove_duplicate_close_phi,
9672 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
9673 (canonicalize_loop_closed_ssa): ... here and simplify.
9674 * graphite-optimize-isl.c: Include tree-vectorizer.h.
9675 (optimize_isl): Use dump_printf_loc to tell when we stopped
9676 optimizing because of an ISL timeout.
9678 2017-09-22 Richard Biener <rguenther@suse.de>
9680 PR tree-optimization/82291
9681 * tree-if-conv.c (predicate_mem_writes): Make sure to
9682 remove writes in blocks predicated with false.
9684 2017-09-22 Richard Biener <rguenther@suse.de>
9686 * sese.c: Include cfganal.h.
9687 (if_region_set_false_region): Remove.
9688 (create_if_region_on_edge): Likewise.
9689 (move_sese_in_condition): Re-implement without destroying
9692 2017-09-22 Richard Biener <rguenther@suse.de>
9694 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
9695 Verify both BBs contain loop PHI nodes before dispatching to
9697 (graphite_regenerate_ast_isl): Do not recompute dominators,
9698 do not verify three times. Restructure for clarity.
9699 * graphite-scop-detection.c (same_close_phi_node,
9700 remove_duplicate_close_phi, make_close_phi_nodes_unique,
9701 defined_in_loop_p, canonicalize_loop_closed_ssa,
9702 canonicalize_loop_closed_ssa_form): Simplify, remove excess
9703 checking and SSA rewrite, move to ...
9704 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
9705 (graphite_initialize): Do not pass in ctx, do not reset the
9706 SCEV cache, compute only dominators.
9707 (graphite_transform_loops): Allocate ISL ctx after
9708 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
9709 Maintain post-dominators only around build_scops.
9710 * sese.c (if_region_set_false_region): Make static. Free
9711 and recompute dominators.
9712 (move_sese_in_condition): Assert we don't get called with
9713 post-dominators computed.
9714 * sese.h (if_region_set_false_region): Remove.
9716 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
9718 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
9719 mode attribute for TARGET_AVX512VL.
9721 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
9723 * config/i386/i386.opt (mprefer-avx256): New option.
9724 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
9726 (ix86_preferred_simd_mode): Return 256-bit AVX modes
9727 for TARGET_PREFER_AVX256.
9728 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
9730 2017-09-21 Jeff Law <law@redhat.com>
9732 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
9733 Fix dump output if the only stack space is for pushed registers.
9735 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9737 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
9740 2017-09-21 Martin Sebor <msebor@redhat.com>
9743 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
9744 code (in C++) or code that triggers warnings.
9746 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
9748 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
9750 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
9752 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
9753 * sched-rgn.c: Ditto.
9754 * sel-sched-ir.c: Ditto.
9756 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
9758 * toplev.h (set_random_seed): Adjust return type.
9759 * toplev.c (init_local_tick): Move eager initialization of random_seed
9760 to get_random_seed. Adjust comment.
9761 (init_random_seed): Inline to get_random_seed, delete.
9762 (get_random_seed): Initialize random_seed lazily.
9763 (set_random_seed): Do not return previous value.
9764 (print_switch_value): Do not call get_random_seed.
9766 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
9768 * cgraph.c (delete_function_version): New, broken out from...
9769 (cgraph_node::delete_function_version): ...here. Rename to
9770 cgraph_node::delete_function_version_by_decl. Update all uses.
9771 (cgraph_node::remove): Call delete_function_version.
9773 2017-09-21 Jakub Jelinek <jakub@redhat.com>
9776 * tree-inline.c (expand_call_inline): Emit clobber stmts for
9777 VAR_DECLs to which addressable non-volatile parameters are mapped
9778 and for id->retvar after the return value assignment. Clear
9779 id->retval and id->retbnd after inlining.
9781 2017-09-21 Richard Biener <rguenther@suse.de>
9783 PR tree-optimization/82276
9784 PR tree-optimization/82244
9785 * tree-vrp.c (build_assert_expr_for): Set
9786 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
9788 (remove_range_assertions): Revert earlier change.
9790 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
9793 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
9795 2017-09-21 Richard Biener <rguenther@suse.de>
9797 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
9798 Restore valid IL after code generation errors.
9799 * graphite.c (graphite_transform_loops): Diagnose code
9800 generation issues as MSG_MISSED_OPTIMIZATION and continue
9801 with processing SCOPs.
9803 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9804 Alan Hayward <alan.hayward@arm.com>
9805 David Sherwood <david.sherwood@arm.com>
9807 * calls.c (compute_argument_addresses): Use simplify_gen_binary
9808 rather than choosing between plus_constant and gen_rtx_<CODE>.
9809 * expr.c (emit_push_insn): Likewise.
9810 (expand_expr_real_2): Likewise.
9812 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9813 Alan Hayward <alan.hayward@arm.com>
9814 David Sherwood <david.sherwood@arm.com>
9816 * loop-unroll.c (split_iv): Call copy_rtx on the step.
9818 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9819 Alan Hayward <alan.hayward@arm.com>
9820 David Sherwood <david.sherwood@arm.com>
9822 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
9823 calling tree_to_uhwi.
9825 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9826 Alan Hayward <alan.hayward@arm.com>
9827 David Sherwood <david.sherwood@arm.com>
9829 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
9830 INTEGER_CST rather than a negative test for ADDR_EXPR.
9832 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
9833 Alan Hayward <alan.hayward@arm.com>
9834 David Sherwood <david.sherwood@arm.com>
9836 * tree-vrp.c (extract_range_from_binary_expr_1): Check
9837 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
9839 2017-09-21 Richard Biener <rguenther@suse.de>
9841 PR tree-optimization/71351
9842 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
9843 graphite_create_new_loop_guard): Remove, fold remaining parts
9845 (translate_isl_ast_node_for): ... here and simplify.
9847 2017-09-21 Jakub Jelinek <jakub@redhat.com>
9850 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
9851 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
9852 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
9853 alternative always use QI mode, for -Os imov (=R,R) alternative
9854 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
9857 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9858 Jeff Law <law@redhat.com>
9860 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
9861 (allocate_stack_space): New function, partially extracted from
9863 (s390_emit_prologue): Track offset to most recent stack probe.
9864 Code to allocate space moved into allocate_stack_space.
9865 Dump actions when no stack is allocated.
9866 (s390_prologue_plus_offset): New function.
9867 (s390_emit_stack_probe): Likewise.
9869 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
9871 * common.opt (Wa, Wl, Wp, g, gz=): Add
9873 (gno-column-info): Remove.
9874 (gcolumn-info): Drop RejectNegative.
9876 (gno-record-gcc-switches): Remove.
9877 (grecord-gcc-switches): Drop RejectNegative.
9878 (gno-split-dwarf): Remove.
9879 (gsplit-dwarf): Drop RejectNegative.
9880 (gno-strict-dwarf): Remove.
9881 (gstrict-dwarf): Drop RejectNegative.
9882 * config/darwin.opt (gfull, gused): Add RejectNegative.
9883 * dwarf2out.c (gen_producer_string): Drop
9884 gno-record-gcc-switches handler.
9885 * optc-gen.awk: Add g to prefixes with negative forms.
9886 * opts-common.c (remapping_prefix_p): New.
9887 (find_opt): Check it.
9888 (generate_canonical_option): Test g prefix.
9889 (option_map): Add -gno- mapping.
9890 (add_misspelling_candidates): Check remapping_prefix_p.
9892 2017-09-20 Jeff Law <law@redhat.com>
9894 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
9895 thinko in stack clash protection support.
9897 * explow.c (compute_stack_clash_protection_loop_data): Use
9898 CONST_INT_P instead of explicit test. Verify object is a
9899 CONST_INT_P before looking at INTVAL.
9900 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
9901 instead of explicit test.
9903 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
9906 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
9907 address instead of to r1 and r11.
9909 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
9911 * config.gcc: Support "knm".
9912 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
9913 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
9915 * config/i386/i386.c (m_KNM): Define.
9916 (processor_target_table): Add "knm".
9918 (ix86_option_override_internal): Add "knm".
9919 (ix86_issue_rate): Add PROCESSOR_KNM.
9920 (ix86_adjust_cost): Ditto.
9921 (ia32_multipass_dfa_lookahead): Ditto.
9922 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
9923 (fold_builtin_cpu): Add M_INTEL_KNM.
9924 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
9925 (processor_type): Add PROCESSOR_KNM.
9926 * config/i386/x86-tune.def: Add m_KNM.
9927 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
9929 2017-09-20 Richard Biener <rguenther@suse.de>
9931 PR tree-optimization/80213
9932 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
9933 are allowed in empty BBs as well.
9934 (canonicalize_loop_closed_ssa): Also look for other complex
9936 (scop_detection::get_sese): Include the loop-closed PHI block
9938 (scop_detection::merge_sese): Remove code adding extra blocks.
9939 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
9940 (build_scops): Assert the final returned scop is invalid.
9942 2017-09-20 Richard Biener <rguenther@suse.de>
9944 PR tree-optimization/82264
9945 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
9947 (vn_phi_lookup): Likewise.
9948 (vn_phi_insert): Likewise.
9950 2017-09-20 Jakub Jelinek <jakub@redhat.com>
9952 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
9953 that fits into uhwi or shwi, add DW_AT_const_value regardless
9954 of early_dwarf without going through RTL, using add_AT_unsigned
9957 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
9958 (DEBUG_LTO_ABBREV_SECTION): Likewise.
9959 (DEBUG_LTO_MACINFO_SECTION): Likewise.
9960 (DEBUG_MACRO_SECTION): Likewise.
9961 (DEBUG_LTO_MACRO_SECTION): Likewise.
9962 (DEBUG_STR_DWO_SECTION): Likewise.
9963 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
9964 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
9965 (DEBUG_LTO_DWO_LINE_SECTION): Define.
9966 (DEBUG_LTO_LINE_STR_SECTION): Define.
9967 (init_sections_and_labels): Initialize debug_line_str_section
9968 variable. Initialize debug_loc_section for -gdwarf-5 to
9969 DEBUG_LOCLISTS_SECTION. Formatting fixes.
9971 2017-09-20 Richard Biener <rguenther@suse.de>
9973 * graphite-sese-to-poly.c (extract_affine): Properly handle
9974 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
9976 2017-09-20 Richard Biener <rguenther@suse.de>
9978 PR tree-optimization/81373
9979 * graphite-scop-detection.c (build_cross_bb_scalars_def):
9980 Force SESE live-out defs to be handled even if they are
9983 2017-09-19 Jeff Law <law@redhat.com>
9985 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
9986 nothing for stack adjustments with REG_STACK_CHECK.
9987 * sched-deps.c (parse_add_or_inc): Reject insns with
9988 REG_STACK_CHECK from dependency breaking.
9989 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
9990 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
9991 * reg-notes.def (STACK_CHECK): New note.
9993 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
9994 (ix86_expand_prologue): Dump stack clash info as needed.
9995 Call ix86_adjust_stack_and_probe_stack_clash as needed.
9997 * function.c (dump_stack_clash_frame_info): New function.
9998 * function.h (dump_stack_clash_frame_info): Prototype.
9999 (enum stack_clash_probes): New enum.
10001 * config/alpha/alpha.c (alpha_expand_prologue): Also check
10002 flag_stack_clash_protection.
10003 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
10004 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
10005 (arm_frame_pointer_required): Likewise.
10006 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
10007 (ia64_expand_prologue): Likewise.
10008 * config/mips/mips.c (mips_expand_prologue): Likewise.
10009 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
10010 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
10011 (sparc_flat_expand_prologue): Likewise.
10012 * config/spu/spu.c (spu_expand_prologue): Likewise.
10014 * explow.c: Include "params.h".
10015 (anti_adjust_stack_and_probe_stack_clash): New function.
10016 (get_stack_check_protect): Likewise.
10017 (compute_stack_clash_protection_loop_data): Likewise.
10018 (emit_stack_clash_protection_loop_start): Likewise.
10019 (emit_stack_clash_protection_loop_end): Likewise.
10020 (allocate_dynamic_stack_space): Use get_stack_check_protect.
10021 Use anti_adjust_stack_and_probe_stack_clash.
10022 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
10023 (emit_stack_clash_protection_loop_start): Likewise.
10024 (emit_stack_clash_protection_loop_end): Likewise.
10025 * rtl.h (get_stack_check_protect): Prototype.
10026 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
10027 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
10028 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
10030 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
10032 * doc/tm.texi: Rebuilt.
10033 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
10034 get_stack_check_protect.
10035 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
10036 * config/arm/arm.c (arm_expand_prologue): Likewise.
10037 (arm_frame_pointer_required): Likewise.
10038 * config/i386/i386.c (ix86_expand_prologue): Likewise.
10039 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
10040 * config/mips/mips.c (mips_expand_prologue): Likewise.
10041 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
10042 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
10043 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
10044 (sparc_flat_expand_prologue): Likewise.
10046 * common.opt (-fstack-clash-protection): New option.
10047 * flag-types.h (enum stack_check_type): Note difference between
10048 -fstack-check= and -fstack-clash-protection.
10049 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
10050 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
10051 * toplev.c (process_options): Issue warnings/errors for cases
10052 not handled with -fstack-clash-protection.
10053 * doc/invoke.texi (-fstack-clash-protection): Document new option.
10054 (-fstack-check): Note additional problem with -fstack-check=generic.
10055 Note that -fstack-check is primarily for Ada and refer users
10056 to -fstack-clash-protection for stack-clash-protection.
10057 Document new params for stack clash protection.
10059 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
10061 * config/i386/i386.c (ix86_split_long_move): Do not handle
10062 address used for LEA in a special way.
10064 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
10066 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
10068 2017-09-19 Martin Sebor <msebor@redhat.com>
10071 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
10072 of incompatible types.
10074 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
10076 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10077 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
10078 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10079 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
10081 2017-09-19 Richard Biener <rguenther@suse.de>
10083 PR tree-optimization/82244
10084 * tree-vrp.c (remove_range_assertions): Do not propagate
10085 a constant to abnormals but replace the assert with a copy.
10087 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
10089 PR rtl-optimization/57878
10090 PR rtl-optimization/68988
10091 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
10092 avoidance test involving non_reload_pseudos. Move frequency test
10093 below the general fragmentation avoidance test.
10095 2017-09-19 Richard Biener <rguenther@suse.de>
10097 PR tree-optimization/69728
10098 * graphite-sese-to-poly.c (schedule_error): New global.
10099 (add_loop_schedule): Handle empty domain by failing the
10101 (build_original_schedule): Handle schedule_error.
10103 2017-09-19 Richard Biener <rguenther@suse.de>
10105 * graphite-scop-detection.c (scop_detection::can_represent_loop):
10106 Do not iterate to sibling loops but only to siblings of inner
10109 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
10112 * config/m68k/m68k.md (moveq feeding equality comparison): Check
10113 that the registers are different.
10115 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
10117 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
10118 to processor_model and "amdfam17h" to arch_names_table.
10119 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
10121 2017-09-18 Jakub Jelinek <jakub@redhat.com>
10124 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
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-vectorizer.h (vect_slp_analyze_operations): Replace parameters
10132 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
10134 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
10135 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
10136 vect_schedule_slp_instance.
10137 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
10138 Update call to vect_slp_analyze_node_operations. Simplify return
10140 (vect_slp_analyze_bb_1): Update call accordingly.
10141 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
10142 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
10143 (vect_schedule_slp): Update call accordingly.
10145 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10146 Alan Hayward <alan.hayward@arm.com>
10147 David Sherwood <david.sherwood@arm.com>
10149 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
10150 with types that aren't in fact scalar.
10152 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10154 * tree-vect-slp.c (vect_record_max_nunits): New function,
10156 (vect_build_slp_tree_1): ...here.
10157 (vect_build_slp_tree_2): Call it for phis too.
10159 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10161 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
10162 to vect_get_vec_def_for_operand when getting the mask operand.
10164 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10165 Alan Hayward <alan.hayward@arm.com>
10166 David Sherwood <david.sherwood@arm.com>
10168 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
10171 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10172 Alan Hayward <alan.hayward@arm.com>
10173 David Sherwood <david.sherwood@arm.com>
10175 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
10176 calculation for vector booleans.
10178 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10179 Alan Hayward <alan.hayward@arm.com>
10180 David Sherwood <david.sherwood@arm.com>
10182 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
10184 (vect_transform_stmt): ...here.
10185 (vect_analyze_stmt): Use it instead of calling
10186 vectorizable_live_operation directly.
10188 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
10190 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
10191 non-SIMT targets in acc vector loops.
10193 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
10195 * configure.ac: Add arc and check if assembler supports gdwarf2.
10196 * configure: Regenerate.
10198 2017-09-18 Richard Biener <rguenther@suse.de>
10200 PR tree-optimization/82220
10201 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
10202 epilogue niters from the min_profitable_iters compute.
10204 2017-09-18 Jakub Jelinek <jakub@redhat.com>
10207 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
10208 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
10209 (ix86_init_pic_reg): Revert 2017-09-01 changes.
10211 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
10214 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
10215 switching to a new text section.
10217 2017-09-18 Richard Biener <rguenther@suse.de>
10219 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
10221 (build_alias_set): Reject aliases with no access function.
10223 2017-09-18 Richard Biener <rguenther@suse.de>
10225 PR tree-optimization/79622
10226 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
10228 (build_cross_bb_scalars_use): Likewise.
10230 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
10232 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
10234 2017-09-18 Alan Modra <amodra@gmail.com>
10237 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
10238 stack_pointer_rtx for count 0. Update comments. Break up
10239 large rtl expression.
10241 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
10243 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
10244 Increase to 20 bytes.
10245 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
10246 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
10247 or avx version of the stub.
10249 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
10252 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
10253 compute the minimum stack alignment. Also update preferred stack
10254 boundary for leaf functions.
10256 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
10258 PR tree-optimization/82228
10259 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
10262 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
10264 * common/config/nds32/nds32-common.c
10265 (nds32_option_optimization_table): Refine formatting.
10266 (nds32_option_optimization_table): Use -fsched-pressure and
10267 -fomit-frame-pointer for specific optimization level.
10269 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
10271 * config/nds32/nds32.c: Refine formatting and comments.
10272 * config/nds32/nds32.h: Likewise.
10273 * config/nds32/nds32.md: Likewise.
10274 * config/nds32/nds32-cost.c: Likewise.
10275 * config/nds32/nds32-isr.c: Likewise.
10276 * config/nds32/nds32-md-auxiliary.c: Likewise.
10277 * config/nds32/nds32-multiple.md: Likewise.
10278 * config/nds32/nds32-predicates.c: Likewise.
10280 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
10281 Jakub Jelinek <jakub@redhat.com>
10283 Add support for -std=c++2a.
10284 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
10286 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
10288 2017-09-15 Steve Ellcey <sellcey@cavium.com>
10291 * doc/extend.texi (Common Function Attributes): Add
10292 references to ARM, AArch64, and S/390 specific attributes.
10293 (Function Specific Option Pragmas): Add AArch64 and S/390
10294 to list of back ends that support the target pragma.
10296 2017-09-15 Nathan Sidwell <nathan@acm.org>
10298 * doc/standards.texi: Fix C++17 description. Update URLs for
10301 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
10303 * common.opt (Wcast-align=strict): New warning option.
10304 * doc/invoke.texi: Document -Wcast-align=strict.
10306 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
10308 * cgraph.h (cgraph_thunk_info): Add comments.
10309 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
10310 assert for VIRTUAL_* arguments stricter.
10312 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
10314 PR tree-optimization/71026
10315 * match.pd: Move RDIV patterns from fold-const.c
10316 * fold-const.c (distribute_real_division): Removed.
10317 (fold_binary_loc): Remove calls to distribute_real_divison.
10319 2017-09-15 Jakub Jelinek <jakub@redhat.com>
10321 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
10322 c++1z and gnu++1z as deprecated. Change other references to
10323 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
10324 Change -Wc++1z-compat to -Wc++17-compat.
10325 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
10326 * dwarf2out.c (highest_c_language): Handle C++17.
10327 (gen_compile_unit_die): Likewise.
10329 2017-09-15 Jakub Jelinek <jakub@redhat.com>
10331 PR rtl-optimization/82192
10332 * combine.c (make_extraction): Don't look through non-paradoxical
10333 SUBREGs or TRUNCATE if pos + len is or might be bigger than
10336 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
10337 Alan Hayward <alan.hayward@arm.com>
10338 David Sherwood <david.sherwood@arm.com>
10340 * target.def (function_arg_offset): New hook.
10341 * targhooks.h (default_function_arg_offset): Declare.
10342 * targhooks.c (default_function_arg_offset): New function.
10343 * function.c (locate_and_pad_parm): Use
10344 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
10345 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
10346 (TARGET_FUNCTION_ARG_OFFSET): ...this.
10347 * doc/tm.texi: Regenerate.
10348 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
10349 * config/spu/spu.c (spu_function_arg_offset): New function.
10350 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
10351 * system.h (FUNCTION_ARG_OFFSET): Poison.
10353 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
10354 Alan Hayard <alan.hayward@arm.com>
10355 David Sherwood <david.sherwood@arm.com>
10357 * target.def (truly_noop_truncation): New hook.
10358 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
10359 than TRULY_NOOP_TRUNCATION.
10360 * hooks.h (hook_bool_uint_uint_true): Declare.
10361 * hooks.c (hook_bool_uint_uint_true): New function.
10362 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
10363 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
10364 * doc/tm.texi: Regenerate.
10365 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
10366 rather than TRULY_NOOP_TRUNCATION in comments.
10367 (simplify_comparison): Likewise.
10368 (record_truncated_value): Likewise.
10369 * expmed.c (extract_bit_field_1): Likewise.
10370 (extract_split_bit_field): Likewise.
10371 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
10372 instead of TRULY_NOOP_TRUNCATION.
10373 * function.c (assign_parm_setup_block): Likewise.
10374 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
10375 * rtlhooks.c: Include target.h.
10376 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
10377 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
10378 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
10379 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
10380 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
10381 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
10382 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
10383 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
10384 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
10385 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
10386 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
10387 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
10388 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
10389 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
10390 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
10391 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
10392 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
10393 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
10394 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
10395 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
10396 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
10397 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
10398 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
10399 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
10400 * config/mips/mips.c (mips_truly_noop_truncation): New function.
10401 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10402 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
10403 rather than TRULY_NOOP_TRUNCATION in comments.
10404 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
10405 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
10406 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
10407 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
10408 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
10409 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
10410 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
10411 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
10412 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
10413 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
10414 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
10415 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
10416 rather than TRULY_NOOP_TRUNCATION in comments.
10417 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
10418 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
10419 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
10420 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
10421 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
10422 TRULY_NOOP_TRUNCATION condition.
10423 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
10424 (TRULY_NOOP_TRUNCATION): Delete.
10425 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
10426 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
10427 * config/spu/spu.c (spu_truly_noop_truncation): New function.
10428 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10429 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
10430 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
10431 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
10432 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
10433 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
10434 rather than TRULY_NOOP_TRUNCATION in comments.
10435 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
10436 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
10437 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
10438 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
10439 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
10440 * system.h (TRULY_NOOP_TRUNCATION): Poison.
10442 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
10445 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
10446 (*cmp_ior): Likewise.
10447 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
10448 (*ior_scc_scc_cmp): Likewise.
10449 (*and_scc_scc): Likewise.
10450 (*and_scc_scc_cmp): Likewise.
10452 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
10453 Alan Hayard <alan.hayward@arm.com>
10454 David Sherwood <david.sherwood@arm.com>
10456 * target.def (can_change_mode_class): New hook.
10457 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
10458 (hard_regno_nregs): Likewise.
10459 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
10460 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
10461 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
10462 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
10463 (LOAD_EXTEND_OP): Update accordingly.
10464 * doc/tm.texi: Regenerate.
10465 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
10466 CANNOT_CHANGE_MODE_CLASS.
10467 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
10468 (REG_CAN_CHANGE_MODE_P): ...this new macro.
10469 * combine.c (simplify_set): Update accordingly.
10470 * emit-rtl.c (validate_subreg): Likewise.
10471 * recog.c (general_operand): Likewise.
10472 * regcprop.c (mode_change_ok): Likewise.
10473 * reload1.c (choose_reload_regs): Likewise.
10474 (inherit_piecemeal_p): Likewise.
10475 * rtlanal.c (simplify_subreg_regno): Likewise.
10476 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
10477 instead of CANNOT_CHANGE_MODE_CLASS.
10478 (reload_cse_simplify_operands): Likewise.
10479 * reload.c (push_reload): Use targetm.can_change_mode_class
10480 instead of CANNOT_CHANGE_MODE_CLASS.
10481 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
10482 REG_CANNOT_CHANGE_MODE_P.
10483 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10484 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
10485 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10486 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10487 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10488 (arm_can_change_mode_class): New function.
10489 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
10490 than CANNOT_CHANGE_MODE_CLASS in comments.
10491 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10492 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
10493 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
10494 (ix86_can_change_mode_class): ...this new function, inverting the
10495 sense of the return value.
10496 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10497 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10498 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10499 (ia64_can_change_mode_class): New function.
10500 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10501 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
10502 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
10503 (m32c_can_change_mode_class): ...this new function, inverting the
10504 sense of the return value.
10505 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10506 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10507 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
10508 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
10509 (mips_can_change_mode_class): ...this new function, inverting the
10510 sense of the return value.
10511 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10512 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10513 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10514 (msp430_can_change_mode_class): New function.
10515 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10516 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
10517 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10518 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10519 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10520 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
10521 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10522 (pa_cannot_change_mode_class): Replace with...
10523 (pa_can_change_mode_class): ...this new function, inverting the
10524 sense of the return value.
10525 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
10526 than CANNOT_CHANGE_MODE_CLASS in comments.
10527 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10528 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
10529 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10530 (pdp11_cannot_change_mode_class): Replace with...
10531 (pdp11_can_change_mode_class): ...this new function, inverting the
10532 sense of the return value.
10533 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10534 * config/powerpcspe/powerpcspe-protos.h
10535 (rs6000_cannot_change_mode_class_ptr): Delete.
10536 * config/powerpcspe/powerpcspe.c
10537 (rs6000_cannot_change_mode_class_ptr): Delete.
10538 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10539 (rs6000_option_override_internal): Assign to
10540 targetm.can_change_mode_class instead of
10541 rs6000_cannot_change_mode_class_ptr.
10542 (rs6000_cannot_change_mode_class): Replace with...
10543 (rs6000_can_change_mode_class): ...this new function, inverting the
10544 sense of the return value.
10545 (rs6000_debug_cannot_change_mode_class): Replace with...
10546 (rs6000_debug_can_change_mode_class): ...this new function.
10547 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10548 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
10549 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10550 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10551 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
10553 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
10554 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10555 (rs6000_option_override_internal): Assign to
10556 targetm.can_change_mode_class instead of
10557 rs6000_cannot_change_mode_class_ptr.
10558 (rs6000_cannot_change_mode_class): Replace with...
10559 (rs6000_can_change_mode_class): ...this new function, inverting the
10560 sense of the return value.
10561 (rs6000_debug_cannot_change_mode_class): Replace with...
10562 (rs6000_debug_can_change_mode_class): ...this new function.
10563 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10564 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
10565 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
10566 (s390_can_change_mode_class): ...this new function, inverting the
10567 sense of the return value.
10568 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10569 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10570 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
10571 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10572 (sh_cannot_change_mode_class): Replace with...
10573 (sh_can_change_mode_class): ...this new function, inverting the
10574 sense of the return value.
10575 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10576 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10577 (sparc_can_change_mode_class): New function.
10578 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10579 * config/spu/spu.c (spu_can_change_mode_class): New function.
10580 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10581 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
10582 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
10583 (visium_can_change_mode_class): New function.
10584 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
10586 2017-09-15 Richard Biener <rguenther@suse.de>
10588 PR tree-optimization/82217
10589 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
10590 but not undefined case.
10592 2017-09-15 Jakub Jelinek <jakub@redhat.com>
10595 * postreload.c (reload_cse_simplify_operands): Skip
10596 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
10598 2017-09-15 Richard Biener <rguenther@suse.de>
10600 PR tree-optimization/68823
10601 * graphite-scop-detection.c (build_alias_set): If we have a
10602 possible dependence check whether we can handle them by just
10603 looking at the DRs DR_ACCESS_FNs.
10604 (build_scops): If build_alias_set fails, fail the SCOP.
10606 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
10608 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
10609 to support float128 built-in functions that require the ISA 3.0
10611 (BU_FLOAT128_3_HW): Likewise.
10612 (SQRTF128): Add support for the IEEE 128-bit square root and fma
10613 built-in functions.
10614 (FMAF128): Likewise.
10616 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
10617 support for built-in functions that need the ISA 3.0 IEEE 128-bit
10618 floating point instructions.
10619 (rs6000_invalid_builtin): Likewise.
10620 (rs6000_builtin_mask_names): Likewise.
10621 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
10622 (RS6000_BTM_FLOAT128_HW): Likewise.
10623 (RS6000_BTM_COMMON): Likewise.
10624 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
10626 * doc/extend.texi (RS/6000 built-in functions): Document the
10627 IEEE 128-bit floating point square root and fused multiply-add
10628 built-in functions.
10630 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
10632 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
10633 reg (r2) isn't in the set of registers defined in the prologue.
10635 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10636 Alan Hayward <alan.hayward@arm.com>
10637 David Sherwood <david.sherwood@arm.com>
10639 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
10640 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
10641 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
10642 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
10643 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
10645 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
10646 max_vectorization_factor.
10647 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
10649 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10650 Alan Hayward <alan.hayward@arm.com>
10651 David Sherwood <david.sherwood@arm.com>
10653 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
10654 (vect_worthwhile_without_simd_p): Declare.
10655 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
10656 (vectorizable_reduction): Use it.
10657 * tree-vect-stmts.c (vectorizable_shift): Likewise.
10658 (vectorizable_operation): Likewise.
10660 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10661 Alan Hayward <alan.hayward@arm.com>
10662 David Sherwood <david.sherwood@arm.com>
10664 * tree-vectorizer.h (vect_get_num_copies): New function.
10665 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
10666 * tree-vect-loop.c (vectorizable_reduction): Likewise.
10667 (vectorizable_induction): Likewise.
10668 (vectorizable_live_operation): Likewise.
10669 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
10670 (vectorizable_bswap): Likewise.
10671 (vectorizable_call): Likewise.
10672 (vectorizable_conversion): Likewise.
10673 (vectorizable_assignment): Likewise.
10674 (vectorizable_shift): Likewise.
10675 (vectorizable_operation): Likewise.
10676 (vectorizable_store): Likewise.
10677 (vectorizable_load): Likewise.
10678 (vectorizable_condition): Likewise.
10679 (vectorizable_comparison): Likewise.
10680 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
10682 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10683 Alan Hayward <alan.hayward@arm.com>
10684 David Sherwood <david.sherwood@arm.com>
10686 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
10687 of vect_init_vector.
10689 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10690 Alan Hayward <alan.hayward@arm.com>
10691 David Sherwood <david.sherwood@arm.com>
10693 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
10694 an inline wrapper that provides a location.
10695 (gimple_build_vector): Likewise.
10696 * gimple-fold.c (gimple_build_vector_from_val): New function.
10697 (gimple_build_vector): Likewise.
10698 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
10699 functions to build the initial value. Always return a gimple value.
10700 (get_initial_defs_for_reduction): Likewise. Only compute
10702 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
10703 vect_init_vector on the results from get_initial_def(s)_for_reduction.
10704 (vectorizable_induction): Use gimple_build_vector rather than
10707 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10708 Alan Hayward <alan.hayward@arm.com>
10709 David Sherwood <david.sherwood@arm.com>
10711 * target.h (vec_perm_indices): New typedef.
10712 (auto_vec_perm_indices): Likewise.
10713 * optabs-query.h: Include target.h
10714 (can_vec_perm_p): Take a vec_perm_indices *.
10715 * optabs-query.c (can_vec_perm_p): Likewise.
10716 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
10717 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
10718 * tree-vect-generic.c (lower_vec_perm): Likewise.
10719 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
10720 (vect_grouped_load_supported): Likewise.
10721 (vect_shift_permute_load_chain): Likewise.
10722 (vect_permute_store_chain): Use auto_vec_perm_indices.
10723 (vect_permute_load_chain): Likewise.
10724 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
10725 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
10726 Update uses of can_vec_perm_p.
10727 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
10728 mode with a number of elements. Take a vec_perm_indices *.
10729 (vect_create_epilog_for_reduction): Update accordingly.
10730 Use auto_vec_perm_indices.
10731 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
10732 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
10733 (vect_transform_slp_perm_load): Likewise.
10734 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
10735 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
10736 (vect_gen_perm_mask_checked): Likewise.
10737 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
10738 (vect_gen_perm_mask_checked): Likewise.
10739 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
10740 (vectorizable_store): Likewise.
10741 (vectorizable_load): Likewise.
10742 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
10743 (vectorizable_bswap): Likewise.
10745 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10746 Alan Hayward <alan.hayward@arm.com>
10747 David Sherwood <david.sherwood@arm.com>
10749 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
10750 * tree.c (build_vector): Likewise.
10751 (build_vector_from_ctor): Update accordingly.
10752 (build_vector_from_val): Likewise.
10753 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
10754 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
10755 * tree-vect-generic.c (add_rshift): Likewise.
10756 (expand_vector_divmod): Likewise.
10757 (optimize_vector_constructor): Likewise.
10758 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
10759 (vect_transform_slp_perm_load): Likewise.
10760 (vect_schedule_slp_instance): Likewise.
10761 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
10762 (vectorizable_call): Likewise.
10763 (vect_gen_perm_mask_any): Likewise. Add elements in order.
10764 * expmed.c (make_tree): Likewise.
10765 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
10766 a vector passed to build_vector.
10767 (fold_convert_const): Likewise.
10768 (exact_inverse): Likewise.
10769 (fold_ternary_loc): Likewise.
10770 (fold_relational_const): Likewise.
10771 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
10772 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
10773 (const_unop): Likewise. Store the reduction accumulator in a
10774 variable rather than an array.
10775 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
10776 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
10777 the new vector, rather than constructing it after the input arrays.
10778 (native_interpret_vector): Use auto_vec<tree> when building
10779 a vector passed to build_vector. Add elements in order.
10780 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
10781 auto_vec<tree> when building a vector passed to build_vector.
10782 (vect_create_epilog_for_reduction): Likewise.
10783 (vectorizable_induction): Likewise.
10784 (get_initial_def_for_reduction): Likewise. Fix indentation of
10786 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
10788 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
10789 passed to build_vector.
10791 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
10792 Alan Hayward <alan.hayward@arm.com>
10793 David Sherwood <david.sherwood@arm.com>
10795 * tree-core.h (tree_base::u): Add an "nelts" field.
10796 (tree_vector): Use VECTOR_CST_NELTS as the length.
10797 * tree.c (tree_size): Likewise.
10798 (make_vector): Initialize VECTOR_CST_NELTS.
10799 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
10800 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
10801 TYPE_VECTOR_SUBPARTS.
10802 * expr.c (const_vector_mask_from_tree): Consistently use "units"
10803 as the number of units, setting it from VECTOR_CST_NELTS.
10804 (const_vector_from_tree): Likewise.
10805 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
10806 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
10807 (fold_negate_expr_1): Likewise.
10808 (fold_convert_const): Likewise.
10809 (const_binop): Likewise. Differentiate the number of output and
10811 (const_unop): Likewise.
10812 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
10813 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
10814 in cases that did the opposite.
10816 2017-09-14 Richard Biener <rguenther@suse.de>
10818 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
10821 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
10823 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
10825 2017-09-14 Jakub Jelinek <jakub@redhat.com>
10828 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
10829 if and where to split a bb, except for splitting before debug insn
10830 sequences followed by non-label real insn. Delete debug insns
10831 in between basic blocks.
10833 * combine.c (make_compound_operation_int): Formatting fixes.
10835 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
10836 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10837 * config/netbsd.h (LINK_EH_SPEC): Likewise.
10838 * config/sol2.h (LINK_EH_SPEC): Likewise.
10839 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10840 * config/s390/linux.h (LINK_SPEC): Likewise.
10841 * config/freebsd.h (LINK_EH_SPEC): Likewise.
10842 * config/openbsd.h (LINK_EH_SPEC): Likewise.
10843 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10844 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
10845 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
10846 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
10847 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
10848 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
10850 2017-09-13 Jakub Jelinek <jakub@redhat.com>
10852 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
10854 (ENDFILE_LINUX_SPEC): Likewise.
10855 (LINK_EH_SPEC): Likewise.
10856 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
10857 (LINK_OS_LINUX_SPEC32): Likewise.
10858 (LINK_OS_LINUX_SPEC64): Likewise.
10859 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
10860 (LINK_OS_LINUX_SPEC): Likewise.
10862 2017-09-13 Martin Liska <mliska@suse.cz>
10864 PR middle-end/82154
10865 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
10866 CASE_HIGH is NULL_TREE.
10868 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
10869 Alan Hayward <alan.hayward@arm.com>
10870 David Sherwood <david.sherwood@arm.com>
10872 * target.def (secondary_memory_needed): New hook.
10873 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
10874 instead of SECONDARY_MEMORY_NEEDED.
10875 (secondary_memory_needed_mode): Likewise.
10876 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
10877 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
10878 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
10879 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
10880 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
10881 * doc/tm.texi: Regenerate.
10882 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
10883 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
10884 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10885 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
10886 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
10887 * config/i386/i386.c (inline_secondary_memory_needed): Put the
10888 mode argument first and change the reg_class arguments to reg_class_t.
10889 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
10890 Make static. Update the call to inline_secondary_memory_needed.
10891 (ix86_register_move_cost): Update the call to
10892 inline_secondary_memory_needed.
10893 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10894 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
10896 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
10897 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
10899 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
10900 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
10901 * config/mips/mips.c (mips_secondary_memory_needed): Make static
10902 and match hook interface. Add comment from mips.h.
10903 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10904 * config/mmix/mmix.md (truncdfsf2): Refer to
10905 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
10907 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
10908 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
10909 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10910 (pa_secondary_memory_needed): New function.
10911 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
10912 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
10913 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10914 (pdp11_secondary_memory_needed): Make static and match hook interface.
10915 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
10916 * config/powerpcspe/powerpcspe-protos.h
10917 (rs6000_secondary_memory_needed_ptr): Delete.
10918 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
10920 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10921 (rs6000_option_override_internal): Assign to
10922 targetm.secondary_memory_needed rather than
10923 rs6000_secondary_memory_needed_ptr.
10924 (rs6000_secondary_memory_needed): Match hook interface.
10925 (rs6000_debug_secondary_memory_needed): Likewise.
10926 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
10927 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
10928 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
10929 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10930 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
10931 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
10933 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
10934 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10935 (rs6000_option_override_internal): Assign to
10936 targetm.secondary_memory_needed rather than
10937 rs6000_secondary_memory_needed_ptr.
10938 (rs6000_secondary_memory_needed): Match hook interface.
10939 (rs6000_debug_secondary_memory_needed): Likewise.
10940 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
10941 * config/s390/s390.c (s390_secondary_memory_needed): New function.
10942 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10943 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
10944 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
10945 (sparc_secondary_memory_needed): New function.
10946 * lra-constraints.c (check_and_process_move): Refer to
10947 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
10949 (curr_insn_transform): Likewise.
10950 (process_alt_operands): Use targetm.secondary_memory_needed
10951 instead of TARGET_SECONDARY_MEMORY_NEEDED.
10952 (check_secondary_memory_needed_p): Likewise.
10953 (choose_split_class): Likewise.
10954 * reload.c: Unconditionally include code that was previously
10955 conditional on SECONDARY_MEMORY_NEEDED.
10956 (push_secondary_reload): Use targetm.secondary_memory_needed
10957 instead of TARGET_SECONDARY_MEMORY_NEEDED.
10958 (push_reload): Likewise.
10959 * reload1.c: Unconditionally include code that was previously
10960 conditional on SECONDARY_MEMORY_NEEDED.
10961 (choose_reload_regs): Use targetm.secondary_memory_needed
10962 instead of TARGET_SECONDARY_MEMORY_NEEDED.
10963 (gen_reload): Likewise.
10964 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
10966 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
10967 Alan Hayward <alan.hayward@arm.com>
10968 David Sherwood <david.sherwood@arm.com>
10970 * target.def (secondary_memory_needed_mode): New hook:
10971 * targhooks.c (default_secondary_memory_needed_mode): Declare.
10972 * targhooks.h (default_secondary_memory_needed_mode): New function.
10973 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
10974 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
10975 * doc/tm.texi: Regenerate.
10976 * lra-constraints.c (check_and_process_move): Use
10977 targetm.secondary_memory_needed_mode instead of
10978 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
10979 (curr_insn_transform): Likewise.
10980 * reload.c (get_secondary_mem): Likewise.
10981 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10982 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
10984 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10985 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10986 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
10987 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10988 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
10990 * config/powerpcspe/powerpcspe-protos.h
10991 (rs6000_secondary_memory_needed_mode): Delete.
10992 * config/powerpcspe/powerpcspe.c
10993 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
10994 (rs6000_secondary_memory_needed_mode): Make static.
10995 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
10996 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
10998 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
11000 (rs6000_secondary_memory_needed_mode): Make static.
11001 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11002 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
11003 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11004 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11005 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
11007 (sparc_secondary_memory_needed_mode): New function.
11008 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
11010 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
11012 * config/aarch64/constraints.md (Umq): New constraint.
11013 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
11015 (mov<mode>): Update condition.
11017 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11019 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
11020 when bitposition is the same.
11022 2017-09-13 Richard Biener <rguenther@suse.de>
11024 * dwarf2out.c (output_die_symbol): Remove.
11025 (output_die): Do not output a DIEs symbol.
11027 2017-09-13 Richard Biener <rguenther@suse.de>
11029 PR middle-end/82128
11030 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
11031 default-def to avoid breaking iterator update with the weird
11032 interaction with cgraph_update_edges_for_call_stmt_node.
11034 2017-09-13 Richard Biener <rguenther@suse.de>
11036 * tree-cfg.c (verify_gimple_assign_binary): Add verification
11037 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
11038 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
11039 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
11041 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
11043 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
11044 Disable pc relative literal load irrespective of
11045 TARGET_FIX_ERR_A53_84341 for default.
11047 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
11049 * config/sparc/sparc.c (output_return): Output the source location of
11050 the insn in the delay slot, if any.
11051 (output_sibcall): Likewise.
11053 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
11056 * common.opt (-static-pie): New alias.
11057 (shared): Negate static-pie.
11058 (-no-pie): Update help text.
11060 (static-pie): New option.
11061 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
11062 -static-pie support.
11063 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
11064 (LINK_EH_SPEC): Likewise.
11065 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11066 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
11067 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
11068 * gcc.c (LINK_COMMAND_SPEC): Likewise.
11069 (init_gcc_specs): Likewise.
11070 (init_spec): Likewise.
11071 (display_help): Update help message for -pie.
11072 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
11075 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
11077 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
11078 (movdi_aarch64): Likewise.
11079 (movti_aarch64): Likewise.
11081 2017-09-12 Simon Wright <simon@pushface.org>
11084 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
11085 calculation of the minor version, always output as 0.
11087 2017-09-12 Jakub Jelinek <jakub@redhat.com>
11090 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
11091 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
11092 on it early, rather than manual conversion late. For
11093 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
11094 instead of performing manual conversion.
11096 2017-09-12 Carl Love <cel@us.ibm.com>
11098 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
11099 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
11101 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11102 VMULOSW): Add definitions.
11103 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11104 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11105 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11106 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11107 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11109 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
11111 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
11113 (movti_aarch64): Likewise.
11114 (movdf_aarch64): Likewise.
11115 (movtf_aarch64): Likewise.
11116 (load_pairdi): Likewise.
11117 (store_pairdi): Likewise.
11118 (load_pairdf): Likewise.
11119 (store_pairdf): Likewise.
11120 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
11121 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
11122 (ldr_got_small_<mode>): Likewise.
11123 (ldr_got_small_28k_<mode>): Likewise.
11124 (ldr_got_tiny): Likewise.
11125 * config/aarch64/iterators.md (ldst_sz): New.
11126 (ldpstp_sz): Likewise.
11127 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
11129 (thunderx_load): Split load_8 to load_16.
11130 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
11132 (thunderx2t99_storepair_basic): Split store_8 to store_16.
11133 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
11134 (xgene1_store_pair): Split store_8 to store_16.
11135 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
11136 (falkor_st_0_st_sd): Split store_8 to store_16.
11138 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
11140 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
11141 and store1/2/3/4 to store_4/8/12/16.
11142 * config/aarch64/aarch64.md: Update for rename.
11143 * config/arm/arm.md: Likewise.: Likewise.
11144 * config/arm/arm.c: Likewise.
11145 * config/arm/thumb1.md: Likewise.
11146 * config/arm/thumb2.md: Likewise.
11147 * config/arm/vfp.md: Likewise.
11148 * config/arm/arm-generic.md: Likewise.
11149 * config/arm/arm1020e.md: Likewise.
11150 * config/arm/arm1026ejs.md: Likewise.
11151 * config/arm/arm1136jfs.md: Likewise.
11152 * config/arm/arm926ejs.md: Likewise.
11153 * config/arm/cortex-a15.md: Likewise.
11154 * config/arm/cortex-a17.md: Likewise.
11155 * config/arm/cortex-a5.md: Likewise.
11156 * config/arm/cortex-a53.md: Likewise.
11157 * config/arm/cortex-a57.md: Likewise.
11158 * config/arm/cortex-a7.md: Likewise.
11159 * config/arm/cortex-a8.md: Likewise.
11160 * config/arm/cortex-a9.md: Likewise.
11161 * config/arm/cortex-m4.md: Likewise.
11162 * config/arm/cortex-m7.md: Likewise.
11163 * config/arm/cortex-r4.md: Likewise.
11164 * config/arm/exynos-m1.md: Likewise.
11165 * config/arm/fa526.md: Likewise.
11166 * config/arm/fa606te.md: Likewise.
11167 * config/arm/fa626te.md: Likewise.
11168 * config/arm/fa726te.md: Likewise.
11169 * config/arm/fmp626.md: Likewise.
11170 * config/arm/iwmmxt.md: Likewise.
11171 * config/arm/ldmstm.md: Likewise.
11172 * config/arm/marvell-pj4.md: Likewise.
11173 * config/arm/xgene1.md: Likewise.
11174 * config/aarch64/thunderx.md: Likewise.
11175 * config/aarch64/thunderx2t99.md: Likewise.
11176 * config/aarch64/falkor.md: Likewise.
11178 2017-09-12 Martin Liska <mliska@suse.cz>
11180 * attribs.c (private_lookup_attribute): New function.
11181 * attribs.h (private_lookup_attribute): Declared here.
11182 (lookup_attribute): Called from this place.
11184 2017-09-12 Richard Biener <rguenther@suse.de>
11186 PR tree-optimization/82157
11187 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
11188 stmts with side-effects.
11190 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11191 Alan Hayward <alan.hayward@arm.com>
11192 David Sherwood <david.sherwood@arm.com>
11194 * target.def (hard_regno_nregs): New hook.
11195 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
11196 * targhooks.h (default_hard_regno_nregs): Declare.
11197 * targhooks.c (default_hard_regno_nregs): New function.
11198 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
11199 (TARGET_HARD_REGNO_NREGS): ...this hook.
11200 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
11201 (CLASS_MAX_NREGS): Likewise.
11202 * doc/tm.texi: Regenerate.
11203 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
11204 instead of HARD_REGNO_NREGS.
11205 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
11206 HARD_REGNO_NREGS in the comment.
11207 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
11208 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
11209 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
11210 Return an unsigned int.
11211 (TARGET_HARD_REGNO_NREGS): Redefine.
11212 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
11213 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
11214 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
11215 (arc_hard_regno_nregs): New function.
11216 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
11217 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
11218 (arm_hard_regno_nregs): New function.
11219 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
11220 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
11221 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
11222 (TARGET_HARD_REGNO_NREGS): Redefine.
11223 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
11224 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
11225 (HARD_REGNO_NREGS): Delete.
11226 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
11227 (cr16_hard_regno_nregs): New function.
11228 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
11229 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
11230 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
11231 (cris_hard_regno_nregs): New function.
11232 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
11233 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
11234 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
11235 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
11236 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
11237 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
11238 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
11239 (frv_hard_regno_nregs): Make static. Take and return an
11241 (frv_class_max_nregs): Remove outdated copy of documentation.
11242 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
11243 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
11244 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
11245 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
11246 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
11247 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
11248 (TARGET_HARD_REGNO_NREGS): Redefine.
11249 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
11250 (CLASS_MAX_NREGS): Update comment.
11251 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
11252 (ia64_hard_regno_nregs): New function.
11253 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
11254 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
11255 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
11256 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
11257 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
11259 (m32c_hard_regno_nregs): Likewise. Make static.
11260 (TARGET_HARD_REGNO_NREGS): Redefine.
11261 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
11262 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
11263 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
11264 (m68k_hard_regno_nregs): New function.
11265 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
11266 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
11267 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
11268 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
11269 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
11270 Take and return an unsigned int.
11271 (TARGET_HARD_REGNO_NREGS): Redefine.
11272 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
11273 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
11274 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
11275 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
11276 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
11277 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
11278 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
11279 (msp430_hard_regno_nregs): Make static. Take and return an
11281 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
11282 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
11283 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
11284 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
11285 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
11286 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
11287 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
11288 (TARGET_HARD_REGNO_NREGS): Redefine.
11289 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
11290 (PA_HARD_REGNO_NREGS): ...this.
11291 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
11292 (PA_HARD_REGNO_NREGS): ...this.
11293 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
11294 (pa_hard_regno_nregs): New function.
11295 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
11296 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
11297 (pdp11_hard_regno_nregs): New function.
11298 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
11299 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
11300 (rs6000_hard_regno_nregs_hook): New function.
11301 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
11302 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
11303 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
11304 Take and return an unsigned int. Move earlier in file.
11305 (TARGET_HARD_REGNO_NREGS): Redefine.
11306 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
11307 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
11308 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
11309 (rl78_hard_regno_nregs): Make static. Take and return an
11311 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
11312 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
11313 (rs6000_hard_regno_nregs_hook): New function.
11314 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
11315 * config/rx/rx.c (rx_hard_regno_nregs): New function.
11316 (TARGET_HARD_REGNO_NREGS): Redefine.
11317 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
11318 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
11319 instead of HARD_REGNO_NREGS.
11320 (s390_hard_regno_nregs): New function.
11321 (s390_hard_regno_mode_ok): Add comment from s390.h.
11322 (TARGET_HARD_REGNO_NREGS): Redefine.
11323 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
11324 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
11325 (sh_hard_regno_nregs): New function.
11326 (sh_pass_in_reg_p): Use it.
11327 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
11328 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
11329 (sparc_hard_regno_nregs): New function.
11330 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
11331 * config/spu/spu.c (spu_hard_regno_nregs): New function.
11332 (spu_function_arg_advance): Use it, supplying a valid register number.
11333 (TARGET_HARD_REGNO_NREGS): Redefine.
11334 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
11335 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
11336 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
11337 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
11338 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
11339 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
11340 (CLASS_MAX_NREGS): Remove copy of old documentation.
11341 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
11342 (visium_hard_regno_nregs): New function.
11343 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
11344 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
11345 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
11346 xtensa_hard_regno_nregs): New function.
11347 * system.h (HARD_REGNO_NREGS): Poison.
11349 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11351 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
11352 hard_regno_nregs instead of HARD_REGNO_NREGS.
11353 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
11354 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
11355 (c6x_expand_epilogue): Likewise.
11356 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
11357 (frv_read_iacc_argument): Likewise.
11358 * config/sh/sh.c: Include regs.h.
11359 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
11360 (regs_used): Likewise.
11361 (output_stack_adjust): Likewise.
11362 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
11363 * expmed.c: Include regs.h.
11364 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
11365 * ree.c: Include regs.h.
11366 (combine_reaching_defs): Use hard_regno_nregs instead of
11368 (add_removable_extension): Likewise.
11370 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11372 * regs.h (hard_regno_nregs): Turn into a function.
11373 (end_hard_regno): Update accordingly.
11374 * caller-save.c (setup_save_areas): Likewise.
11375 (save_call_clobbered_regs): Likewise.
11376 (replace_reg_with_saved_mem): Likewise.
11377 (insert_restore): Likewise.
11378 (insert_save): Likewise.
11379 * combine.c (can_change_dest_mode): Likewise.
11380 (move_deaths): Likewise.
11381 (distribute_notes): Likewise.
11382 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
11383 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
11384 (rs6000_split_multireg_move): Likewise.
11385 (rs6000_register_move_cost): Likewise.
11386 (rs6000_memory_move_cost): Likewise.
11387 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
11388 (rs6000_split_multireg_move): Likewise.
11389 (rs6000_register_move_cost): Likewise.
11390 (rs6000_memory_move_cost): Likewise.
11391 * cselib.c (cselib_reset_table): Likewise.
11392 (cselib_lookup_1): Likewise.
11393 * emit-rtl.c (set_mode_and_regno): Likewise.
11394 * function.c (aggregate_value_p): Likewise.
11395 * ira-color.c (setup_profitable_hard_regs): Likewise.
11396 (check_hard_reg_p): Likewise.
11397 (calculate_saved_nregs): Likewise.
11398 (assign_hard_reg): Likewise.
11399 (improve_allocation): Likewise.
11400 (calculate_spill_cost): Likewise.
11401 * ira-emit.c (modify_move_list): Likewise.
11402 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
11403 (ira_hard_reg_in_set_p): Likewise.
11404 * ira.c (setup_reg_mode_hard_regset): Likewise.
11405 (clarify_prohibited_class_mode_regs): Likewise.
11406 (check_allocation): Likewise.
11407 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11408 (lra_setup_reg_renumber): Likewise.
11409 (setup_try_hard_regno_pseudos): Likewise.
11410 (spill_for): Likewise.
11411 (assign_hard_regno): Likewise.
11412 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
11413 * lra-constraints.c (in_class_p): Likewise.
11414 (lra_constraint_offset): Likewise.
11415 (simplify_operand_subreg): Likewise.
11416 (lra_constraints): Likewise.
11417 (split_reg): Likewise.
11418 (split_if_necessary): Likewise.
11419 (invariant_p): Likewise.
11420 (inherit_in_ebb): Likewise.
11421 * lra-lives.c (process_bb_lives): Likewise.
11422 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
11423 (get_hard_regs): Likewise.
11424 (do_remat): Likewise.
11425 * lra-spills.c (assign_spill_hard_regs): Likewise.
11426 * mode-switching.c (create_pre_exit): Likewise.
11427 * postreload.c (reload_combine_recognize_pattern): Likewise.
11428 * recog.c (peep2_find_free_register): Likewise.
11429 * regcprop.c (kill_value_regno): Likewise.
11430 (set_value_regno): Likewise.
11431 (copy_value): Likewise.
11432 (maybe_mode_change): Likewise.
11433 (find_oldest_value_reg): Likewise.
11434 (copyprop_hardreg_forward_1): Likewise.
11435 * regrename.c (check_new_reg_p): Likewise.
11436 (regrename_do_replace): Likewise.
11437 * reload.c (push_reload): Likewise.
11438 (combine_reloads): Likewise.
11439 (find_dummy_reload): Likewise.
11440 (operands_match_p): Likewise.
11441 (find_reloads): Likewise.
11442 (find_equiv_reg): Likewise.
11443 (reload_adjust_reg_for_mode): Likewise.
11444 * reload1.c (count_pseudo): Likewise.
11445 (count_spilled_pseudo): Likewise.
11446 (find_reg): Likewise.
11447 (clear_reload_reg_in_use): Likewise.
11448 (free_for_value_p): Likewise.
11449 (allocate_reload_reg): Likewise.
11450 (choose_reload_regs): Likewise.
11451 (reload_adjust_reg_for_temp): Likewise.
11452 (emit_reload_insns): Likewise.
11453 (delete_output_reload): Likewise.
11454 * rtlanal.c (subreg_get_info): Likewise.
11455 * sched-deps.c (sched_analyze_reg): Likewise.
11456 * sel-sched.c (init_regs_for_mode): Likewise.
11457 (mark_unavailable_hard_regs): Likewise.
11458 (choose_best_reg_1): Likewise.
11459 (verify_target_availability): Likewise.
11460 * valtrack.c (dead_debug_insert_temp): Likewise.
11461 * var-tracking.c (track_loc_p): Likewise.
11462 (emit_note_insn_var_location): Likewise.
11463 * varasm.c (make_decl_rtl): Likewise.
11464 * reginfo.c (choose_hard_reg_mode): Likewise.
11465 (init_reg_modes_target): Refer directly to
11466 this_target_regs->x_hard_regno_nregs.
11468 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11470 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
11471 instead of hard_regno_nregs.
11473 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11475 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
11476 end_hard_regno instead of hard_regno_nregs.
11477 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
11478 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
11479 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
11480 * ira-color.c (improve_allocation): Likewise.
11481 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11482 * lra-lives.c (mark_regno_live): Likewise.
11483 (mark_regno_dead): Likewise.
11484 * lra-remat.c (operand_to_remat): Likewise.
11485 * lra.c (collect_non_operand_hard_regs): Likewise.
11486 * postreload.c (reload_combine_note_store): Likewise.
11487 (move2add_valid_value_p): Likewise.
11488 * reload.c (regno_clobbered_p): Likewise.
11490 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11492 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
11494 * config/v850/v850.c (v850_reorg): Likewise.
11495 * reload.c (refers_to_regno_for_reload_p): Likewise.
11496 (find_equiv_reg): Likewise.
11497 * reload1.c (reload_reg_reaches_end_p): Likewise.
11499 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11501 * caller-save.c (add_used_regs): Use REG_NREGS instead of
11503 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
11504 * config/arm/arm.c (output_move_neon): Likewise.
11505 (arm_attr_length_move_neon): Likewise.
11506 (neon_split_vcombine): Likewise.
11507 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
11508 (c6x_mark_reg_written): Likewise.
11509 (c6x_dwarf_register_span): Likewise.
11510 * config/i386/i386.c (ix86_save_reg): Likewise.
11511 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
11512 (rws_access_reg): Likewise.
11513 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
11514 * mode-switching.c (create_pre_exit): Likewise.
11515 * ree.c (combine_reaching_defs): Likewise.
11516 (add_removable_extension): Likewise.
11517 * regcprop.c (find_oldest_value_reg): Likewise.
11518 (copyprop_hardreg_forward_1): Likewise.
11519 * reload.c (reload_inner_reg_of_subreg): Likewise.
11520 (push_reload): Likewise.
11521 (combine_reloads): Likewise.
11522 (find_dummy_reload): Likewise.
11523 (reload_adjust_reg_for_mode): Likewise.
11524 * reload1.c (find_reload_regs): Likewise.
11525 (forget_old_reloads_1): Likewise.
11526 (reload_reg_free_for_value_p): Likewise.
11527 (reload_adjust_reg_for_temp): Likewise.
11528 (emit_reload_insns): Likewise.
11529 (delete_output_reload): Likewise.
11530 * sel-sched.c (choose_best_reg_1): Likewise.
11531 (choose_best_pseudo_reg): Likewise.
11533 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11534 Alan Hayward <alan.hayward@arm.com>
11535 David Sherwood <david.sherwood@arm.com>
11537 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
11538 * target.def (slow_unaligned_access): New hook.
11539 * targhooks.h (default_slow_unaligned_access): Declare.
11540 * targhooks.c (default_slow_unaligned_access): New function.
11541 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
11542 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
11543 * doc/tm.texi: Regenerate.
11544 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
11545 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
11546 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
11548 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
11549 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
11551 (rs6000_slow_unaligned_access): New function.
11552 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
11553 (expand_block_compare): Likewise.
11554 (expand_strn_compare): Likewise.
11555 (rs6000_rtx_costs): Likewise.
11556 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
11557 (riscv_slow_unaligned_access): Likewise.
11558 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
11559 (riscv_slow_unaligned_access_p): ...this and make static.
11560 (riscv_option_override): Update accordingly.
11561 (riscv_slow_unaligned_access): New function.
11562 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
11563 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
11564 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
11565 (rs6000_slow_unaligned_access): New function.
11566 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
11567 (rs6000_rtx_costs): Likewise.
11568 * config/rs6000/rs6000-string.c (expand_block_compare)
11569 (expand_strn_compare): Use targetm.slow_unaligned_access instead
11570 of SLOW_UNALIGNED_ACCESS.
11571 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
11572 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
11573 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
11574 of SLOW_UNALIGNED_ACCESS.
11575 * expmed.c (simple_mem_bitfield_p): Likewise.
11576 * expr.c (alignment_for_piecewise_move): Likewise.
11577 (emit_group_load_1): Likewise.
11578 (emit_group_store): Likewise.
11579 (copy_blkmode_from_reg): Likewise.
11580 (emit_push_insn): Likewise.
11581 (expand_assignment): Likewise.
11582 (store_field): Likewise.
11583 (expand_expr_real_1): Likewise.
11584 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
11585 * lra-constraints.c (simplify_operand_subreg): Likewise.
11586 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11587 * gimple-ssa-store-merging.c: Likewise in block comment at start
11589 * tree-ssa-strlen.c: Include target.h.
11590 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
11591 of SLOW_UNALIGNED_ACCESS.
11592 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
11594 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11596 PR rtl-optimization/82185
11597 * expmed.c (emit_store_flag_int): Only test tem if it has been
11600 2017-09-12 Richard Biener <rguenther@suse.de>
11602 PR middle-end/82149
11603 * match.pd ((FTYPE) N CMP CST): Fix typo.
11605 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
11607 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
11609 (mips_near_type_p): Add 'short_call' attribute as a synonym
11611 * doc/extend.texi (short_call): Document new function attribute.
11613 2017-09-12 Jakub Jelinek <jakub@redhat.com>
11616 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
11617 assertion check that in the condition.
11618 (get_atomic_generic_size): Likewise. Before testing if parameter
11619 has pointer type, if it has array type, call for C++
11620 default_conversion to perform array-to-pointer conversion.
11622 2017-09-12 Richard Biener <rguenther@suse.de>
11624 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
11625 for operations we cannot scalarize.
11627 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
11629 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
11630 vectors heap vectors. Clean up comments.
11631 Make visited_bbs a reference.
11632 (profitable_jump_thread_path): Make GC
11633 vectors heap vectors. Clean up comments.
11635 (convert_and_register_jump_thread_path): Make GC vectors heap
11637 (check_subpath_and_update_thread_path): Same. Clean up comments.
11638 Make visited_bbs a reference.
11639 (handle_phi): Abstract common code to to
11640 register_jump_thread_path_if_profitable.
11641 Rename VAR_BB to DEF_BB.
11643 Make GC vectors heap vectors.
11644 Make visited_bbs a reference.
11645 (handle_assignment): Same.
11646 (register_jump_thread_path_if_profitable): New.
11647 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
11649 Make GC vectors heap vectors. Clean up comments.
11650 Make visited_bbs a reference.
11651 (find_jump_threads_backwards): Make visited_bbs live in the stack.
11652 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
11655 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
11658 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
11659 words of E_DImode object are reachable by xtensa_uimm8x4 access.
11661 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
11663 Revert r251800 and r251799.
11665 2017-09-11 Martin Jambor <mjambor@suse.cz>
11668 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
11669 arguments in advance.
11670 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
11671 use it to find predecessor edges.
11672 (naive_outof_ssa): Collect vector of predecessors.
11674 2017-09-08 Jason Merrill <jason@redhat.com>
11676 PR c++/70029 - ICE with ref-qualifier and -flto
11677 * langhooks.h (struct lang_hooks_for_types): Add
11678 copy_lang_qualifiers.
11679 * attribs.c (build_type_attribute_qual_variant): Use it.
11680 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
11682 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
11683 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
11685 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
11688 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
11689 (*mulsi3_sp64): New instruction.
11690 (mulsi3): New expander.
11692 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
11694 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
11696 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11698 * sancov.c: Include memmodel.h.
11700 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
11703 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
11706 2017-09-07 Carl Love <cel@us.ibm.com>
11708 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
11709 the sldi instruction.
11711 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
11713 * sancov.c: Include tm_p.h.
11715 2017-09-07 Jakub Jelinek <jakub@redhat.com>
11718 * output.h (switch_to_other_text_partition): New declaration.
11719 * varasm.c (switch_to_other_text_partition): New function.
11720 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
11721 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
11722 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
11723 to the other text partition before emitting LCL label and switch back
11724 after emitting the word after it.
11726 2017-09-07 Richard Biener <rguenther@suse.de>
11728 * passes.def (pass_split_crit_edges): Remove instance before PRE.
11729 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
11730 critical edges here, after loop init.
11731 (pass_data_pre): Remove PROP_no_crit_edges flags.
11732 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
11733 for valueization of call args to avoid leaking VN_TOP.
11734 (visit_use): Assert we do not visit default defs.
11735 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
11736 Use error_mark_node to more easily detect leaking VN_TOP.
11737 All default-defs are varying, not VN_TOP. Mark them visited.
11738 (run_scc_vn): Make code match comment.
11740 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11742 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
11743 OPTION_MASK_FLOAT128_KEYWORD.
11744 (POWERPC_MASKS): Likewise.
11745 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
11746 support for the -mfloat128-type option, and make -mfloat128
11747 default on PowerPC Linux systems. Define or undefine
11748 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
11749 Define __float128 to be __ieee128 if IEEE 128-bit support is
11750 enabled, or undefine it.
11751 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
11752 Delete defining __FLOAT128_TYPE__.
11753 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
11754 -mfloat128-type option and make -mfloat128 default on PowerPC
11756 (TARGET_FLOAT128_TYPE): Likewise.
11757 (-mfloat128-type): Likewise.
11758 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11759 Delete the -mfloat128-type option and make -mfloat128 default on
11760 PowerPC Linux systems. Always use __ieee128 to be the keyword for
11761 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
11762 128-bit floating point is enabled. Change tests from using
11763 -mfloat128-type to -mfloat128.
11764 (rs6000_mangle_type): Use the correct mangling for the __float128
11765 type even if normal long double is restricted to 64-bits.
11766 (floatn_mode): Enable the _Float128 type by default on VSX Linux
11768 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
11769 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
11770 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
11772 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
11773 documentation for -mfloat128.
11775 2017-09-06 Olivier Hainque <hainque@adacore.com>
11777 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
11779 2017-09-06 Wish Wu <wishwu007@gmail.com>
11780 Jakub Jelinek <jakub@redhat.com>
11782 * asan.c (initialize_sanitizer_builtins): Add
11783 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
11784 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
11785 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
11786 BT_FN_VOID_UINT64_PTR variables.
11787 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
11788 (BT_FN_VOID_UINT16_UINT16): Likewise.
11789 (BT_FN_VOID_UINT32_UINT32): Likewise.
11790 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
11791 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
11792 (BT_FN_VOID_UINT64_PTR): Likewise.
11793 * common.opt (flag_sanitize_coverage): New variable.
11794 (fsanitize-coverage=trace-pc): Remove.
11795 (fsanitize-coverage=): Add.
11796 * flag-types.h (enum sanitize_coverage_code): New enum.
11797 * fold-const.c (fold_range_test): Disable non-short-circuit
11798 optimization if flag_sanitize_coverage.
11799 (fold_truth_andor): Likewise.
11800 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
11801 * opts.c (COVERAGE_SANITIZER_OPT): Define.
11802 (coverage_sanitizer_opts): New array.
11803 (get_closest_sanitizer_option): Add OPTS argument, handle also
11804 OPT_fsanitize_coverage_.
11805 (parse_sanitizer_options): Adjusted to also handle
11806 OPT_fsanitize_coverage_.
11807 (common_handle_option): Add OPT_fsanitize_coverage_.
11808 * sancov.c (instrument_comparison, instrument_switch): New function.
11809 (sancov_pass): Add trace-cmp support.
11810 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
11811 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
11812 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
11813 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
11814 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
11815 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
11816 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
11817 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
11818 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
11819 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
11821 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
11823 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
11824 error. Only quit immediately if parsing is complete.
11825 (BEGIN): Initialize fatal_err and parse_done.
11826 (begin fpu, end fpu): Check number of arguments.
11827 (begin arch, end arch): Likewise.
11828 (begin cpu, end cpu): Likewise.
11829 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
11830 (optalias): Likewise.
11832 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
11834 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
11835 * config/arm/arm-isa.h: Delete. Move definitions to ...
11836 * arm-cpus.in: ... here. Use new feature and fgroup values.
11837 * config/arm/arm.c (arm_option_override): Use lower case for feature
11839 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
11840 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
11841 * config/arm/parsecpu.awk (END): Add new command 'isa'.
11843 (print_isa_bits_for): New function.
11844 (gen_isa): New function.
11845 (gen_comm_data): Use print_isa_bits_for.
11846 (define feature): New keyword.
11847 (define fgroup): New keyword.
11848 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
11849 (arm-isa.h): Add rule to generate file.
11850 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
11851 case for feature bit names.
11853 2017-09-06 Richard Biener <rguenther@suse.de>
11855 * tree-ssa-pre.c (NECESSARY): Remove.
11856 (create_expression_by_pieces): Do not touch pass-local flags.
11857 (insert_into_preds_of_block): Likewise.
11858 (do_pre_regular_insertion): Likewise.
11859 (eliminate_insert): Likewise.
11860 (eliminate_dom_walker::before_dom_children): Likewise.
11861 (fini_eliminate): Do not look at inserted_exprs.
11862 (mark_operand_necessary): Remove.
11863 (remove_dead_inserted_code): Replace with simple work-list
11864 algorithm based on inserted_exprs and SSA uses.
11865 (pass_pre::execute): Re-order fini_eliminate and
11866 remove_dead_inserted_code.
11868 2017-09-06 Olivier Hainque <hainque@adacore.com>
11870 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
11871 for VxWorks 7. Adjust surrounding comments.
11873 2017-09-06 Richard Biener <rguenther@suse.de>
11875 * gimple-ssa-strength-reduction.c
11876 (find_candidates_dom_walker::before_dom_children): Also allow
11879 2017-09-06 Richard Biener <rguenther@suse.de>
11881 PR tree-optimization/82108
11882 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
11883 for gap in the non-permutation SLP case.
11885 2017-09-06 Martin Jambor <mjambor@suse.cz>
11887 PR tree-optimization/82078
11888 * tree-sra.c (sort_and_splice_var_accesses): Move call to
11889 add_access_to_work_queue...
11890 (build_accesses_from_assign): ...here.
11891 (propagate_all_subaccesses): Make sure racc is the group
11892 representative, if there is one.
11894 2017-09-06 Jakub Jelinek <jakub@redhat.com>
11896 PR middle-end/82095
11897 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
11900 2017-09-06 Richard Biener <rguenther@suse.de>
11902 * gimple-ssa-strength-reduction.c
11903 (find_candidates_dom_walker::before_doom_children): Use a
11904 type and not a mode check.
11906 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
11909 * config/arm/predicates.md (arm_general_adddi_operand): Create new
11911 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
11913 2017-09-05 Jeff Law <law@redhat.com>
11915 PR tree-optimization/64910
11916 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
11917 cases where we have 3 or more operands.
11919 2017-09-05 Jakub Jelinek <jakub@redhat.com>
11921 PR middle-end/81768
11922 * omp-low.c (lower_omp_for): Recompute tree invariant if
11923 gimple_omp_for_initial/final is ADDR_EXPR.
11925 PR middle-end/81768
11926 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
11927 into gimple val before gimplification fo the COND_EXPR.
11929 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
11931 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
11932 REGION_COPY argument.
11933 (thread_through_all_blocks): Remove unused argument to
11934 duplicate_thread_path.
11936 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11937 Alan Hayward <alan.hayward@arm.com>
11938 David Sherwood <david.sherwood@arm.com>
11940 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
11941 Take a scalar_mode rather than a machine_mode.
11942 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11943 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
11944 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
11945 (aarch64_gen_adjusted_ldpstp): Likewise.
11946 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
11948 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11949 Alan Hayward <alan.hayward@arm.com>
11950 David Sherwood <david.sherwood@arm.com>
11952 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
11953 Take a scalar_int_mode instead of a machine_mode.
11954 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
11955 (aarch64_output_scalar_simd_mov_immediate): Likewise.
11956 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
11957 (aarch64_simd_attr_length_rglist): Delete.
11958 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
11959 a scalar_int_mode instead of a machine_mode.
11960 (aarch64_add_offset): Likewise.
11961 (aarch64_internal_mov_immediate): Likewise
11962 (aarch64_add_constant_internal): Likewise.
11963 (aarch64_add_constant): Likewise.
11964 (aarch64_movw_imm): Likewise.
11965 (aarch64_rtx_arith_op_extract_p): Likewise.
11966 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
11967 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
11968 Remove assert that the mode isn't a vector.
11969 (aarch64_output_scalar_simd_mov_immediate): Likewise.
11970 (aarch64_expand_mov_immediate): Update calls after above changes.
11971 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
11972 (aarch64_and_bitmask_imm): Check for scalar integer modes.
11973 (aarch64_move_imm): Likewise.
11974 (aarch64_can_const_movi_rtx_p): Likewise.
11975 (aarch64_strip_extend): Likewise.
11976 (aarch64_extr_rtx_p): Likewise.
11977 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
11978 a CONST_INT when the mode parameter is VOIDmode.
11979 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
11981 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11983 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
11984 * stor-layout.c (bitwise_mode_for_mode): Likewise.
11985 (bitwise_type_for_mode): Update accordingly.
11987 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11989 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
11990 * stor-layout.c (mode_for_size_tree): Likewise.
11991 (mode_for_array): Update accordingly.
11992 (layout_decl): Likewise.
11993 (compute_record_mode): Likewise. Only set the mode once.
11995 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
11997 * target.def (get_mask_mode): Change return type to opt_mode.
11999 * doc/tm.texi: Regenerate.
12000 * targhooks.h (default_get_mask_mode): Return an opt_mode.
12001 * targhooks.c (default_get_mask_mode): Likewise.
12002 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12003 * optabs-query.c (can_vec_mask_load_store_p): Update use of
12004 targetm.get_mask_mode.
12005 * tree.c (build_truth_vector_type): Likewise.
12007 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12009 * machmode.h (mode_for_vector): Return an opt_mode.
12010 * stor-layout.c (mode_for_vector): Likewise.
12011 (mode_for_int_vector): Update accordingly.
12012 (layout_type): Likewise.
12013 * config/i386/i386.c (emit_memmov): Likewise.
12014 (ix86_expand_set_or_movmem): Likewise.
12015 (ix86_expand_vector_init): Likewise.
12016 (ix86_get_mask_mode): Likewise.
12017 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
12019 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
12020 * expmed.c (extract_bit_field_1): Likewise.
12021 * expr.c (expand_expr_real_2): Likewise.
12022 * optabs-query.c (can_vec_perm_p): Likewise.
12023 (can_vec_mask_load_store_p): Likewise.
12024 * optabs.c (expand_vec_perm): Likewise.
12025 * targhooks.c (default_get_mask_mode): Likewise.
12026 * tree-vect-stmts.c (vectorizable_store): Likewise.
12027 (vectorizable_load): Likewise.
12028 (get_vectype_for_scalar_type_and_size): Likewise.
12030 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12032 * machmode.h (mode_for_int_vector): New function.
12033 * stor-layout.c (mode_for_int_vector): Likewise.
12034 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
12035 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
12036 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12037 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12038 (s390_expand_vcond): Likewise.
12040 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12042 * machmode.h (opt_machine_mode): New type.
12043 (opt_mode<T>): Allow construction from anything that can be
12045 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
12046 (mode_for_size): Return an opt_machine_mode.
12047 * stor-layout.c (mode_for_size): Likewise.
12048 (mode_for_size_tree): Update call accordingly.
12049 (bitwise_mode_for_mode): Likewise.
12050 (make_fract_type): Likewise.
12051 (make_accum_type): Likewise.
12052 * caller-save.c (replace_reg_with_saved_mem): Update call
12054 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12055 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12056 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12057 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12058 * expmed.c (extract_bit_field_1): Likewise.
12059 * reload.c (get_secondary_mem): Likewise.
12060 * varasm.c (assemble_integer): Likewise.
12061 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
12064 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12066 * machmode.h (decimal_float_mode_for_size): New function.
12067 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
12068 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
12069 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
12070 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
12071 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
12072 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
12074 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12076 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
12077 (get_builtin_sync_mode): Likewise.
12078 (expand_ifn_atomic_compare_exchange): Likewise.
12079 (expand_builtin_atomic_clear): Likewise.
12080 (expand_builtin_atomic_test_and_set): Likewise.
12081 (fold_builtin_atomic_always_lock_free): Likewise.
12082 * calls.c (compute_argument_addresses): Likewise.
12083 (emit_library_call_value_1): Likewise.
12084 (store_one_arg): Likewise.
12085 * combine.c (combine_instructions): Likewise.
12086 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
12087 * config/arm/arm.c (arm_function_value): Likewise.
12088 (aapcs_allocate_return_reg): Likewise.
12089 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
12090 * config/i386/i386.c (construct_container): Likewise.
12091 (ix86_gimplify_va_arg): Likewise.
12092 (ix86_expand_sse_cmp): Likewise.
12093 (emit_memmov): Likewise.
12094 (emit_memset): Likewise.
12095 (expand_small_movmem_or_setmem): Likewise.
12096 (ix86_expand_pextr): Likewise.
12097 (ix86_expand_pinsr): Likewise.
12098 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
12099 * config/microblaze/microblaze.c (microblaze_block_move_straight):
12101 * config/mips/mips.c (mips_function_value_1) Likewise.
12102 (mips_block_move_straight): Likewise.
12103 (mips_expand_ins_as_unaligned_store): Likewise.
12104 * config/powerpcspe/powerpcspe.c
12105 (rs6000_darwin64_record_arg_advance_flush): Likewise.
12106 (rs6000_darwin64_record_arg_flush): Likewise.
12107 * config/rs6000/rs6000.c
12108 (rs6000_darwin64_record_arg_advance_flush): Likewise.
12109 (rs6000_darwin64_record_arg_flush): Likewise.
12110 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
12111 (sparc_function_value_1): Likewise.
12112 * config/spu/spu.c (adjust_operand): Likewise.
12113 (spu_emit_branch_or_set): Likewise.
12114 (arith_immediate_p): Likewise.
12115 * emit-rtl.c (gen_lowpart_common): Likewise.
12116 * expr.c (expand_expr_real_1): Likewise.
12117 * function.c (assign_parm_setup_block): Likewise.
12118 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
12119 * reload1.c (alter_reg): Likewise.
12120 * stor-layout.c (mode_for_vector): Likewise.
12121 (layout_type): Likewise.
12123 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12125 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
12126 (spu_convert_move): Likewise.
12127 * lower-subreg.c (resolve_simple_move): Likewise.
12129 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12132 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
12133 define_insn to a define_expand.
12134 (altivec_vsum2sws_direct): New define_insn.
12135 (altivec_vsumsws): Convert from a define_insn to a define_expand.
12137 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
12139 * config/arm/arm.c (arm_option_params_internal): Improve setting of
12142 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
12147 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
12148 realign stack if stack alignment needed is less than incoming
12151 2017-09-05 Marek Polacek <polacek@redhat.com>
12154 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
12157 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
12159 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
12161 2017-09-05 Richard Biener <rguenther@suse.de>
12163 PR tree-optimization/82084
12164 * fold-const.c (can_native_encode_string_p): Handle wide characters.
12166 2017-09-05 Richard Biener <rguenther@suse.de>
12168 PR tree-optimization/82102
12169 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
12171 2017-09-05 Martin Liska <mliska@suse.cz>
12173 PR tree-optimization/82032
12174 * tree-cfg.c (generate_range_test): New function.
12175 * tree-cfg.h (generate_range_test): Declared here.
12176 * tree-cfgcleanup.c (convert_single_case_switch): New function.
12177 (cleanup_control_expr_graph): Use it.
12178 * tree-switch-conversion.c (try_switch_expansion): Remove
12180 (emit_case_nodes): Use generate_range_test.
12182 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
12185 * config/i386/i386.md (*<btsc><mode>_mask): Add
12186 TARGET_USE_BT to insn constraint.
12187 (*btr<mode>_mask): Ditto.
12189 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
12191 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
12192 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
12194 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
12197 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
12198 TARGET_NEON and TARGET_IWMMXT.
12199 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
12200 TARGET_NEON and TARGET_IWMMXT.
12201 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
12203 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
12205 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
12206 (ix86_rewrite_tls_address): Ditto.
12207 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
12208 (ix86_rewrite_tls_address_1): Ditto.
12209 (ix86_rewrite_tls_address): Ditto.
12210 * config/i386/predicates.md (tls_address_pattern): New predicate.
12211 * config/i386/i386.md (TLS address splitter): New splitter.
12213 2017-09-04 Richard Biener <rguenther@suse.de>
12215 PR tree-optimization/82084
12216 * fold-const.h (can_native_encode_string_p): Declare.
12217 * fold-const.c (can_native_encode_string_p): Factor out from ...
12218 (native_encode_string): ... here.
12219 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
12220 vectorizing stores from constants we later cannot handle.
12222 2017-09-04 Marek Polacek <polacek@redhat.com>
12225 * doc/invoke.texi: Update -Wtautological-compare documentation.
12227 2017-09-04 Jeff Law <law@redhat.com>
12229 PR tree-optimization/64910
12230 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
12231 swap the first and last operand if the last is a constant.
12233 2017-09-04 Marek Polacek <polacek@redhat.com>
12236 * convert.c (do_narrow): When sanitizing signed integer overflows,
12237 bail out for signed types.
12238 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
12240 2017-09-04 Richard Biener <rguenther@suse.de>
12242 PR tree-optimization/82060
12243 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
12244 Move devirtualization after stmt folding and before EH/AB/noreturn
12245 cleanup to get the stmt refs canonicalized. Use a bool instead
12246 of gimple_modified_p since that doesn't work for NOPs. Schedule
12247 NOPs generated by folding for removal.
12249 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12250 Alan Hayward <alan.hayward@arm.com>
12251 David Sherwood <david.sherwood@arm.com>
12253 * coretypes.h (pad_direction): New enum.
12254 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
12255 (FUNCTION_ARG_PADDING): Likewise.
12256 * target.def (function_arg_padding): New hook.
12257 * targhooks.h (default_function_arg_padding): Declare.
12258 * targhooks.c (default_function_arg_padding): New function.
12259 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
12260 (TARGET_FUNCTION_ARG_PADDING): ...this.
12261 * doc/tm.texi: Regenerate.
12262 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
12263 instead of direction.
12264 (compute_argument_addresses): Likewise.
12265 (load_register_parameters): Likewise.
12266 (emit_library_call_value_1): Likewise.
12267 (store_one_arg): Use targetm.calls.function_arg_padding instead
12268 of FUNCTION_ARG_PADDING.
12269 (must_pass_in_stack_var_size_or_pad): Likewise.
12270 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
12271 (emit_group_store): Likewise.
12272 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
12273 instead of FUNCTION_ARG_PADDING.
12274 (emit_push_insn): Likewise, and propagate enum change throughout
12276 * function.h (direction): Delete.
12277 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
12279 * function.c (assign_parm_find_stack_rtl): Likewise.
12280 (assign_parm_setup_block_p): Likewise.
12281 (assign_parm_setup_block): Likewise.
12282 (gimplify_parameters): Likewise.
12283 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
12284 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
12286 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
12287 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12288 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
12289 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
12290 (aarch64_function_arg_padding): ...this new function.
12291 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
12292 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12293 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
12294 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12295 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
12296 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12297 (arm_pad_arg_upward): Replace with...
12298 (arm_function_arg_padding): ...this new function.
12299 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
12301 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
12302 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
12303 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12304 (ia64_hpux_function_arg_padding): Replace with...
12305 (ia64_function_arg_padding): ...this new function. Use pad_direction
12306 instead of direction. Check for TARGET_HPUX.
12307 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
12308 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12309 (iq2000_function_arg_padding): New function.
12310 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
12311 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
12312 (mips_function_arg_padding): ...this new function.
12313 (mips_pad_reg_upward): Update accordingly.
12314 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12315 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
12316 targetm.calls.function_arg_padding.
12317 (FUNCTION_ARG_PADDING): Delete.
12318 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12319 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
12320 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
12321 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
12322 (nios2_block_reg_padding): Return pad_direction instead of direction.
12323 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
12324 instead of direction.
12325 (nios2_function_arg_padding): Likewise. Make static.
12326 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12327 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
12328 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
12329 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
12330 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12331 (pa_function_arg_padding): Make static. Return pad_direction instead
12333 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
12334 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
12335 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
12336 instead of direction. Use targetm.calls.function_arg_padding.
12337 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
12338 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
12339 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
12340 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
12341 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
12343 (function_arg_padding): Rename to...
12344 (rs6000_function_arg_padding): ...this. Make static. Return
12345 pad_direction instead of direction.
12346 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
12347 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
12348 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
12349 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
12350 instead of direction. Use targetm.calls.function_arg_padding.
12351 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
12352 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
12353 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
12354 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
12355 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12356 (function_arg_padding): Rename to...
12357 (rs6000_function_arg_padding): ...this. Make static. Return
12358 pad_direction instead of direction.
12359 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
12360 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
12361 * config/s390/s390.c (s390_function_arg_padding): New function.
12362 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12363 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
12364 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
12365 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12366 (function_arg_padding): Rename to...
12367 (sparc_function_arg_padding): ...this. Make static. Return
12368 pad_direction instead of direction.
12369 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
12370 * config/spu/spu.c (spu_function_arg_padding): New function.
12371 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12372 * system.h (FUNCTION_ARG_PADDING): Poison.
12374 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12375 Alan Hayward <alan.hayward@arm.com>
12376 David Sherwood <david.sherwood@arm.com>
12378 * target.def (modes_tieable_p): New hook.
12379 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
12380 (TARGET_MODES_TIEABLE_P): ...this.
12381 * doc/tm.texi.in: Regenerate.
12382 * hooks.h (hook_bool_mode_mode_true): Declare.
12383 * hooks.c (hook_bool_mode_mode_true): New function.
12384 * combine.c (subst): Use targetm.modes_tieable_p instead of
12386 * dse.c (find_shift_sequence): Likewise.
12387 * expmed.c (extract_low_bits): Likewise.
12388 * lower-subreg.c: Include target.h.
12389 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
12391 * rtlanal.c (rtx_cost): Likewise.
12392 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
12393 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
12394 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
12395 (TARGET_MODES_TIEABLE_P): Redefine.
12396 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
12397 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
12398 (TARGET_MODES_TIEABLE_P): Redefine.
12399 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
12400 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
12401 (arc_modes_tieable_p): New function.
12402 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
12403 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
12404 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
12405 (arm_modes_tieable_p): Make static.
12406 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
12407 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
12408 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
12409 (TARGET_MODES_TIEABLE_P): Redefine.
12410 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
12411 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
12412 (TARGET_MODES_TIEABLE_P): Redefine.
12413 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
12414 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
12415 (cr16_modes_tieable_p): New function.
12416 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
12417 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
12418 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
12419 (TRULY_NOOP_TRUNCATION): Update comment.
12420 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
12421 (TRULY_NOOP_TRUNCATION): Update comment.
12422 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
12423 (frv_modes_tieable_p): New function.
12424 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
12425 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
12426 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
12427 (TARGET_MODES_TIEABLE_P): Redefine.
12428 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
12429 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
12430 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
12431 (TARGET_MODES_TIEABLE_P): Redefine.
12432 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
12433 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
12434 (ia64_modes_tieable_p): New function.
12435 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
12436 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
12437 (iq2000_modes_tieable_p): New function.
12438 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
12439 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
12440 (lm32_modes_tieable_p): New function.
12441 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
12442 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
12443 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
12444 (TARGET_MODES_TIEABLE_P): Redefine.
12445 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
12446 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
12447 (m32r_modes_tieable_p): New function.
12448 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
12449 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
12450 (m68k_modes_tieable_p): New function.
12451 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
12452 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
12453 (mcore_modes_tieable_p): New function.
12454 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
12455 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
12457 (TARGET_MODES_TIEABLE_P): Redefine.
12458 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
12459 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
12460 * config/mips/mips.c (mips_modes_tieable_p): Make static.
12461 (TARGET_MODES_TIEABLE_P): Redefine.
12462 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
12463 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
12464 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
12465 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
12466 (mn10300_modes_tieable_p): ...this and make static.
12467 (TARGET_MODES_TIEABLE_P): Redefine.
12468 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
12469 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
12470 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
12471 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
12472 (msp430_modes_tieable_p): Make static.
12473 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
12474 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
12475 (TARGET_MODES_TIEABLE_P): Redefine.
12476 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
12477 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
12478 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
12479 (TARGET_MODES_TIEABLE_P): Redefine.
12480 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
12481 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
12482 * config/pa/pa.c (pa_modes_tieable_p): Make static.
12483 (TARGET_MODES_TIEABLE_P): Redefine.
12484 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
12485 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
12486 (pdp11_modes_tieable_p): New function.
12487 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
12488 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
12489 (rs6000_modes_tieable_p): New function.
12490 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
12491 * config/powerpcspe/powerpcspe.md: Update comment.
12492 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
12493 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
12494 (TARGET_MODES_TIEABLE_P): Redefine.
12495 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
12496 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
12497 (rl78_modes_tieable_p): New function.
12498 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
12499 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
12500 (rs6000_modes_tieable_p): New function.
12501 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
12502 * config/rs6000/rs6000.md: Update comment.
12503 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
12504 * config/rx/rx.c (rx_modes_tieable_p): New function.
12505 (TARGET_MODES_TIEABLE_P): Redefine.
12506 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
12507 * config/s390/s390.c (s390_modes_tieable_p): New function.
12508 (TARGET_MODES_TIEABLE_P): Redefine.
12509 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
12510 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
12511 (sh_modes_tieable_p): New function.
12512 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
12513 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
12514 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
12515 (sparc_modes_tieable_p): Make static.
12516 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
12517 * config/spu/spu.c (spu_modes_tieable_p): New function.
12518 (TARGET_MODES_TIEABLE_P): Redefine.
12519 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
12520 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
12521 (TARGET_MODES_TIEABLE_P): Redefine.
12522 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
12523 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
12524 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
12525 * config/v850/v850.c (v850_modes_tieable_p): New function.
12526 (TARGET_MODES_TIEABLE_P): Redefine.
12527 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
12528 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
12529 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
12530 (visium_modes_tieable_p): New function.
12531 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
12532 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
12533 (xtensa_modes_tieable_p): New function.
12534 * system.h (MODES_TIEABLE_P): Poison.
12536 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12537 Alan Hayward <alan.hayward@arm.com>
12538 David Sherwood <david.sherwood@arm.com>
12540 * target.def (hard_regno_mode_ok): New hook.
12541 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
12542 (TARGET_HARD_REGNO_MODE_OK): ...this.
12543 * doc/tm.texi.in: Regenerate.
12544 * hooks.h (hook_bool_uint_mode_true): Declare.
12545 * hooks.c (hook_bool_uint_mode_true): New function.
12546 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
12547 HARD_REGNO_MODE_OK.
12548 * genpreds.c (write_insn_preds_c): Add an include of target.h.
12549 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
12550 instead of HARD_REGNO_MODE_OK.
12551 * caller-save.c: Include target.h.
12552 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
12553 HARD_REGNO_MODE_OK.
12554 * combine.c (can_combine_p): Likewise.
12555 (combinable_i3pat): Likewise.
12556 (can_change_dest_mode): Likewise.
12557 * expr.c (init_expr_target): Likewise.
12558 (convert_move): Likewise.
12559 (convert_modes): Likewise.
12560 * ira.c (setup_prohibited_class_mode_regs): Likewise.
12561 (setup_prohibited_mode_move_regs): Likewise.
12562 * ira.h (target_ira): Likewise.
12563 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12564 * lra-constraints.c (process_alt_operands): Likewise.
12565 (split_reg): Likewise.
12566 * recog.c (peep2_find_free_register): Likewise.
12567 * ree.c (combine_reaching_defs): Likewise.
12568 * regcprop.c (maybe_mode_change): Likewise.
12569 * reginfo.c (init_reg_sets_1): Likewise.
12570 (choose_hard_reg_mode): Likewise.
12571 (simplifiable_subregs): Likewise.
12572 * regrename.c (check_new_reg_p): Likewise.
12573 * reload.c (find_valid_class): Likewise.
12574 (find_valid_class_1): Likewise.
12575 (reload_inner_reg_of_subreg): Likewise.
12576 (push_reload): Likewise.
12577 (combine_reloads): Likewise.
12578 (find_dummy_reload): Likewise.
12579 (find_reloads): Likewise.
12580 * reload1.c (find_reg): Likewise.
12581 (set_reload_reg): Likewise.
12582 (allocate_reload_reg): Likewise.
12583 (choose_reload_regs): Likewise.
12584 (reload_adjust_reg_for_temp): Likewise.
12585 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
12586 (simplify_subreg_regno): Likewise.
12587 * sel-sched.c (init_regs_for_mode): Likewise.
12588 * varasm.c (make_decl_rtl): Likewise.
12589 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
12590 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
12591 HARD_REGNO_MODE_OK.
12592 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
12593 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
12594 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12595 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
12596 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
12597 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12598 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
12599 (arc_mode_class): Delete.
12600 (HARD_REGNO_MODE_OK): Delete.
12601 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12602 (arc_hard_regno_mode_ok): Rename old array to...
12603 (arc_hard_regno_mode_ok_modes): ...this.
12604 (arc_conditional_register_usage): Update accordingly.
12605 (arc_mode_class): Make static.
12606 (arc_hard_regno_mode_ok): New function.
12607 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
12608 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
12609 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12610 (arm_hard_regno_mode_ok): Make static.
12611 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
12612 HARD_REGNO_MODE_OK.
12613 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
12614 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
12615 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
12617 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12618 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
12619 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
12620 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
12621 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
12622 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12623 * config/bfin/predicates.md (valid_reg_operand): Use
12624 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
12625 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
12626 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
12627 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12628 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
12629 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
12630 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12631 (cr16_hard_regno_mode_ok): Make static and return a bool.
12632 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
12633 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12634 (cris_hard_regno_mode_ok): New function.
12635 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
12636 (epiphany_mode_class): Delete.
12637 (HARD_REGNO_MODE_OK): Delete.
12638 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
12639 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12640 (hard_regno_mode_ok): Rename to...
12641 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
12642 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
12643 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
12644 HARD_REGNO_MODE_OK.
12645 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
12646 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
12647 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12648 (frv_hard_regno_mode_ok): Make static and return a bool.
12649 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
12650 HARD_REGNO_MODE_OK.
12651 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
12652 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
12653 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
12654 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
12656 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12657 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
12658 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
12659 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
12661 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12662 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
12663 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12664 (ia64_hard_regno_mode_ok): New function.
12665 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
12666 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12667 (iq2000_hard_regno_mode_ok): New function.
12668 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
12669 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12670 (lm32_hard_regno_mode_ok): New function.
12671 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
12672 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
12673 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
12674 instead of HARD_REGNO_MODE_OK.
12675 (m32c_hard_regno_ok): Rename to...
12676 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
12677 (m32c_cannot_change_mode_class): Update accordingly.
12678 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12679 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
12680 (m32r_mode_class): Delete.
12681 (HARD_REGNO_MODE_OK): Delete.
12682 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12683 (m32r_hard_regno_mode_ok): Rename to...
12684 (m32r_hard_regno_modes): ...this.
12685 (m32r_mode_class): Make static.
12686 (m32r_hard_regno_mode_ok): New function.
12687 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
12688 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
12689 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12690 (m68k_hard_regno_mode_ok): Make static.
12691 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
12692 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12693 (mcore_hard_regno_mode_ok): New function.
12694 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
12695 (HARD_REGNO_MODE_OK): Delete.
12696 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
12698 (microblaze_hard_regno_mode_ok_p): ...this and make static.
12699 (microblaze_hard_regno_mode_ok): New function.
12700 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12701 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
12702 (mips_hard_regno_mode_ok): Delete.
12703 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
12704 (mips_hard_regno_mode_ok_p): ...this and make static.
12705 (mips_hard_regno_mode_ok_p): Rename to...
12706 (mips_hard_regno_mode_ok_uncached): ...this.
12707 (mips_hard_regno_mode_ok): New function.
12708 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
12709 of HARD_REGNO_MODE_OK.
12710 (mips_option_override): Update after above name changes.
12711 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12712 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
12713 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
12714 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
12715 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
12716 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12717 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
12718 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
12719 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
12720 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12721 (msp430_hard_regno_mode_ok): Make static and return a bool.
12722 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
12723 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
12724 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
12726 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12727 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
12728 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
12729 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
12730 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
12731 (PA_HARD_REGNO_MODE_OK): ...this
12732 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
12733 (PA_HARD_REGNO_MODE_OK): ...this.
12734 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12735 (pa_hard_regno_mode_ok): New function.
12736 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
12737 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12738 (pdp11_hard_regno_mode_ok): New function.
12739 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
12740 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
12742 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
12744 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12745 (rs6000_hard_regno_mode_ok): Rename to...
12746 (rs6000_hard_regno_mode_ok_uncached): ...this.
12747 (rs6000_init_hard_regno_mode_ok): Update accordingly.
12748 (rs6000_hard_regno_mode_ok): New function.
12749 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
12750 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
12751 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
12752 (riscv_hard_regno_mode_ok): ...this and make static.
12753 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12754 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
12755 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
12756 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12757 (rl78_hard_regno_mode_ok): Make static and return bool.
12758 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
12759 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
12761 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
12762 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12763 (rs6000_hard_regno_mode_ok): Rename to...
12764 (rs6000_hard_regno_mode_ok_uncached): ...this.
12765 (rs6000_init_hard_regno_mode_ok): Update accordingly.
12766 (rs6000_hard_regno_mode_ok): New function.
12767 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
12768 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
12769 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12770 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
12771 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
12772 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
12773 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12774 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
12775 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
12776 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12777 (sh_hard_regno_mode_ok): Make static.
12778 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
12779 instead of HARD_REGNO_MODE_OK.
12780 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
12781 (sparc_mode_class): Delete.
12782 (HARD_REGNO_MODE_OK): Delete.
12783 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12784 (hard_regno_mode_classes): Make static.
12785 (sparc_mode_class): Likewise.
12786 (sparc_hard_regno_mode_ok): New function.
12787 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
12788 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
12789 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
12791 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12792 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
12793 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
12794 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
12795 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
12796 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12797 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
12798 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
12799 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
12800 (visium_hard_regno_mode_ok): New function.
12801 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
12802 instead of HARD_REGNO_MODE_OK.
12803 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
12804 (HARD_REGNO_MODE_OK): Delete.
12805 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
12806 (xtensa_hard_regno_mode_ok_p): ...this and make static.
12807 (xtensa_option_override): Update accordingly.
12808 (TARGET_HARD_REGNO_MODE_OK): Redefine.
12809 (xtensa_hard_regno_mode_ok): New function.
12810 * system.h (HARD_REGNO_MODE_OK): Poison.
12812 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12813 Alan Hayward <alan.hayward@arm.com>
12814 David Sherwood <david.sherwood@arm.com>
12816 * target.def (hard_regno_call_part_clobbered): New hook.
12817 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
12818 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
12819 * doc/tm.texi: Regenerate.
12820 * hooks.h (hook_bool_uint_mode_false): Declare.
12821 * hooks.c (hook_bool_uint_mode_false): New function.
12822 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12823 * cselib.c (cselib_process_insn): Use
12824 targetm.hard_regno_call_part_clobbered instead of
12825 HARD_REGNO_CALL_PART_CLOBBERED.
12826 * ira-conflicts.c (ira_build_conflicts): Likewise.
12827 * ira-costs.c (ira_tune_allocno_costs): Likewise.
12828 * lra-constraints.c (need_for_call_save_p): Likewise.
12829 * lra-lives.c: Include target.h.
12830 (check_pseudos_live_through_calls): Use
12831 targetm.hard_regno_call_part_clobbered instead of
12832 HARD_REGNO_CALL_PART_CLOBBERED.
12833 * regcprop.c: Include target.h.
12834 (copyprop_hardreg_forward_1): Use
12835 targetm.hard_regno_call_part_clobbered instead of
12836 HARD_REGNO_CALL_PART_CLOBBERED.
12837 * reginfo.c (choose_hard_reg_mode): Likewise.
12838 * regrename.c (check_new_reg_p): Likewise.
12839 * reload.c (find_equiv_reg): Likewise.
12840 * reload1.c (emit_reload_insns): Likewise.
12841 * sched-deps.c (deps_analyze_insn): Likewise.
12842 * sel-sched.c (init_regs_for_mode): Likewise.
12843 (mark_unavailable_hard_regs): Likewise.
12844 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
12845 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12846 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
12848 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12849 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12850 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
12852 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
12854 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12855 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12856 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
12858 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12859 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12860 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
12862 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12863 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
12865 * config/powerpcspe/powerpcspe.c
12866 (rs6000_hard_regno_call_part_clobbered): New function.
12867 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12868 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12869 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
12871 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12872 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12873 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
12875 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
12876 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12877 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
12879 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12880 Alan Hayward <alan.hayward@arm.com>
12881 David Sherwood <david.sherwood@arm.com>
12883 * rtl.h (subreg_memory_offset): Declare.
12884 * emit-rtl.c (subreg_memory_offset): New function.
12885 * expmed.c (store_bit_field_1): Use it.
12886 * expr.c (undefined_operand_subword_p): Likewise.
12887 * simplify-rtx.c (simplify_subreg): Likewise.
12889 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
12891 PR rtl-optimization/57448
12894 * optabs.c (expand_atomic_load): Place compiler memory barriers if
12895 using atomic_load pattern.
12896 (expand_atomic_store): Likewise.
12898 2017-09-04 Jakub Jelinek <jakub@redhat.com>
12901 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
12902 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
12903 handling. Use replace_call_with_value with NULL instead of
12904 gsi_replace, unlink_stmt_vdef and release_defs.
12906 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
12909 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
12911 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12914 * rtl.h (emit_library_call_value_1): Declare.
12915 (emit_library_call): Replace declaration with a series of overloads.
12916 Remove the parameter count argument.
12917 (emit_library_call_value): Likewise.
12918 * calls.c (emit_library_call_value_1): Make global. Replace varargs
12919 with an "rtx_mode_t *".
12920 (emit_library_call_value): Delete.
12921 (emit_library_call): Likewise.
12922 * asan.c (asan_emit_stack_protection): Update calls accordingly.
12923 (asan_emit_allocas_unpoison): Likewise.
12924 * builtins.c (expand_builtin_powi): Likewise.
12925 (expand_asan_emit_allocas_unpoison): Likewise.
12926 * cfgexpand.c (expand_main_function): Likewise.
12927 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
12928 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
12929 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
12930 * config/arm/arm.c (arm_trampoline_init): Likewise.
12931 (arm_call_tls_get_addr): Likewise.
12932 (arm_expand_divmod_libfunc): Likewise.
12933 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
12934 (smulsi3_highpart): Likewise.
12935 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
12936 (c6x_expand_compare): Likewise.
12937 (c6x_expand_movmem): Likewise.
12938 * config/frv/frv.c (frv_trampoline_init): Likewise.
12939 * config/i386/i386.c (ix86_trampoline_init): Likewise.
12940 (ix86_expand_divmod_libfunc): Likewise.
12941 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
12942 (ia64_expand_compare): Likewise.
12943 (ia64_profile_hook): Likewise.
12944 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
12945 (nonlocal_goto): Likewise.
12946 (restore_stack_nonlocal): Likewise.
12947 * config/m32r/m32r.c (block_move_call): Likewise.
12948 (m32r_trampoline_init): Likewise.
12949 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
12950 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
12951 (m68k_call_m68k_read_tp): Likewise.
12952 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
12953 (microblaze_expand_divide): Likewise.
12954 * config/mips/mips.h (mips_args): Likewise.
12955 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
12956 (MIPS_ICACHE_SYNC): Likewise.
12957 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
12958 (nios2_trampoline_init): Likewise.
12959 * config/pa/pa.c (hppa_tls_call): Likewise.
12960 (pa_trampoline_init): Likewise.
12961 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
12962 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
12963 (expand_strn_compare): Likewise.
12964 (rs6000_generate_compare): Likewise.
12965 (rs6000_expand_float128_convert): Likewise.
12966 (output_profile_hook): Likewise.
12967 (rs6000_trampoline_init): Likewise.
12968 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
12969 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
12970 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
12971 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
12972 (rs6000_generate_compare): Likewise.
12973 (rs6000_expand_float128_convert): Likewise.
12974 (output_profile_hook): Likewise.
12975 (rs6000_trampoline_init): Likewise.
12976 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
12977 * config/sh/sh.c (sh_trampoline_init): Likewise.
12978 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
12979 (sparc_emit_float_lib_cmp): Likewise.
12980 (sparc32_initialize_trampoline): Likewise.
12981 (sparc64_initialize_trampoline): Likewise.
12982 (sparc_profile_hook): Likewise.
12983 * config/spu/spu.c (ea_load_store): Likewise.
12984 * config/spu/spu.md (floatunssidf2): Likewise.
12985 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
12986 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
12987 * config/visium/visium.c (expand_block_move_4): Likewise.
12988 (expand_block_move_2): Likewise.
12989 (expand_block_move_1): Likewise.
12990 (expand_block_set_4): Likewise.
12991 (expand_block_set_2): Likewise.
12992 (expand_block_set_1): Likewise.
12993 (visium_trampoline_init): Likewise.
12994 (visium_profile_hook): Likewise.
12995 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
12996 (xtensa_setup_frame_addresses): Likewise.
12997 (xtensa_trampoline_init): Likewise.
12998 * except.c (sjlj_emit_function_enter): Likewise.
12999 (sjlj_emit_function_exit): Likewise.
13000 * explow.c (allocate_dynamic_stack_space): Likewise.
13001 (probe_stack_range): Likewise.
13002 * expr.c (convert_mode_scalar): Likewise.
13003 * optabs.c (expand_binop): Likewise.
13004 (expand_twoval_binop_libfunc): Likewise.
13005 (expand_unop): Likewise.
13006 (prepare_cmp_insn): Likewise.
13007 (prepare_float_lib_cmp): Likewise.
13008 (expand_float): Likewise.
13009 (expand_fix): Likewise.
13010 (expand_fixed_convert): Likewise.
13011 (maybe_emit_sync_lock_test_and_set): Likewise.
13012 (expand_atomic_compare_and_swap): Likewise.
13013 (expand_mem_thread_fence): Likewise.
13014 (expand_atomic_fetch_op): Likewise.
13016 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
13018 * doc/generic.texi (OpenACC): Adjust URL.
13019 * doc/invoke.texi (C Dialect Options): Ditto.
13021 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
13023 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
13024 predicate for operand 1. Add (m,<S>) constraint.
13025 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
13026 Prevent memory operand 1 with register operand 2.
13028 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
13030 PR rtl-optimization/82024
13031 * combine.c (try_combine): If the combination result is a PARALLEL,
13032 and we only need to retain the SET in there that would be placed
13033 at I2, check that we can place that at I3 instead, before doing so.
13035 2017-09-01 Jakub Jelinek <jakub@redhat.com>
13038 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
13040 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
13041 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
13044 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
13045 Jeff Law <law@redhat.com>
13047 * varasm.c (bss_initializer_p): Do not put constants into .bss
13048 (categorize_decl_for_section): Handle bss_initializer_p returning
13049 false when DECL_INITIAL is NULL.
13051 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13054 * config/s390/s390.c (s390_can_inline_p): New function.
13056 2017-09-01 Jeff Law <law@redhat.com>
13058 PR tree-optimization/82052
13059 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
13060 Always initialize the returned slot after a hash table miss
13061 when INSERT is true.
13063 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
13065 * config/s390/s390.md (mem_signal_fence): Remove.
13066 * doc/md.texi (mem_signal_fence): Remove.
13067 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
13069 * target-insns.def (mem_signal_fence): Remove.
13071 2017-09-01 Jakub Jelinek <jakub@redhat.com>
13074 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
13077 * asan.c (create_odr_indicator): Strip name encoding from assembler
13078 name before appending it after __odr_asan_.
13080 2017-09-01 Martin Liska <mliska@suse.cz>
13082 PR tree-optimization/82059
13083 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
13084 frequency only when an edge is redirected.
13086 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13088 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
13089 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
13090 (arc_conditional_register_usage): Remove ARC600 lp_count
13092 (arc_file_start): Emit Tag_ARC_CPU_variation.
13093 (arc_can_use_doloop_p): New conditions to use ZOLs.
13094 (hwloop_fail): New function.
13095 (hwloop_optimize): Likewise.
13096 (hwloop_pattern_reg): Likewise.
13097 (arc_doloop_hooks): New struct, to be used with reorg_loops.
13098 (arc_reorg_loops): New function, calls reorg_loops.
13099 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
13100 (arc600_corereg_hazard): Remove ZOL checking, case handled by
13102 (arc_loop_hazard): Remove function, functionality moved into
13104 (arc_hazard): Remove arc_loop_hazard call.
13105 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
13106 into hwloop_optimize.
13107 (arc_label_align): Remove ZOL handling.
13108 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
13109 * config/arc/arc.md (doloop_begin): Remove pattern.
13110 (doloop_begin_i): Likewise.
13111 (doloop_end_i): Likewise.
13112 (doloop_fallback): Likewise.
13113 (doloop_fallback_m): Likewise.
13114 (doloop_end): Reimplement expand.
13115 (arc_lp): New pattern for LP instruction.
13116 (loop_end): New pattern.
13117 (loop_fail): Likewise.
13118 (decrement_and_branch_until_zero): Likewise.
13119 * config/arc/arc.opt (mlpc-width): New option.
13120 * doc/invoke.texi (mlpc-width): Document option.
13122 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13124 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
13125 (arc_ccfsm_advance): Fix checking for delay slots.
13126 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
13128 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13130 * config/arc/arc.md (movqi_insn): Add stores to save constant long
13132 (movhi_insn): Update store instruction constraint which are saving
13133 6-bit short immediates.
13134 (movsi_insn): Consider also short scaled load operations.
13135 (zero_extendhisi2_i): Use Usd constraint instead of T.
13136 (extendhisi2_i): Add q constraint.
13137 (arc_clzsi2): Add type and length attributes.
13138 (arc_ctzsi2): Likewise.
13139 * config/arc/constraints.md (Usc): Update constraint, the
13140 assembler can parse two relocations for a single instruction.
13142 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13144 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
13145 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
13147 2017-08-31 Olivier Hainque <hainque@adacore.com>
13149 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
13150 match as powerpc-wrs-vxworks*.
13152 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
13154 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
13155 register constraint for by-element operand.
13156 (aarch64_mls_elt_merge<mode>): Likewise.
13158 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13160 * config/arc/arc.c (arc_can_follow_jump): Check for short
13163 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13165 * config.gcc: Use g.opt for arc.
13166 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
13167 functionality moved to ...
13168 (legitimate_scaled_address_p): New function, ...here.
13169 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
13170 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
13171 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
13173 (arc_override_options): Handle G option.
13174 (arc_output_pic_addr_const): Correct function definition.
13175 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
13176 (arc_decl_anon_ns_mem_p): Delete.
13177 (arc_in_small_data_p): Overhaul this function to take into
13178 consideration the value given via G option.
13179 (arc_rewrite_small_data_1): Renamed and corrected old
13180 arc_rewrite_small_data function.
13181 (arc_rewrite_small_data): New function.
13182 (small_data_pattern): Don't use pic_offset_table_rtx.
13183 * config/arc/arc.h (CC1_SPEC): Recognize G option.
13184 * config/arc/simdext.md (movmisalignv2hi): Use
13185 prepare_move_operands function.
13187 (movmisalign*): Likewise.
13188 * doc/invoke.texi (ARC options): Document -G option.
13190 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13192 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
13194 * config/arc/arc.c (arc_print_operand): Output scalled address for
13195 sdata whenever is possible.
13196 (arc_in_small_data_p): Allow sdata for 64bit datum when double
13197 load/stores are available.
13198 (compact_sda_memory_operand): Check for the alignment required by
13199 code density instructions.
13200 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
13202 * config/arc/constraints.md (Usd): Update constraint.
13203 (Us0): New constraint.
13204 (Usc): Update constraint.
13206 2017-08-31 Richard Biener <rguenther@suse.de>
13208 PR middle-end/82054
13209 * dwarf2out.c (dwarf2out_early_global_decl): Process each
13210 function only once.
13212 2017-08-31 Tamar Christina <tamar.christina@arm.com>
13214 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
13215 Resize type_signature.
13217 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13218 Alan Hayward <alan.hayward@arm.com>
13219 David Sherwood <david.sherwood@arm.com>
13221 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
13222 subregs whose inner modes can be stored in GPRs.
13223 (aarch64_classify_index): Likewise.
13225 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13226 Alan Hayward <alan.hayward@arm.com>
13227 David Sherwood <david.sherwood@arm.com>
13229 * config/aarch64/iterators.md (V_cmp_result): Rename to...
13230 (V_INT_EQUIV): ...this.
13231 (v_cmp_result): Rename to...
13232 (v_int_equiv): ...this.
13233 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
13234 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
13235 (copysign<mode>3): Likewise.
13236 (aarch64_simd_bsl<mode>_internal): Likewise.
13237 (aarch64_simd_bsl<mode>): Likewise.
13238 (vec_cmp<mode><mode>): Likewise.
13239 (vcond<mode><mode>): Likewise.
13240 (vcond<v_cmp_mixed><mode>): Likewise.
13241 (vcondu<mode><v_cmp_mixed>): Likewise.
13242 (aarch64_cm<optab><mode>): Likewise.
13243 (aarch64_cmtst<mode>): Likewise.
13244 (aarch64_fac<optab><mode>): Likewise.
13245 (vec_perm_const<mode>): Likewise.
13246 (vcond_mask_<mode><v_cmp_result>): Rename to...
13247 (vcond_mask_<mode><v_int_equiv>): ...this.
13248 (vec_cmp<mode><v_cmp_result>): Rename to...
13249 (vec_cmp<mode><v_int_equiv>): ...this.
13251 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13252 Alan Hayward <alan.hayward@arm.com>
13253 David Sherwood <david.sherwood@arm.com>
13255 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
13257 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
13258 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
13259 (UNSPEC_LD4_DREG): New unspecs.
13260 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
13261 (aarch64_ld2<mode>_dreg_be): Replace with...
13262 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
13264 (aarch64_ld3<mode>_dreg_le)
13265 (aarch64_ld3<mode>_dreg_be): Replace with...
13266 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
13268 (aarch64_ld4<mode>_dreg_le)
13269 (aarch64_ld4<mode>_dreg_be): Replace with...
13270 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
13273 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13275 PR tree-optimization/81987
13276 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
13277 insert an initializer in a location not dominated by the stride
13280 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
13282 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
13283 on the entire header of the finally block in the fallthru case.
13285 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
13287 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
13289 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
13291 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
13292 rs6000_emit_move_from_cr and call renamed function.
13293 (rs6000_emit_prologue): Call renamed functions.
13294 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
13295 movesi_from_cr, remove volatile CRs.
13297 2017-08-30 Jon Beniston <jon@beniston.com>
13298 Richard Biener <rguenther@suse.de>
13300 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
13301 of VECTOR_MODE_P check.
13302 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
13303 element vector types.
13305 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13307 * df.h (df_read_modify_subreg_p): Remove in favor of...
13308 * rtl.h (read_modify_subreg_p): ...this new function. Take a
13309 const_rtx instead of an rtx.
13310 * cprop.c (local_cprop_find_used_regs): Update accordingly.
13311 * df-problems.c (df_word_lr_mark_ref): Likewise.
13312 * ira-lives.c (mark_pseudo_reg_live): Likewise.
13313 (mark_pseudo_reg_dead): Likewise.
13314 (mark_ref_dead): Likewise.
13315 * reginfo.c (init_subregs_of_mode): Likewise.
13316 * sched-deps.c (sched_analyze_1): Likewise.
13317 * df-scan.c (df_def_record_1): Likewise.
13318 (df_uses_record): Likewise.
13319 (df_read_modify_subreg_p): Remove in favor of...
13320 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
13321 const_rtx instead of an rtx.
13323 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13324 Alan Hayward <alan.hayward@arm.com>
13325 David Sherwood <david.sherwood@arm.com>
13327 * rtl.h (partial_subreg_p): New function.
13328 * caller-save.c (save_call_clobbered_regs): Use it.
13329 * calls.c (expand_call): Likewise.
13330 * combine.c (combinable_i3pat): Likewise.
13331 (simplify_set): Likewise.
13332 (make_extraction): Likewise.
13333 (make_compound_operation_int): Likewise.
13334 (gen_lowpart_or_truncate): Likewise.
13335 (force_to_mode): Likewise.
13336 (make_field_assignment): Likewise.
13337 (reg_truncated_to_mode): Likewise.
13338 (record_truncated_value): Likewise.
13339 (move_deaths): Likewise.
13340 * cse.c (record_jump_cond): Likewise.
13341 (cse_insn): Likewise.
13342 * cselib.c (cselib_lookup_1): Likewise.
13343 * expmed.c (extract_bit_field_using_extv): Likewise.
13344 * function.c (assign_parm_setup_reg): Likewise.
13345 * ifcvt.c (noce_convert_multiple_sets): Likewise.
13346 * ira-build.c (create_insn_allocnos): Likewise.
13347 * lra-coalesce.c (merge_pseudos): Likewise.
13348 * lra-constraints.c (match_reload): Likewise.
13349 (simplify_operand_subreg): Likewise.
13350 (curr_insn_transform): Likewise.
13351 * lra-lives.c (process_bb_lives): Likewise.
13352 * lra.c (new_insn_reg): Likewise.
13353 (lra_substitute_pseudo): Likewise.
13354 * regcprop.c (mode_change_ok): Likewise.
13355 (maybe_mode_change): Likewise.
13356 (copyprop_hardreg_forward_1): Likewise.
13357 * reload.c (push_reload): Likewise.
13358 (find_reloads): Likewise.
13359 (find_reloads_subreg_address): Likewise.
13360 * reload1.c (alter_reg): Likewise.
13361 (eliminate_regs_1): Likewise.
13362 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13364 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
13366 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
13367 back to if statements, including unpack.
13369 2017-08-30 Martin Liska <mliska@suse.cz>
13371 PR inline-asm/82001
13372 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
13374 (func_checker::compare_asm_inputs_outputs): ... this function.
13375 (func_checker::compare_gimple_asm): Use the function to compare
13376 also ASM constrains.
13377 * ipa-icf-gimple.h: Rename the function.
13379 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13380 Alan Hayward <alan.hayward@arm.com>
13381 David Sherwood <david.sherwood@arm.com>
13383 * coretypes.h (complex_mode): New type.
13384 * gdbhooks.py (build_pretty_printer): Handle it.
13385 * machmode.h (complex_mode): New class.
13386 (complex_mode::includes_p): New function.
13387 (is_complex_int_mode): Likewise.
13388 (is_complex_float_mode): Likewise.
13389 * genmodes.c (get_mode_class): Handle complex mode classes.
13390 * function.c (expand_function_end): Use is_complex_int_mode.
13392 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13393 Alan Hayward <alan.hayward@arm.com>
13394 David Sherwood <david.sherwood@arm.com>
13396 * coretypes.h (scalar_mode_pod): New typedef.
13397 * gdbhooks.py (build_pretty_printer): Handle it.
13398 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
13399 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
13400 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
13401 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
13402 as_a <scalar_mode>.
13404 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13405 Alan Hayward <alan.hayward@arm.com>
13406 David Sherwood <david.sherwood@arm.com>
13408 * machmode.h (mode_for_vector): Take a scalar_mode instead
13410 * stor-layout.c (mode_for_vector): Likewise.
13411 * explow.c (promote_mode): Use as_a <scalar_mode>.
13412 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
13414 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13415 Alan Hayward <alan.hayward@arm.com>
13416 David Sherwood <david.sherwood@arm.com>
13418 * target.def (preferred_simd_mode): Take a scalar_mode
13419 instead of a machine_mode.
13420 * targhooks.h (default_preferred_simd_mode): Likewise.
13421 * targhooks.c (default_preferred_simd_mode): Likewise.
13422 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
13423 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
13424 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
13425 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
13426 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
13427 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
13428 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
13429 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
13431 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
13432 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
13433 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
13434 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
13435 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
13436 * doc/tm.texi: Regenerate.
13437 * optabs-query.c (can_vec_mask_load_store_p): Return false for
13440 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13441 Alan Hayward <alan.hayward@arm.com>
13442 David Sherwood <david.sherwood@arm.com>
13444 * target.def (scalar_mode_supported_p): Take a scalar_mode
13445 instead of a machine_mode.
13446 * targhooks.h (default_scalar_mode_supported_p): Likewise.
13447 * targhooks.c (default_scalar_mode_supported_p): Likewise.
13448 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
13449 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13450 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
13451 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
13452 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
13453 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
13454 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
13455 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
13456 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
13457 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
13458 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
13459 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
13461 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
13462 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
13463 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13464 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13465 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
13467 * doc/tm.texi: Regenerate.
13469 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13470 Alan Hayward <alan.hayward@arm.com>
13471 David Sherwood <david.sherwood@arm.com>
13473 * coretypes.h (opt_scalar_mode): New typedef.
13474 * gdbhooks.py (build_pretty_printers): Handle it.
13475 * machmode.h (mode_iterator::get_2xwider): Add overload for
13477 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
13479 * expr.c (convert_mode_scalar): Likewise.
13480 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13481 * optabs.c (expand_float): Likewise.
13482 (expand_fix): Likewise.
13483 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13485 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13486 Alan Hayward <alan.hayward@arm.com>
13487 David Sherwood <david.sherwood@arm.com>
13489 * optabs.c (expand_float): Explicitly check for scalars before
13490 using a branching expansion.
13491 (expand_fix): Likewise.
13493 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13494 Alan Hayward <alan.hayward@arm.com>
13495 David Sherwood <david.sherwood@arm.com>
13497 * expr.c (convert_mode): Split scalar handling out into...
13498 (convert_mode_scalar): ...this new function. Treat the modes
13501 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13502 Alan Hayward <alan.hayward@arm.com>
13503 David Sherwood <david.sherwood@arm.com>
13505 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
13507 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
13509 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13510 Alan Hayward <alan.hayward@arm.com>
13511 David Sherwood <david.sherwood@arm.com>
13513 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
13514 rather than a machine_mode.
13515 (fixed_from_string): Likewise.
13516 (fixed_convert): Likewise.
13517 (fixed_convert_from_int): Likewise.
13518 (fixed_convert_from_real): Likewise.
13519 (real_convert_from_fixed): Likewise.
13520 * fixed-value.c (fixed_from_double_int): Likewise.
13521 (fixed_from_string): Likewise.
13522 (fixed_convert): Likewise.
13523 (fixed_convert_from_int): Likewise.
13524 (fixed_convert_from_real): Likewise.
13525 (real_convert_from_fixed): Likewise.
13526 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
13528 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13529 Alan Hayward <alan.hayward@arm.com>
13530 David Sherwood <david.sherwood@arm.com>
13532 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
13533 of separate mode class checks. Do not allow vector modes here.
13534 (immed_wide_int_const): Use as_a <scalar_mode>.
13535 * explow.c (trunc_int_for_mode): Likewise.
13536 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
13537 (wi::shwi): Likewise.
13538 (wi::min_value): Likewise.
13539 (wi::max_value): Likewise.
13540 * dwarf2out.c (loc_descriptor): Likewise.
13541 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
13542 for CONST_WIDE_INT.
13544 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13545 Alan Hayward <alan.hayward@arm.com>
13546 David Sherwood <david.sherwood@arm.com>
13548 * tree.h (SCALAR_TYPE_MODE): New macro.
13549 * expr.c (expand_expr_addr_expr_1): Use it.
13550 (expand_expr_real_2): Likewise.
13551 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
13552 (fold_convert_const_fixed_from_int): Likewise.
13553 (fold_convert_const_fixed_from_real): Likewise.
13554 (native_encode_fixed): Likewise
13555 (native_encode_complex): Likewise
13556 (native_encode_vector): Likewise.
13557 (native_interpret_fixed): Likewise.
13558 (native_interpret_real): Likewise.
13559 (native_interpret_complex): Likewise.
13560 (native_interpret_vector): Likewise.
13561 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
13562 (simd_clone_adjust_argument_types): Likewise.
13563 (simd_clone_init_simd_arrays): Likewise.
13564 (simd_clone_adjust): Likewise.
13565 * stor-layout.c (layout_type): Likewise.
13566 * tree.c (build_minus_one_cst): Likewise.
13567 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
13568 * tree-inline.c (estimate_move_cost): Likewise.
13569 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
13570 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
13571 (vectorizable_reduction): Likewise.
13572 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
13573 (vect_recog_mixed_size_cond_pattern): Likewise.
13574 (check_bool_pattern): Likewise.
13575 (adjust_bool_pattern): Likewise.
13576 (search_type_for_mask_1): Likewise.
13577 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
13578 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13579 (vectorizable_load): Likewise.
13580 (vectorizable_store): Likewise.
13581 * ubsan.c (ubsan_encode_value): Likewise.
13582 * varasm.c (output_constant): Likewise.
13584 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13585 Alan Hayward <alan.hayward@arm.com>
13586 David Sherwood <david.sherwood@arm.com>
13588 * coretypes.h (scalar_mode): New class.
13589 * machmode.h (scalar_mode): Likewise.
13590 (scalar_mode::includes_p): New function.
13591 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
13592 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
13593 * genmodes.c (get_mode_class): Handle remaining scalar modes.
13594 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
13595 * expmed.c (store_bit_field_1): Likewise.
13596 (extract_bit_field_1): Likewise.
13597 * expr.c (write_complex_part): Likewise.
13598 (read_complex_part): Likewise.
13599 (emit_move_complex_push): Likewise.
13600 (expand_expr_real_2): Likewise.
13601 * function.c (assign_parm_setup_reg): Likewise.
13602 (assign_parms_unsplit_complex): Likewise.
13603 * optabs.c (expand_binop): Likewise.
13604 * rtlanal.c (subreg_get_info): Likewise.
13605 * simplify-rtx.c (simplify_immed_subreg): Likewise.
13606 * varasm.c (output_constant_pool_2): Likewise.
13608 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13609 Alan Hayward <alan.hayward@arm.com>
13610 David Sherwood <david.sherwood@arm.com>
13612 * expmed.c (extract_high_half): Use scalar_int_mode and remove
13614 (expmed_mult_highpart_optab): Likewise.
13615 (expmed_mult_highpart): Likewise.
13617 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13618 Alan Hayward <alan.hayward@arm.com>
13619 David Sherwood <david.sherwood@arm.com>
13621 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
13622 instead of a machine_mode.
13623 (builtin_memset_read_str): Likewise.
13624 * builtins.c (c_readstr): Likewise.
13625 (builtin_memcpy_read_str): Likewise.
13626 (builtin_strncpy_read_str): Likewise.
13627 (builtin_memset_read_str): Likewise.
13628 (builtin_memset_gen_str): Likewise.
13629 (expand_builtin_signbit): Use scalar_int_mode for local variables.
13630 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
13631 instead of a machine_mode.
13632 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
13634 (make_extraction): Likewise.
13635 (try_widen_shift_mode): Take and return scalar_int_modes instead
13637 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
13638 a scalar_int_mode instead of a machine_mode.
13639 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
13640 (avr_addr_space_pointer_mode): Likewise.
13641 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
13642 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
13643 (msp430_unwind_word_mode): Likewise.
13644 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
13645 (spu_addr_space_pointer_mode): Likewise.
13646 (spu_addr_space_address_mode): Likewise.
13647 (spu_libgcc_cmp_return_mode): Likewise.
13648 (spu_libgcc_shift_count_mode): Likewise.
13649 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
13650 (rl78_addr_space_pointer_mode): Likewise.
13651 (fl78_unwind_word_mode): Likewise.
13652 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
13654 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
13655 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
13656 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
13657 (mips_valid_pointer_mode): Likewise.
13658 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
13659 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
13660 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
13662 (ft32_addr_space_address_mode): Likewise.
13663 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
13664 scalar_int_mode instead of a machine_mode.
13665 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
13667 (m32c_addr_space_address_mode): Likewise.
13668 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
13669 (rs6000_eh_return_filter_mode): Likewise.
13670 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
13671 (rs6000_eh_return_filter_mode): Likewise.
13672 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
13673 (s390_libgcc_shift_count_mode): Likewise.
13674 (s390_unwind_word_mode): Likewise.
13675 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
13677 * target.def (mode_rep_extended): Likewise.
13678 (valid_pointer_mode): Likewise.
13679 (addr_space.valid_pointer_mode): Likewise.
13680 (eh_return_filter_mode): Return a scalar_int_mode rather than
13682 (libgcc_cmp_return_mode): Likewise.
13683 (libgcc_shift_count_mode): Likewise.
13684 (unwind_word_mode): Likewise.
13685 (addr_space.pointer_mode): Likewise.
13686 (addr_space.address_mode): Likewise.
13687 * doc/tm.texi: Regenerate.
13688 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
13690 (do_jump): Use scalar_int_mode for local variables.
13691 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
13692 rather than a machine_mode.
13693 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
13694 (scompare_loc_descriptor_wide): Likewise.
13695 (scompare_loc_descriptor_narrow): Likewise.
13696 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
13698 * except.c (sjlj_emit_dispatch_table): Likewise.
13699 (expand_builtin_eh_copy_values): Likewise.
13700 * explow.c (convert_memory_address_addr_space_1): Likewise.
13701 Take a scalar_int_mode rather than a machine_mode.
13702 (convert_memory_address_addr_space): Take a scalar_int_mode rather
13703 than a machine_mode.
13704 (memory_address_addr_space): Use scalar_int_mode for local variables.
13705 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
13706 rather than a machine_mode.
13707 * expmed.c (mask_rtx): Likewise.
13708 (init_expmed_one_conv): Likewise.
13709 (expand_mult_highpart_adjust): Likewise.
13710 (extract_high_half): Likewise.
13711 (expmed_mult_highpart_optab): Likewise.
13712 (expmed_mult_highpart): Likewise.
13713 (expand_smod_pow2): Likewise.
13714 (expand_sdiv_pow2): Likewise.
13715 (emit_store_flag_int): Likewise.
13716 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
13718 (extract_low_bits): Likewise.
13719 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
13721 * expr.c (pieces_addr::adjust): Likewise.
13722 (can_store_by_pieces): Likewise.
13723 (store_by_pieces): Likewise.
13724 (clear_by_pieces_1): Likewise.
13725 (expand_expr_addr_expr_1): Likewise.
13726 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
13727 (expand_expr_real_1): Likewise.
13728 (try_casesi): Likewise.
13729 * final.c (shorten_branches): Likewise.
13730 * fold-const.c (fold_convert_const_int_from_fixed): Change the
13731 type of "mode" to machine_mode.
13732 * internal-fn.c (expand_arith_overflow_result_store): Take a
13733 scalar_int_mode rather than a machine_mode.
13734 (expand_mul_overflow): Use scalar_int_mode for local variables.
13735 * loop-doloop.c (doloop_modify): Likewise.
13736 (doloop_optimize): Likewise.
13737 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
13738 than a machine_mode.
13739 (expand_doubleword_shift_condmove): Likewise.
13740 (expand_doubleword_shift): Likewise.
13741 (expand_doubleword_clz): Likewise.
13742 (expand_doubleword_popcount): Likewise.
13743 (expand_doubleword_parity): Likewise.
13744 (expand_absneg_bit): Use scalar_int_mode for local variables.
13745 (prepare_float_lib_cmp): Likewise.
13746 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
13747 rather than a machine_mode.
13748 (convert_memory_address_addr_space): Likewise.
13749 (get_mode_bounds): Likewise.
13750 (get_address_mode): Return a scalar_int_mode rather than a
13752 * rtlanal.c (get_address_mode): Likewise.
13753 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
13754 than a machine_mode.
13755 * targhooks.c (default_mode_rep_extended): Likewise.
13756 (default_valid_pointer_mode): Likewise.
13757 (default_addr_space_valid_pointer_mode): Likewise.
13758 (default_eh_return_filter_mode): Return a scalar_int_mode rather
13759 than a machine_mode.
13760 (default_libgcc_cmp_return_mode): Likewise.
13761 (default_libgcc_shift_count_mode): Likewise.
13762 (default_unwind_word_mode): Likewise.
13763 (default_addr_space_pointer_mode): Likewise.
13764 (default_addr_space_address_mode): Likewise.
13765 * targhooks.h (default_eh_return_filter_mode): Likewise.
13766 (default_libgcc_cmp_return_mode): Likewise.
13767 (default_libgcc_shift_count_mode): Likewise.
13768 (default_unwind_word_mode): Likewise.
13769 (default_addr_space_pointer_mode): Likewise.
13770 (default_addr_space_address_mode): Likewise.
13771 (default_mode_rep_extended): Take a scalar_int_mode rather than
13773 (default_valid_pointer_mode): Likewise.
13774 (default_addr_space_valid_pointer_mode): Likewise.
13775 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
13777 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
13778 rather than a machine_mode.
13779 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
13780 for local variables.
13781 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13782 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
13783 than a machine_mode.
13785 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13786 Alan Hayward <alan.hayward@arm.com>
13787 David Sherwood <david.sherwood@arm.com>
13789 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
13790 the mode argument to scalar_int_mode.
13791 (do_jump_by_parts_zero_rtx): Likewise.
13792 (do_jump_by_parts_equality_rtx): Likewise.
13793 (do_jump_by_parts_greater): Take a mode argument.
13794 (do_jump_by_parts_equality): Likewise.
13795 (do_jump_1): Update calls accordingly.
13797 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13798 Alan Hayward <alan.hayward@arm.com>
13799 David Sherwood <david.sherwood@arm.com>
13801 * is-a.h (safe_dyn_cast): New function.
13802 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
13803 (jump_table_for_label): Likewise.
13804 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
13805 instead of an rtx_insn *.
13806 (shorten_branches): Use dyn_cast instead of LABEL_P and
13807 JUMP_TABLE_DATA_P. Use jump_table_for_label and
13808 rtx_jump_table_data::get_data_mode.
13809 (final_scan_insn): Likewise.
13811 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13812 Alan Hayward <alan.hayward@arm.com>
13813 David Sherwood <david.sherwood@arm.com>
13815 * combine.c (try_combine): Use is_a <scalar_int_mode> when
13816 trying to combine a full-register integer set with a subreg
13819 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13820 Alan Hayward <alan.hayward@arm.com>
13821 David Sherwood <david.sherwood@arm.com>
13823 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
13824 that is always either address_mode or pointer_mode.
13826 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13827 Alan Hayward <alan.hayward@arm.com>
13828 David Sherwood <david.sherwood@arm.com>
13830 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
13831 when the two are known to be equal.
13833 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13834 Alan Hayward <alan.hayward@arm.com>
13835 David Sherwood <david.sherwood@arm.com>
13837 * simplify-rtx.c (simplify_const_unary_operation): Use
13838 is_a <scalar_int_mode> instead of checking for a nonzero
13839 precision. Forcibly convert op_mode to a scalar_int_mode
13840 in that case. More clearly differentiate the operand and
13841 result modes and use the former when deciding what the value
13842 of a count-bits operation should be. Use is_int_mode instead
13843 of checking for a MODE_INT. Remove redundant check for whether
13844 this mode has a zero precision.
13846 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13847 Alan Hayward <alan.hayward@arm.com>
13848 David Sherwood <david.sherwood@arm.com>
13850 * optabs.c (widen_leading): Change the type of the mode argument
13851 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
13852 (widen_bswap): Likewise.
13853 (expand_parity): Likewise.
13854 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
13855 (expand_ffs): Likewise.
13856 (epand_unop): Check for scalar integer modes before calling the
13859 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13860 Alan Hayward <alan.hayward@arm.com>
13861 David Sherwood <david.sherwood@arm.com>
13863 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
13865 (expand_expr_real_1): Update call accordingly.
13867 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13868 Alan Hayward <alan.hayward@arm.com>
13869 David Sherwood <david.sherwood@arm.com>
13871 * expmed.c (store_bit_field_using_insv): Add op0_mode and
13872 value_mode arguments. Use scalar_int_mode internally.
13873 (store_bit_field_1): Rename the new integer mode from imode
13874 to op0_mode and use it instead of GET_MODE (op0). Update calls
13875 to store_split_bit_field, store_bit_field_using_insv and
13876 store_fixed_bit_field.
13877 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
13878 Use scalar_int_mode internally. Use a bit count rather than a mode
13879 when calculating the largest bit size for get_best_mode.
13880 Update calls to store_split_bit_field and store_fixed_bit_field_1.
13881 (store_fixed_bit_field_1): Add mode and value_mode arguments.
13882 Remove assertion that OP0 has a scalar integer mode.
13883 (store_split_bit_field): Add op0_mode and value_mode arguments.
13884 Update calls to extract_fixed_bit_field.
13885 (extract_bit_field_using_extv): Add an op0_mode argument.
13886 Use scalar_int_mode internally.
13887 (extract_bit_field_1): Rename the new integer mode from imode to
13888 op0_mode and use it instead of GET_MODE (op0). Update calls to
13889 extract_split_bit_field, extract_bit_field_using_extv and
13890 extract_fixed_bit_field.
13891 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
13892 to extract_split_bit_field and extract_fixed_bit_field_1.
13893 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
13894 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
13895 on the target mode.
13896 (extract_split_bit_field): Add an op0_mode argument. Update call
13897 to extract_fixed_bit_field.
13899 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13900 Alan Hayward <alan.hayward@arm.com>
13901 David Sherwood <david.sherwood@arm.com>
13903 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
13904 * explow.c (hard_function_value): Likewise.
13905 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
13906 convert_to_mode call outside the loop.
13907 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
13908 for the mode iterator. Require the mode specified by max_pieces
13910 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
13912 (copy_blkmode_to_reg): Likewise.
13913 (set_storage_via_setmem): Likewise.
13914 * optabs.c (prepare_cmp_insn): Likewise.
13915 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13916 * stor-layout.c (finish_bitfield_representative): Likewise.
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 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
13923 * expr.c (convert_move): Use them.
13924 (convert_modes): Likewise.
13925 (store_expr_with_bounds): Likewise.
13927 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13928 Alan Hayward <alan.hayward@arm.com>
13929 David Sherwood <david.sherwood@arm.com>
13931 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
13932 parameter for the mode of "x". Remove the "known_x", "known_mode"
13933 and "known_ret" arguments. Change the type of the mode argument
13934 to scalar_int_mode.
13935 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
13936 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
13937 (reg_num_sign_bit_copies_for_combine): Likewise.
13938 * rtlanal.c (nonzero_bits1): Likewise.
13939 (num_sign_bit_copies1): Likewise.
13940 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
13941 (reg_num_sign_bit_copies_general): Likewise.
13942 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
13943 (reg_nonzero_bits_general): Likewise.
13945 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13946 Alan Hayward <alan.hayward@arm.com>
13947 David Sherwood <david.sherwood@arm.com>
13949 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
13950 than in subroutines. Return 1 for non-integer modes.
13951 (cached_num_sign_bit_copies): Change the type of the mode parameter
13952 to scalar_int_mode.
13953 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
13954 classes. Handle CONST_INT_P first and then check whether X also
13955 has a scalar integer mode. Check the same thing for inner registers
13956 of a SUBREG and for values that are being extended or truncated.
13958 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13959 Alan Hayward <alan.hayward@arm.com>
13960 David Sherwood <david.sherwood@arm.com>
13962 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
13963 in subroutines. Return the mode mask for non-integer modes.
13964 (cached_nonzero_bits): Change the type of the mode parameter
13965 to scalar_int_mode.
13966 (nonzero_bits1): Likewise. Remove early exit for other mode
13967 classes. Handle CONST_INT_P first and then check whether X
13968 also has a scalar integer mode.
13970 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13971 Alan Hayward <alan.hayward@arm.com>
13972 David Sherwood <david.sherwood@arm.com>
13974 * expr.c (widest_int_mode_for_size): Make the comment match the code.
13975 Return a scalar_int_mode and assert that the size is greater than
13977 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
13978 (op_by_pieces_d::op_by_pieces_d): Likewise.
13979 (op_by_pieces_d::run): Likewise.
13980 (can_store_by_pieces): Likewise.
13982 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13983 Alan Hayward <alan.hayward@arm.com>
13984 David Sherwood <david.sherwood@arm.com>
13986 * combine.c (extract_left_shift): Add a mode argument and update
13988 (make_compound_operation_int): Change the type of the mode parameter
13989 to scalar_int_mode and update the call to extract_left_shift.
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 * combine.c (simplify_and_const_int): Change the type of the mode
13996 parameter to scalar_int_mode.
13997 (simplify_and_const_int_1): Likewise. Update recursive call.
13999 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14000 Alan Hayward <alan.hayward@arm.com>
14001 David Sherwood <david.sherwood@arm.com>
14003 * combine.c (simplify_compare_const): Check that the mode is a
14004 scalar_int_mode (rather than VOIDmode) before testing its
14006 (simplify_comparison): Move COMPARISON_P handling out of the
14007 loop and restrict the latter part of the loop to scalar_int_modes.
14008 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
14009 and when considering SUBREG_REGs. Use is_int_mode instead of
14010 checking GET_MODE_CLASS against MODE_INT.
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 * combine.c (try_widen_shift_mode): Move check for equal modes to...
14017 (simplify_shift_const_1): ...here. Use scalar_int_mode for
14018 shift_unit_mode and for modes involved in scalar shifts.
14020 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14021 Alan Hayward <alan.hayward@arm.com>
14022 David Sherwood <david.sherwood@arm.com>
14024 * combine.c (force_int_to_mode): New function, split out from...
14025 (force_to_mode): ...here. Keep xmode up-to-date and use it
14026 instead of GET_MODE (x).
14028 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14029 Alan Hayward <alan.hayward@arm.com>
14030 David Sherwood <david.sherwood@arm.com>
14032 * optabs-query.h (extraction_insn::struct_mode): Change type to
14033 opt_scalar_int_mode and update comment.
14034 (extraction_insn::field_mode): Change type to scalar_int_mode.
14035 (extraction_insn::pos_mode): Likewise.
14036 * combine.c (make_extraction): Update accordingly.
14037 * optabs-query.c (get_traditional_extraction_insn): Likewise.
14038 (get_optab_extraction_insn): Likewise.
14039 * recog.c (simplify_while_replacing): Likewise.
14040 * expmed.c (narrow_bit_field_mem): Change the type of the mode
14041 parameter to opt_scalar_int_mode.
14043 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14044 Alan Hayward <alan.hayward@arm.com>
14045 David Sherwood <david.sherwood@arm.com>
14047 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
14048 to a scalar_int_mode instead of a machine_mode.
14049 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
14050 (get_best_mode): Return a boolean and use a pointer argument to store
14051 the selected mode. Replace the limit mode parameter with a bit limit.
14052 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
14053 for the values returned by bit_field_mode_iterator::next_mode.
14054 (store_bit_field): Update call to get_best_mode.
14055 (store_fixed_bit_field): Likewise.
14056 (extract_fixed_bit_field): Likewise.
14057 * expr.c (optimize_bitfield_assignment_op): Likewise.
14058 * fold-const.c (optimize_bit_field_compare): Likewise.
14059 (fold_truth_andor_1): Likewise.
14060 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
14061 Update for new type of m_mode.
14062 (get_best_mode): As above.
14064 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14065 Alan Hayward <alan.hayward@arm.com>
14066 David Sherwood <david.sherwood@arm.com>
14068 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
14069 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
14070 (store_bit_field): Check is_a <scalar_int_mode> before calling
14071 strict_volatile_bitfield_p.
14072 (extract_bit_field): Likewise.
14074 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14075 Alan Hayward <alan.hayward@arm.com>
14076 David Sherwood <david.sherwood@arm.com>
14078 * target.def (cstore_mode): Return a scalar_int_mode.
14079 * doc/tm.texi: Regenerate.
14080 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
14081 * targhooks.h (default_cstore_mode): Likewise.
14082 * targhooks.c (default_cstore_mode): Likewise, using a forced
14084 * expmed.c (emit_cstore): Expect the target of the cstore to be
14087 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14088 Alan Hayward <alan.hayward@arm.com>
14089 David Sherwood <david.sherwood@arm.com>
14091 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
14093 (niter_desc): Likewise mode.
14094 (iv_analyze): Add a mode parameter.
14096 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
14097 and change its type to scalar_int_mode.
14098 * loop-iv.c: Update commentary at head of file.
14099 (iv_constant): Pass the mode paraeter before the rtx it describes
14100 and change its type to scalar_int_mode. Remove VOIDmode handling.
14101 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
14102 (iv_extend): Likewise.
14103 (shorten_into_mode): Likewise.
14104 (iv_add): Use scalar_int_mode.
14105 (iv_mult): Likewise.
14106 (iv_shift): Likewise.
14107 (canonicalize_iv_subregs): Likewise.
14108 (get_biv_step_1): Pass the outer_mode parameter before the rtx
14109 it describes and change its mode to scalar_int_mode. Also change
14110 the type of the returned inner_mode to scalar_int_mode.
14111 (get_biv_step): Likewise, turning outer_mode from a pointer
14112 into a direct parameter. Update call to get_biv_step_1.
14113 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
14114 iv_constant and get_biv_step.
14115 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
14116 and change its type to scalar_int_mode. Don't initialise iv->mode
14117 to VOIDmode and remove later checks for its still being VOIDmode.
14118 Update calls to iv_analyze_op and iv_analyze_expr. Check
14119 is_a <scalar_int_mode> when changing the mode under consideration.
14120 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
14121 Update call to iv_analyze_expr.
14122 (iv_analyze_op): Add a mode parameter. Reject subregs whose
14123 inner register is not also a scalar_int_mode. Update call to
14125 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
14126 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
14127 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
14128 separate mode class checks. Update calls to iv_analyze. Remove
14129 fix-up of VOIDmodes after iv_analyze_biv.
14130 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
14131 don't have a scalar_int_mode. Update call to biv_p.
14133 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14134 Alan Hayward <alan.hayward@arm.com>
14135 David Sherwood <david.sherwood@arm.com>
14137 * cfgexpand.c (convert_debug_memory_address): Use
14138 as_a <scalar_int_mode>.
14139 * combine.c (expand_compound_operation): Likewise.
14140 (make_extraction): Likewise.
14141 (change_zero_ext): Likewise.
14142 (simplify_comparison): Likewise.
14143 * cse.c (cse_insn): Likewise.
14144 * dwarf2out.c (minmax_loc_descriptor): Likewise.
14145 (mem_loc_descriptor): Likewise.
14146 (loc_descriptor): Likewise.
14147 * expmed.c (init_expmed_one_mode): Likewise.
14148 (synth_mult): Likewise.
14149 (emit_store_flag_1): Likewise.
14150 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
14151 of a comparison with size.
14152 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
14153 (reduce_to_bit_field_precision): Likewise.
14154 * function.c (expand_function_end): Likewise.
14155 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
14156 * loop-doloop.c (doloop_modify): Likewise.
14157 * optabs.c (expand_binop): Likewise.
14158 (expand_unop): Likewise.
14159 (expand_copysign_absneg): Likewise.
14160 (prepare_cmp_insn): Likewise.
14161 (maybe_legitimize_operand): Likewise.
14162 * recog.c (const_scalar_int_operand): Likewise.
14163 * rtlanal.c (get_address_mode): Likewise.
14164 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14165 (simplify_cond_clz_ctz): Likewise.
14166 * tree-nested.c (get_nl_goto_field): Likewise.
14167 * tree.c (build_vector_type_for_mode): Likewise.
14168 * var-tracking.c (use_narrower_mode): Likewise.
14170 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14171 Alan Hayward <alan.hayward@arm.com>
14172 David Sherwood <david.sherwood@arm.com>
14174 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
14175 * builtins.c (expand_builtin_signbit): Use it.
14176 * cfgexpand.c (expand_debug_expr): Likewise.
14177 * dojump.c (do_jump): Likewise.
14178 (do_compare_and_jump): Likewise.
14179 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
14180 * expmed.c (make_tree): Likewise.
14181 * expr.c (expand_expr_real_2): Likewise.
14182 (expand_expr_real_1): Likewise.
14183 (try_casesi): Likewise.
14184 * fold-const-call.c (fold_const_call_ss): Likewise.
14185 * fold-const.c (unextend): Likewise.
14186 (extract_muldiv_1): Likewise.
14187 (fold_single_bit_test): Likewise.
14188 (native_encode_int): Likewise.
14189 (native_encode_string): Likewise.
14190 (native_interpret_int): Likewise.
14191 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
14192 * internal-fn.c (expand_addsub_overflow): Likewise.
14193 (expand_neg_overflow): Likewise.
14194 (expand_mul_overflow): Likewise.
14195 (expand_arith_overflow): Likewise.
14196 * match.pd: Likewise.
14197 * stor-layout.c (layout_type): Likewise.
14198 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
14199 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
14200 * tree-ssanames.c (get_range_info): Likewise.
14201 * tree-switch-conversion.c (array_value_type) Likewise.
14202 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
14203 (vect_recog_divmod_pattern): Likewise.
14204 (vect_recog_mixed_size_cond_pattern): Likewise.
14205 * tree-vrp.c (extract_range_basic): Likewise.
14206 (simplify_float_conversion_using_ranges): Likewise.
14207 * tree.c (int_fits_type_p): Likewise.
14208 * ubsan.c (instrument_bool_enum_load): Likewise.
14209 * varasm.c (mergeable_string_section): Likewise.
14210 (narrowing_initializer_constant_valid_p): Likewise.
14211 (output_constant): Likewise.
14213 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14214 Alan Hayward <alan.hayward@arm.com>
14215 David Sherwood <david.sherwood@arm.com>
14217 * machmode.h (NARROWEST_INT_MODE): New macro.
14218 * expr.c (alignment_for_piecewise_move): Use it instead of
14219 GET_CLASS_NARROWEST_MODE (MODE_INT).
14220 (push_block): Likewise.
14221 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
14223 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14225 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14226 Alan Hayward <alan.hayward@arm.com>
14227 David Sherwood <david.sherwood@arm.com>
14229 * postreload.c (move2add_valid_value_p): Change the type of the
14230 mode parameter to scalar_int_mode.
14231 (move2add_use_add2_insn): Add a mode parameter and use it instead
14233 (move2add_use_add3_insn): Likewise.
14234 (reload_cse_move2add): Update accordingly.
14236 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14237 Alan Hayward <alan.hayward@arm.com>
14238 David Sherwood <david.sherwood@arm.com>
14240 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
14242 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
14243 * optabs.c (expand_unop): Likewise.
14245 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14246 Alan Hayward <alan.hayward@arm.com>
14247 David Sherwood <david.sherwood@arm.com>
14249 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
14250 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
14251 (popcount_loc_descriptor): Likewise.
14252 (bswap_loc_descriptor): Likewise.
14253 (rotate_loc_descriptor): Likewise.
14254 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
14255 calling the functions above.
14257 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14258 Alan Hayward <alan.hayward@arm.com>
14259 David Sherwood <david.sherwood@arm.com>
14261 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
14263 (try_combine): Likewise.
14264 (simplify_if_then_else): Likewise.
14265 * cse.c (cse_insn): Likewise.
14266 * dwarf2out.c (mem_loc_descriptor): Likewise.
14267 * emit-rtl.c (gen_lowpart_common): Likewise.
14268 * simplify-rtx.c (simplify_truncation): Likewise.
14269 (simplify_binary_operation_1): Likewise.
14270 (simplify_const_relational_operation): Likewise.
14271 (simplify_ternary_operation): Likewise.
14272 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
14274 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14275 Alan Hayward <alan.hayward@arm.com>
14276 David Sherwood <david.sherwood@arm.com>
14278 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
14279 * reload.c (push_reload): Likewise.
14280 (find_reloads): Likewise.
14282 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14283 Alan Hayward <alan.hayward@arm.com>
14284 David Sherwood <david.sherwood@arm.com>
14286 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
14287 (make_compound_operation_int): Likewise.
14288 (change_zero_ext): Likewise.
14289 * expr.c (convert_move): Likewise.
14290 (convert_modes): Likewise.
14291 * fwprop.c (forward_propagate_subreg): Likewise.
14292 * loop-iv.c (get_biv_step_1): Likewise.
14293 * optabs.c (widen_operand): Likewise.
14294 * postreload.c (move2add_valid_value_p): Likewise.
14295 * recog.c (simplify_while_replacing): Likewise.
14296 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14297 (simplify_binary_operation_1): Likewise. Remove redundant
14298 mode equality check.
14300 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14301 Alan Hayward <alan.hayward@arm.com>
14302 David Sherwood <david.sherwood@arm.com>
14304 * combine.c (combine_simplify_rtx): Add checks for
14305 is_a <scalar_int_mode>.
14306 (simplify_if_then_else): Likewise.
14307 (make_field_assignment): Likewise.
14308 (simplify_comparison): Likewise.
14309 * ifcvt.c (noce_try_bitop): Likewise.
14310 * loop-invariant.c (canonicalize_address_mult): Likewise.
14311 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14313 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14314 Alan Hayward <alan.hayward@arm.com>
14315 David Sherwood <david.sherwood@arm.com>
14317 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
14318 is_a <scalar_int_mode> instead of != BLKmode.
14320 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14321 Alan Hayward <alan.hayward@arm.com>
14322 David Sherwood <david.sherwood@arm.com>
14324 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
14325 instead of != VOIDmode.
14326 * combine.c (if_then_else_cond): Likewise.
14327 (change_zero_ext): Likewise.
14328 * dwarf2out.c (mem_loc_descriptor): Likewise.
14329 (loc_descriptor): Likewise.
14330 * rtlanal.c (canonicalize_condition): Likewise.
14331 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
14333 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14334 Alan Hayward <alan.hayward@arm.com>
14335 David Sherwood <david.sherwood@arm.com>
14337 * simplify-rtx.c (simplify_binary_operation_1): Use
14338 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
14340 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14341 Alan Hayward <alan.hayward@arm.com>
14342 David Sherwood <david.sherwood@arm.com>
14344 * wide-int.h (int_traits<unsigned char>) New class.
14345 (int_traits<unsigned short>) Likewise.
14346 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
14347 Use GET_MODE_UNIT_PRECISION and remove redundant test for
14349 * combine.c (set_nonzero_bits_and_sign_copies): Use
14350 is_a <scalar_int_mode>.
14351 (find_split_point): Likewise.
14352 (combine_simplify_rtx): Likewise.
14353 (simplify_logical): Likewise.
14354 (expand_compound_operation): Likewise.
14355 (expand_field_assignment): Likewise.
14356 (make_compound_operation): Likewise.
14357 (extended_count): Likewise.
14358 (change_zero_ext): Likewise.
14359 (simplify_comparison): Likewise.
14360 * dwarf2out.c (scompare_loc_descriptor): Likewise.
14361 (ucompare_loc_descriptor): Likewise.
14362 (minmax_loc_descriptor): Likewise.
14363 (mem_loc_descriptor): Likewise.
14364 (loc_descriptor): Likewise.
14365 * expmed.c (init_expmed_one_mode): Likewise.
14366 * lra-constraints.c (lra_constraint_offset): Likewise.
14367 * optabs.c (prepare_libcall_arg): Likewise.
14368 * postreload.c (move2add_note_store): Likewise.
14369 * reload.c (operands_match_p): Likewise.
14370 * rtl.h (load_extend_op): Likewise.
14371 * rtlhooks.c (gen_lowpart_general): Likewise.
14372 * simplify-rtx.c (simplify_truncation): Likewise.
14373 (simplify_unary_operation_1): Likewise.
14374 (simplify_binary_operation_1): Likewise.
14375 (simplify_const_binary_operation): Likewise.
14376 (simplify_const_relational_operation): Likewise.
14377 (simplify_subreg): Likewise.
14378 * stor-layout.c (bitwise_mode_for_mode): Likewise.
14379 * var-tracking.c (adjust_mems): Likewise.
14380 (prepare_call_arguments): Likewise.
14382 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14383 Alan Hayward <alan.hayward@arm.com>
14384 David Sherwood <david.sherwood@arm.com>
14386 * machmode.h (is_int_mode): New fuction.
14387 * combine.c (find_split_point): Use it.
14388 (combine_simplify_rtx): Likewise.
14389 (simplify_if_then_else): Likewise.
14390 (simplify_set): Likewise.
14391 (simplify_shift_const_1): Likewise.
14392 (simplify_comparison): Likewise.
14393 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
14394 * cse.c (notreg_cost): Likewise.
14395 (cse_insn): Likewise.
14396 * cselib.c (cselib_lookup_1): Likewise.
14397 * dojump.c (do_jump_1): Likewise.
14398 (do_compare_rtx_and_jump): Likewise.
14399 * dse.c (get_call_args): Likewise.
14400 * dwarf2out.c (rtl_for_decl_init): Likewise.
14401 (native_encode_initializer): Likewise.
14402 * expmed.c (emit_store_flag_1): Likewise.
14403 (emit_store_flag): Likewise.
14404 * expr.c (convert_modes): Likewise.
14405 (store_field): Likewise.
14406 (expand_expr_real_1): Likewise.
14407 * fold-const.c (fold_read_from_constant_string): Likewise.
14408 * gimple-ssa-sprintf.c (get_format_string): Likewise.
14409 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
14410 * optabs.c (expand_binop): Likewise.
14411 (expand_unop): Likewise.
14412 (expand_abs_nojump): Likewise.
14413 (expand_one_cmpl_abs_nojump): Likewise.
14414 * simplify-rtx.c (mode_signbit_p): Likewise.
14415 (val_signbit_p): Likewise.
14416 (val_signbit_known_set_p): Likewise.
14417 (val_signbit_known_clear_p): Likewise.
14418 (simplify_relational_operation_1): Likewise.
14419 * tree.c (vector_type_mode): Likewise.
14421 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14422 Alan Hayward <alan.hayward@arm.com>
14423 David Sherwood <david.sherwood@arm.com>
14425 * machmode.h (smallest_mode_for_size): Fix formatting.
14426 (smallest_int_mode_for_size): New function.
14427 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
14428 instead of smallest_mode_for_size.
14429 * combine.c (make_extraction): Likewise.
14430 * config/arc/arc.c (arc_expand_movmem): Likewise.
14431 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
14432 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
14433 * config/s390/s390.c (s390_expand_insv): Likewise.
14434 * config/sparc/sparc.c (assign_int_registers): Likewise.
14435 * config/spu/spu.c (spu_function_value): Likewise.
14436 (spu_function_arg): Likewise.
14437 * coverage.c (get_gcov_type): Likewise.
14438 (get_gcov_unsigned_t): Likewise.
14439 * dse.c (find_shift_sequence): Likewise.
14440 * expmed.c (store_bit_field_1): Likewise.
14441 * expr.c (convert_move): Likewise.
14442 (store_field): Likewise.
14443 * internal-fn.c (expand_arith_overflow): Likewise.
14444 * optabs-query.c (get_best_extraction_insn): Likewise.
14445 * optabs.c (expand_twoval_binop_libfunc): Likewise.
14446 * stor-layout.c (layout_type): Likewise.
14447 (initialize_sizetypes): Likewise.
14448 * targhooks.c (default_get_mask_mode): Likewise.
14449 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
14451 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14452 Alan Hayward <alan.hayward@arm.com>
14453 David Sherwood <david.sherwood@arm.com>
14455 * machmode.h (opt_mode::else_blk): New function.
14456 (int_mode_for_mode): Declare.
14457 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
14458 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
14460 * cfgexpand.c (expand_debug_expr): Likewise.
14461 * combine.c (gen_lowpart_or_truncate): Likewise.
14462 (gen_lowpart_for_combine): Likewise.
14463 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
14464 * config/avr/avr.c (avr_to_int_mode): Likewise.
14465 (avr_out_plus_1): Likewise.
14466 (avr_out_plus): Likewise.
14467 (avr_out_round): Likewise.
14468 * config/i386/i386.c (ix86_split_to_parts): Likewise.
14469 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
14470 (s390_expand_vcond): Likewise.
14471 * config/spu/spu.c (spu_split_immediate): Likewise.
14472 (spu_expand_mov): Likewise.
14473 * dse.c (get_stored_val): Likewise.
14474 * expmed.c (store_bit_field_1): Likewise.
14475 (convert_extracted_bit_field): Use int_mode_for_mode instead of
14477 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
14478 (extract_low_bits): Likewise.
14479 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
14480 handling rather than repeating the check.
14481 (emit_group_store): Likewise.
14482 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
14483 * optabs.c (expand_absneg_bit): Likewise.
14484 (expand_copysign_absneg): Likewise.
14485 (expand_copysign_bit): Likewise.
14486 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
14487 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
14488 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
14489 * var-tracking.c (prepare_call_arguments): Likewise.
14490 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
14491 int_mode_for_mode instead of mode_for_size.
14492 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
14494 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14495 Alan Hayward <alan.hayward@arm.com>
14496 David Sherwood <david.sherwood@arm.com>
14498 * machmode.h (int_mode_for_size): New function.
14499 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
14500 instead of mode_for_size.
14501 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
14503 * combine.c (expand_field_assignment): Use int_mode_for_size
14504 instead of mode_for_size.
14505 (make_extraction): Likewise.
14506 (simplify_shift_const_1): Likewise.
14507 (simplify_comparison): Likewise.
14508 * dojump.c (do_jump): Likewise.
14509 * dwarf2out.c (mem_loc_descriptor): Likewise.
14510 * emit-rtl.c (init_derived_machine_modes): Likewise.
14511 * expmed.c (flip_storage_order): Likewise.
14512 (convert_extracted_bit_field): Likewise.
14513 * expr.c (copy_blkmode_from_reg): Likewise.
14514 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
14515 * internal-fn.c (expand_mul_overflow): Likewise.
14516 * lower-subreg.c (simple_move): Likewise.
14517 * optabs-libfuncs.c (init_optabs): Likewise.
14518 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14519 * tree.c (vector_type_mode): Likewise.
14520 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
14521 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
14522 * tree-vect-generic.c (expand_vector_parallel): Likewise.
14523 * tree-vect-stmts.c (vectorizable_load): Likewise.
14524 (vectorizable_store): Likewise.
14526 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14527 Alan Hayward <alan.hayward@arm.com>
14528 David Sherwood <david.sherwood@arm.com>
14530 * coretypes.h (pod_mode): New type.
14531 (scalar_int_mode_pod): New typedef.
14532 * machmode.h (pod_mode): New class.
14533 (int_n_data_t::m): Change type to scalar_int_mode_pod.
14534 * genmodes.c (emit_mode_int_n): Update accordingly.
14535 * lower-subreg.h (target_lower_subreg): Change type to
14536 scalar_int_mode_pod.
14537 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
14538 scalar_int_mode_pod.
14540 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14541 Alan Hayward <alan.hayward@arm.com>
14542 David Sherwood <david.sherwood@arm.com>
14544 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
14545 machine_mode to scalar_int_mode.
14546 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
14547 (rs6000_option_override_internal): Remove cast to int.
14548 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
14549 machine_mode to scalar_int_mode.
14550 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
14551 (rs6000_option_override_internal): Remove cast to int.
14552 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
14553 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
14555 * config/s390/s390.c (s390_expand_builtin): Likewise.
14556 * coretypes.h (scalar_int_mode): New type.
14557 (opt_scalar_int_mode): New typedef.
14558 * machmode.h (scalar_int_mode): New class.
14559 (scalar_int_mode::includes_p): New function.
14560 (byte_mode): Change type to scalar_int_mode.
14561 (word_mode): Likewise.
14562 (ptr_mode): Likewise.
14563 * emit-rtl.c (byte_mode): Likewise.
14564 (word_mode): Likewise.
14565 (ptr_mode): Likewise.
14566 (init_derived_machine_modes): Update accordingly.
14567 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
14568 and MODE_PARTIAL_INT.
14569 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
14570 opt_scalar_int_mode.
14572 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14573 Alan Hayward <alan.hayward@arm.com>
14574 David Sherwood <david.sherwood@arm.com>
14576 * target.def (libgcc_floating_mode_supported_p): Take a
14578 * doc/tm.texi: Regenerate.
14579 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
14581 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
14582 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
14585 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14586 Alan Hayward <alan.hayward@arm.com>
14587 David Sherwood <david.sherwood@arm.com>
14589 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
14590 * doc/tm.texi: Regenerate.
14591 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
14592 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
14593 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
14594 * targhooks.h (default_floatn_mode): Likewise.
14595 * targhooks.c (default_floatn_mode): Likewise.
14596 * tree.c (build_common_tree_nodes): Update accordingly.
14598 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14599 Alan Hayward <alan.hayward@arm.com>
14600 David Sherwood <david.sherwood@arm.com>
14602 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
14603 (mode_iterator::iterate_p): Likewise.
14604 (mode_iterator::get_wider): Likewise.
14605 * expr.c (init_expr_target): Use opt_scalar_float_mode.
14607 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14608 Alan Hayward <alan.hayward@arm.com>
14609 David Sherwood <david.sherwood@arm.com>
14611 * coretypes.h (opt_scalar_float_mode): New typedef.
14612 * machmode.h (float_mode_for_size): New function.
14613 * emit-rtl.c (double_mode): Delete.
14614 (init_emit_once): Use float_mode_for_size.
14615 * stor-layout.c (layout_type): Likewise.
14616 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
14618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14619 Alan Hayward <alan.hayward@arm.com>
14620 David Sherwood <david.sherwood@arm.com>
14622 * output.h (assemble_real): Take a scalar_float_mode.
14623 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
14624 * config/arm/arm.md (consttable_4): Likewise.
14625 (consttable_8): Likewise.
14626 (consttable_16): Likewise.
14627 * config/mips/mips.md (consttable_float): Likewise.
14628 * config/s390/s390.c (s390_output_pool_entry): Likewise.
14629 * varasm.c (assemble_real): Take a scalar_float_mode.
14630 (output_constant_pool_2): Update accordingly.
14631 (output_constant): Likewise.
14633 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14634 Alan Hayward <alan.hayward@arm.com>
14635 David Sherwood <david.sherwood@arm.com>
14637 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
14638 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
14639 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
14640 (native_encode_real): Likewise.
14641 (native_interpret_real): Likewise.
14642 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
14643 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14645 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14646 Alan Hayward <alan.hayward@arm.com>
14647 David Sherwood <david.sherwood@arm.com>
14649 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
14650 <scalar_float_mode>. Simplify.
14651 (gen_extend_conv_libfunc): Likewise.
14653 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14654 Alan Hayward <alan.hayward@arm.com>
14655 David Sherwood <david.sherwood@arm.com>
14657 * coretypes.h (scalar_float_mode): New type.
14658 * machmode.h (mode_traits::from_int): Use machine_mode if
14659 USE_ENUM_MODES is defined.
14660 (is_a): New function.
14662 (dyn_cast): Likewise.
14663 (scalar_float_mode): New class.
14664 (scalar_float_mode::includes_p): New function.
14665 (is_float_mode): Likewise.
14666 * gdbhooks.py (MachineModePrinter): New class.
14667 (build_pretty_printer): Use it for scalar_float_mode.
14668 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
14669 (format_helper::format_helper): Turn into a template.
14670 * genmodes.c (get_mode_class): New function.
14671 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
14672 or machine_mode if none.
14673 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
14674 as_a <scalar_float_mode>.
14675 * dwarf2out.c (mem_loc_descriptor): Likewise.
14676 (insert_float): Likewise.
14677 (add_const_value_attribute): Likewise.
14678 * simplify-rtx.c (simplify_immed_subreg): Likewise.
14679 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
14680 (expand_unop): Update accordingly.
14681 (expand_abs_nojump): Likewise.
14682 (expand_copysign_absneg): Take a scalar_float_mode.
14683 (expand_copysign_bit): Likewise.
14684 (expand_copysign): Update accordingly.
14686 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14687 Alan Hayward <alan.hayward@arm.com>
14688 David Sherwood <david.sherwood@arm.com>
14690 * coretypes.h (opt_mode): New class.
14691 * machmode.h (opt_mode): Likewise.
14692 (opt_mode::else_void): New function.
14693 (opt_mode::require): Likewise.
14694 (opt_mode::exists): Likewise.
14695 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
14696 (GET_MODE_2XWIDER_MODE): Likewise.
14697 (mode_iterator::get_wider): Update accordingly.
14698 (mode_iterator::get_2xwider): Likewise.
14699 (mode_iterator::get_known_wider): Likewise, turning into a template.
14700 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
14701 forcing a wider mode to exist.
14702 * config/cr16/cr16.h (LONG_REG_P): Likewise.
14703 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
14704 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
14705 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
14706 * lower-subreg.c (init_lower_subreg): Likewise.
14707 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
14708 on the final iteration.
14709 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
14710 a wider mode exists before asking for a move pattern.
14711 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
14712 forcing a wider mode to exist.
14713 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
14714 returning false if no such mode exists.
14715 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
14716 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
14717 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
14718 Avoid checking for a MODE_INT if we already know the mode is not a
14720 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
14721 forcing a wider mode to exist.
14722 (expmed_mult_highpart_optab): Likewise.
14723 (expmed_mult_highpart): Likewise.
14724 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
14726 * lto-streamer-in.c (lto_input_mode_table): Likewise.
14727 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
14728 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
14729 * internal-fn.c (expand_mul_overflow): Update use of
14730 GET_MODE_2XWIDER_MODE.
14731 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14732 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
14733 GET_MODE_WIDER_MODE.
14734 (convert_plusminus_to_widen): Likewise.
14735 * tree-switch-conversion.c (array_value_type): Likewise.
14736 * var-tracking.c (emit_note_insn_var_location): Likewise.
14737 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14738 Return false inside rather than outside the loop if no wider mode
14740 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
14741 and GET_MODE_2XWIDER_MODE
14742 (can_compare_p): Use else_void.
14743 * gdbhooks.py (OptMachineModePrinter): New class.
14744 (build_pretty_printer): Use it for opt_mode.
14746 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14747 Alan Hayward <alan.hayward@arm.com>
14748 David Sherwood <david.sherwood@arm.com>
14750 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
14751 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
14753 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14754 Alan Hayward <alan.hayward@arm.com>
14755 David Sherwood <david.sherwood@arm.com>
14757 * machmode.h (mode_traits): New structure.
14758 (get_narrowest_mode): New function.
14759 (mode_iterator::start): Likewise.
14760 (mode_iterator::iterate_p): Likewise.
14761 (mode_iterator::get_wider): Likewise.
14762 (mode_iterator::get_known_wider): Likewise.
14763 (mode_iterator::get_2xwider): Likewise.
14764 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
14765 (FOR_EACH_MODE): Likewise.
14766 (FOR_EACH_MODE_FROM): Likewise.
14767 (FOR_EACH_MODE_UNTIL): Likewise.
14768 (FOR_EACH_WIDER_MODE): Likewise.
14769 (FOR_EACH_2XWIDER_MODE): Likewise.
14770 * builtins.c (expand_builtin_strlen): Use new mode iterators.
14771 * combine.c (simplify_comparison): Likewise
14772 * config/i386/i386.c (type_natural_mode): Likewise.
14773 * cse.c (cse_insn): Likewise.
14774 * dse.c (find_shift_sequence): Likewise.
14775 * emit-rtl.c (init_derived_machine_modes): Likewise.
14776 (init_emit_once): Likewise.
14777 * explow.c (hard_function_value): Likewise.
14778 * expmed.c (extract_fixed_bit_field_1): Likewise.
14779 (extract_bit_field_1): Likewise.
14780 (expand_divmod): Likewise.
14781 (emit_store_flag_1): Likewise.
14782 * expr.c (init_expr_target): Likewise.
14783 (convert_move): Likewise.
14784 (alignment_for_piecewise_move): Likewise.
14785 (widest_int_mode_for_size): Likewise.
14786 (emit_block_move_via_movmem): Likewise.
14787 (copy_blkmode_to_reg): Likewise.
14788 (set_storage_via_setmem): Likewise.
14789 (compress_float_constant): Likewise.
14790 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14791 * optabs-query.c (get_best_extraction_insn): Likewise.
14792 * optabs.c (expand_binop): Likewise.
14793 (expand_twoval_unop): Likewise.
14794 (expand_twoval_binop): Likewise.
14795 (widen_leading): Likewise.
14796 (widen_bswap): Likewise.
14797 (expand_parity): Likewise.
14798 (expand_unop): Likewise.
14799 (prepare_cmp_insn): Likewise.
14800 (prepare_float_lib_cmp): Likewise.
14801 (expand_float): Likewise.
14802 (expand_fix): Likewise.
14803 (expand_sfix_optab): Likewise.
14804 * postreload.c (move2add_use_add2_insn): Likewise.
14805 * reg-stack.c (reg_to_stack): Likewise.
14806 * reginfo.c (choose_hard_reg_mode): Likewise.
14807 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
14808 * stor-layout.c (mode_for_size): Likewise.
14809 (smallest_mode_for_size): Likewise.
14810 (mode_for_vector): Likewise.
14811 (finish_bitfield_representative): Likewise.
14812 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
14813 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
14814 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14815 * var-tracking.c (prepare_call_arguments): Likewise.
14817 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14818 Alan Hayward <alan.hayward@arm.com>
14819 David Sherwood <david.sherwood@arm.com>
14821 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
14822 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
14823 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
14824 * machmode.h (mode_size): Move earlier in file.
14825 (mode_precision): Likewise.
14826 (mode_inner): Likewise.
14827 (mode_nunits): Likewise.
14828 (mode_unit_size): Likewise.
14829 (unit_unit_precision): Likewise.
14830 (mode_wider): Likewise.
14831 (mode_2xwider): Likewise.
14832 (machine_mode): New class.
14833 (mode_to_bytes): New function.
14834 (mode_to_bits): Likewise.
14835 (mode_to_precision): Likewise.
14836 (mode_to_inner): Likewise.
14837 (mode_to_unit_size): Likewise.
14838 (mode_to_unit_precision): Likewise.
14839 (mode_to_nunits): Likewise.
14840 (GET_MODE_SIZE): Use mode_to_bytes.
14841 (GET_MODE_BITSIZE): Use mode_to_bits.
14842 (GET_MODE_PRECISION): Use mode_to_precision.
14843 (GET_MODE_INNER): Use mode_to_inner.
14844 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
14845 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
14846 (GET_MODE_NUNITS): Use mode_to_nunits.
14847 * system.h (ALWAYS_INLINE): New macro.
14848 * config/powerpcspe/powerpcspe-c.c
14849 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
14850 int for arg1_mode and arg2_mode.
14852 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14853 Alan Hayward <alan.hayward@arm.com>
14854 David Sherwood <david.sherwood@arm.com>
14856 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
14857 Prefix mode names with E_ in case statements.
14858 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14859 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
14860 (aarch64_split_simd_move): Likewise.
14861 (aarch64_gen_storewb_pair): Likewise.
14862 (aarch64_gen_loadwb_pair): Likewise.
14863 (aarch64_gen_store_pair): Likewise.
14864 (aarch64_gen_load_pair): Likewise.
14865 (aarch64_get_condition_code_1): Likewise.
14866 (aarch64_constant_pool_reload_icode): Likewise.
14867 (get_rsqrte_type): Likewise.
14868 (get_rsqrts_type): Likewise.
14869 (get_recpe_type): Likewise.
14870 (get_recps_type): Likewise.
14871 (aarch64_gimplify_va_arg_expr): Likewise.
14872 (aarch64_simd_container_mode): Likewise.
14873 (aarch64_emit_load_exclusive): Likewise.
14874 (aarch64_emit_store_exclusive): Likewise.
14875 (aarch64_expand_compare_and_swap): Likewise.
14876 (aarch64_gen_atomic_cas): Likewise.
14877 (aarch64_emit_bic): Likewise.
14878 (aarch64_emit_atomic_swap): Likewise.
14879 (aarch64_emit_atomic_load_op): Likewise.
14880 (aarch64_evpc_trn): Likewise.
14881 (aarch64_evpc_uzp): Likewise.
14882 (aarch64_evpc_zip): Likewise.
14883 (aarch64_evpc_ext): Likewise.
14884 (aarch64_evpc_rev): Likewise.
14885 (aarch64_evpc_dup): Likewise.
14886 (aarch64_gen_ccmp_first): Likewise.
14887 (aarch64_gen_ccmp_next): Likewise.
14888 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
14889 (alpha_emit_xfloating_libcall): Likewise.
14890 (emit_insxl): Likewise.
14891 (alpha_arg_type): Likewise.
14892 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
14893 (arc_preferred_simd_mode): Likewise.
14894 (arc_secondary_reload): Likewise.
14895 (get_arc_condition_code): Likewise.
14896 (arc_print_operand): Likewise.
14897 (arc_legitimate_constant_p): Likewise.
14898 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14899 * config/arc/arc.md (casesi_load): Likewise.
14900 (casesi_compact_jump): Likewise.
14901 * config/arc/predicates.md (proper_comparison_operator): Likewise.
14902 (cc_use_register): Likewise.
14903 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
14904 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
14905 (arm_init_iwmmxt_builtins): Likewise.
14906 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
14907 (neon_expand_vector_init): Likewise.
14908 (arm_attr_length_move_neon): Likewise.
14909 (maybe_get_arm_condition_code): Likewise.
14910 (arm_emit_vector_const): Likewise.
14911 (arm_preferred_simd_mode): Likewise.
14912 (arm_output_iwmmxt_tinsr): Likewise.
14913 (thumb1_output_casesi): Likewise.
14914 (thumb2_output_casesi): Likewise.
14915 (arm_emit_load_exclusive): Likewise.
14916 (arm_emit_store_exclusive): Likewise.
14917 (arm_expand_compare_and_swap): Likewise.
14918 (arm_evpc_neon_vuzp): Likewise.
14919 (arm_evpc_neon_vzip): Likewise.
14920 (arm_evpc_neon_vrev): Likewise.
14921 (arm_evpc_neon_vtrn): Likewise.
14922 (arm_evpc_neon_vext): Likewise.
14923 (arm_validize_comparison): Likewise.
14924 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
14925 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
14926 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
14927 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
14928 (c6x_preferred_simd_mode): Likewise.
14929 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
14930 (epiphany_rtx_costs): Likewise.
14931 * config/epiphany/predicates.md (proper_comparison_operator):
14933 * config/frv/frv.c (condexec_memory_operand): Likewise.
14934 (frv_emit_move): Likewise.
14935 (output_move_single): Likewise.
14936 (output_condmove_single): Likewise.
14937 (frv_hard_regno_mode_ok): Likewise.
14938 (frv_matching_accg_mode): Likewise.
14939 * config/h8300/h8300.c (split_adds_subs): Likewise.
14940 (h8300_rtx_costs): Likewise.
14941 (h8300_print_operand): Likewise.
14942 (compute_mov_length): Likewise.
14943 (output_logical_op): Likewise.
14944 (compute_logical_op_length): Likewise.
14945 (compute_logical_op_cc): Likewise.
14946 (h8300_shift_needs_scratch_p): Likewise.
14947 (output_a_shift): Likewise.
14948 (compute_a_shift_length): Likewise.
14949 (compute_a_shift_cc): Likewise.
14950 (expand_a_rotate): Likewise.
14951 (output_a_rotate): Likewise.
14952 * config/i386/i386.c (classify_argument): Likewise.
14953 (function_arg_advance_32): Likewise.
14954 (function_arg_32): Likewise.
14955 (function_arg_64): Likewise.
14956 (function_value_64): Likewise.
14957 (ix86_gimplify_va_arg): Likewise.
14958 (ix86_legitimate_constant_p): Likewise.
14959 (put_condition_code): Likewise.
14960 (split_double_mode): Likewise.
14961 (ix86_avx256_split_vector_move_misalign): Likewise.
14962 (ix86_expand_vector_logical_operator): Likewise.
14963 (ix86_split_idivmod): Likewise.
14964 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
14965 (ix86_build_const_vector): Likewise.
14966 (ix86_build_signbit_mask): Likewise.
14967 (ix86_match_ccmode): Likewise.
14968 (ix86_cc_modes_compatible): Likewise.
14969 (ix86_expand_branch): Likewise.
14970 (ix86_expand_sse_cmp): Likewise.
14971 (ix86_expand_sse_movcc): Likewise.
14972 (ix86_expand_int_sse_cmp): Likewise.
14973 (ix86_expand_vec_perm_vpermi2): Likewise.
14974 (ix86_expand_vec_perm): Likewise.
14975 (ix86_expand_sse_unpack): Likewise.
14976 (ix86_expand_int_addcc): Likewise.
14977 (ix86_split_to_parts): Likewise.
14978 (ix86_vectorize_builtin_gather): Likewise.
14979 (ix86_vectorize_builtin_scatter): Likewise.
14980 (avx_vpermilp_parallel): Likewise.
14981 (inline_memory_move_cost): Likewise.
14982 (ix86_tieable_integer_mode_p): Likewise.
14983 (x86_maybe_negate_const_int): Likewise.
14984 (ix86_expand_vector_init_duplicate): Likewise.
14985 (ix86_expand_vector_init_one_nonzero): Likewise.
14986 (ix86_expand_vector_init_one_var): Likewise.
14987 (ix86_expand_vector_init_concat): Likewise.
14988 (ix86_expand_vector_init_interleave): Likewise.
14989 (ix86_expand_vector_init_general): Likewise.
14990 (ix86_expand_vector_set): Likewise.
14991 (ix86_expand_vector_extract): Likewise.
14992 (emit_reduc_half): Likewise.
14993 (ix86_emit_i387_round): Likewise.
14994 (ix86_mangle_type): Likewise.
14995 (ix86_expand_round_sse4): Likewise.
14996 (expand_vec_perm_blend): Likewise.
14997 (canonicalize_vector_int_perm): Likewise.
14998 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
14999 (expand_vec_perm_1): Likewise.
15000 (expand_vec_perm_interleave3): Likewise.
15001 (expand_vec_perm_even_odd_pack): Likewise.
15002 (expand_vec_perm_even_odd_1): Likewise.
15003 (expand_vec_perm_broadcast_1): Likewise.
15004 (ix86_vectorize_vec_perm_const_ok): Likewise.
15005 (ix86_expand_vecop_qihi): Likewise.
15006 (ix86_expand_mul_widen_hilo): Likewise.
15007 (ix86_expand_sse2_abs): Likewise.
15008 (ix86_expand_pextr): Likewise.
15009 (ix86_expand_pinsr): Likewise.
15010 (ix86_preferred_simd_mode): Likewise.
15011 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
15012 * config/i386/sse.md (*andnot<mode>3): Likewise.
15013 (<mask_codefor><code><mode>3<mask_name>): Likewise.
15014 (*<code><mode>3): Likewise.
15015 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
15016 (ia64_expand_atomic_op): Likewise.
15017 (ia64_arg_type): Likewise.
15018 (ia64_mode_to_int): Likewise.
15019 (ia64_scalar_mode_supported_p): Likewise.
15020 (ia64_vector_mode_supported_p): Likewise.
15021 (expand_vec_perm_broadcast): Likewise.
15022 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
15023 (iq2000_function_arg_advance): Likewise.
15024 (iq2000_function_arg): Likewise.
15025 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
15026 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
15027 (m68k_libcall_value): Likewise.
15028 (m68k_function_value): Likewise.
15029 (sched_attr_op_type): Likewise.
15030 * config/mcore/mcore.c (mcore_output_move): Likewise.
15031 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
15033 (microblaze_function_arg): Likewise.
15034 * config/mips/mips.c (mips16_build_call_stub): Likewise.
15035 (mips_print_operand): Likewise.
15036 (mips_mode_ok_for_mov_fmt_p): Likewise.
15037 (mips_vector_mode_supported_p): Likewise.
15038 (mips_preferred_simd_mode): Likewise.
15039 (mips_expand_vpc_loongson_even_odd): Likewise.
15040 (mips_expand_vec_unpack): Likewise.
15041 (mips_expand_vi_broadcast): Likewise.
15042 (mips_expand_vector_init): Likewise.
15043 (mips_expand_vec_reduc): Likewise.
15044 (mips_expand_msa_cmp): Likewise.
15045 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
15046 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
15047 (cc_flags_for_mode): Likewise.
15048 * config/msp430/msp430.c (msp430_print_operand): Likewise.
15049 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
15050 (nds32_output_casesi_pc_relative): Likewise.
15051 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15052 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
15053 (nvptx_gen_unpack): Likewise.
15054 (nvptx_gen_pack): Likewise.
15055 (nvptx_gen_shuffle): Likewise.
15056 (nvptx_gen_wcast): Likewise.
15057 (nvptx_preferred_simd_mode): Likewise.
15058 * config/pa/pa.c (pa_secondary_reload): Likewise.
15059 * config/pa/predicates.md (base14_operand): Likewise.
15060 * config/powerpcspe/powerpcspe-c.c
15061 (altivec_resolve_overloaded_builtin): Likewise.
15062 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
15064 (rs6000_preferred_simd_mode): Likewise.
15065 (output_vec_const_move): Likewise.
15066 (rs6000_expand_vector_extract): Likewise.
15067 (rs6000_split_vec_extract_var): Likewise.
15068 (reg_offset_addressing_ok_p): Likewise.
15069 (rs6000_legitimate_offset_address_p): Likewise.
15070 (rs6000_legitimize_address): Likewise.
15071 (rs6000_emit_set_const): Likewise.
15072 (rs6000_const_vec): Likewise.
15073 (rs6000_emit_move): Likewise.
15074 (spe_build_register_parallel): Likewise.
15075 (rs6000_darwin64_record_arg_recurse): Likewise.
15076 (swap_selector_for_mode): Likewise.
15077 (spe_init_builtins): Likewise.
15078 (paired_init_builtins): Likewise.
15079 (altivec_init_builtins): Likewise.
15080 (do_load_for_compare): Likewise.
15081 (rs6000_generate_compare): Likewise.
15082 (rs6000_expand_float128_convert): Likewise.
15083 (emit_load_locked): Likewise.
15084 (emit_store_conditional): Likewise.
15085 (rs6000_output_function_epilogue): Likewise.
15086 (rs6000_handle_altivec_attribute): Likewise.
15087 (rs6000_function_value): Likewise.
15088 (emit_fusion_gpr_load): Likewise.
15089 (emit_fusion_p9_load): Likewise.
15090 (emit_fusion_p9_store): Likewise.
15091 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
15092 (fusion_gpr_mem_load): Likewise.
15093 (fusion_addis_mem_combo_load): Likewise.
15094 (fusion_addis_mem_combo_store): Likewise.
15095 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
15096 (fusion_gpr_mem_load): Likewise.
15097 (fusion_addis_mem_combo_load): Likewise.
15098 (fusion_addis_mem_combo_store): Likewise.
15099 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15101 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
15102 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15103 (rs6000_preferred_simd_mode): Likewise.
15104 (output_vec_const_move): Likewise.
15105 (rs6000_expand_vector_extract): Likewise.
15106 (rs6000_split_vec_extract_var): Likewise.
15107 (reg_offset_addressing_ok_p): Likewise.
15108 (rs6000_legitimate_offset_address_p): Likewise.
15109 (rs6000_legitimize_address): Likewise.
15110 (rs6000_emit_set_const): Likewise.
15111 (rs6000_const_vec): Likewise.
15112 (rs6000_emit_move): Likewise.
15113 (rs6000_darwin64_record_arg_recurse): Likewise.
15114 (swap_selector_for_mode): Likewise.
15115 (paired_init_builtins): Likewise.
15116 (altivec_init_builtins): Likewise.
15117 (rs6000_expand_float128_convert): Likewise.
15118 (emit_load_locked): Likewise.
15119 (emit_store_conditional): Likewise.
15120 (rs6000_output_function_epilogue): Likewise.
15121 (rs6000_handle_altivec_attribute): Likewise.
15122 (rs6000_function_value): Likewise.
15123 (emit_fusion_gpr_load): Likewise.
15124 (emit_fusion_p9_load): Likewise.
15125 (emit_fusion_p9_store): Likewise.
15126 * config/rx/rx.c (rx_gen_move_template): Likewise.
15127 (flags_from_mode): Likewise.
15128 * config/s390/predicates.md (s390_alc_comparison): Likewise.
15129 (s390_slb_comparison): Likewise.
15130 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
15131 (s390_vector_mode_supported_p): Likewise.
15132 (s390_cc_modes_compatible): Likewise.
15133 (s390_match_ccmode_set): Likewise.
15134 (s390_canonicalize_comparison): Likewise.
15135 (s390_emit_compare_and_swap): Likewise.
15136 (s390_branch_condition_mask): Likewise.
15137 (s390_rtx_costs): Likewise.
15138 (s390_secondary_reload): Likewise.
15139 (__SECONDARY_RELOAD_CASE): Likewise.
15140 (s390_expand_cs): Likewise.
15141 (s390_preferred_simd_mode): Likewise.
15142 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
15143 * config/sh/sh.c (sh_print_operand): Likewise.
15144 (dump_table): Likewise.
15145 (sh_secondary_reload): Likewise.
15146 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15147 * config/sh/sh.md (casesi_worker_1): Likewise.
15148 (casesi_worker_2): Likewise.
15149 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
15150 (fcc_comparison_operator): Likewise.
15151 * config/sparc/sparc.c (sparc_expand_move): Likewise.
15152 (emit_soft_tfmode_cvt): Likewise.
15153 (sparc_preferred_simd_mode): Likewise.
15154 (output_cbranch): Likewise.
15155 (sparc_print_operand): Likewise.
15156 (sparc_expand_vec_perm_bmask): Likewise.
15157 (vector_init_bshuffle): Likewise.
15158 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
15159 (spu_vector_mode_supported_p): Likewise.
15160 (spu_expand_insv): Likewise.
15161 (spu_emit_branch_or_set): Likewise.
15162 (spu_handle_vector_attribute): Likewise.
15163 (spu_builtin_splats): Likewise.
15164 (spu_builtin_extract): Likewise.
15165 (spu_builtin_promote): Likewise.
15166 (spu_expand_sign_extend): Likewise.
15167 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
15168 (tilegx_simd_int): Likewise.
15169 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
15170 (tilepro_simd_int): Likewise.
15171 * config/v850/v850.c (const_double_split): Likewise.
15172 (v850_print_operand): Likewise.
15173 (ep_memory_offset): Likewise.
15174 * config/vax/vax.c (vax_rtx_costs): Likewise.
15175 (vax_output_int_move): Likewise.
15176 (vax_output_int_add): Likewise.
15177 (vax_output_int_subtract): Likewise.
15178 * config/visium/predicates.md (visium_branch_operator): Likewise.
15179 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
15180 (visium_print_operand_address): Likewise.
15181 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15182 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
15183 (xtensa_expand_conditional_branch): Likewise.
15184 (xtensa_copy_incoming_a7): Likewise.
15185 (xtensa_output_literal): Likewise.
15186 * dfp.c (decimal_real_maxval): Likewise.
15187 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
15189 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15190 Alan Hayward <alan.hayward@arm.com>
15191 David Sherwood <david.sherwood@arm.com>
15193 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
15194 (mode_nunits_inline): Likewise.
15195 (mode_inner_inline): Likewise.
15196 (mode_unit_size_inline): Likewise.
15197 (mode_unit_precision_inline): Likewise.
15198 (emit_insn_modes_h): Likewise. Also emit a #define of the
15200 (emit_mode_wider): Add an E_ prefix to mode names.
15201 (emit_mode_complex): Likewise.
15202 (emit_mode_inner): Likewise.
15203 (emit_mode_adjustments): Likewise.
15204 (emit_mode_int_n): Likewise.
15205 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
15206 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
15207 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
15208 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
15209 (CRC32_BUILTIN, ENTRY): Likewise.
15210 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
15211 (aarch64_pop_regs): Likewise.
15212 (aarch64_process_components): Likewise.
15213 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
15214 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
15215 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
15216 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
15217 * config/arm/arm.c (arm_init_libfuncs): Likewise.
15218 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
15220 * config/i386/i386-builtin.def (pcmpestr): Likewise.
15221 (pcmpistr): Likewise.
15222 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
15223 * config/mmix/mmix.c (mmix_output_condition): Likewise.
15224 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
15226 * config/rl78/rl78.c (mduc_regs): Likewise.
15227 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
15228 (htm_expand_builtin): Likewise.
15229 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
15230 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
15231 * config/xtensa/xtensa.c (print_operand): Likewise.
15232 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
15233 (NUM_MODE_VECTOR_INT): Likewise.
15234 * genoutput.c (null_operand): Likewise.
15235 (output_operand_data): Likewise.
15236 * genrecog.c (print_parameter_value): Likewise.
15237 * lra.c (debug_operand_data): Likewise.
15239 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15241 * dwarf2out.c (scompare_loc_descriptor_wide)
15242 (scompare_loc_descriptor_narrow): New functions, split out from...
15243 (scompare_loc_descriptor): ...here.
15244 * expmed.c (emit_store_flag_int): New function, split out from...
15245 (emit_store_flag): ...here.
15247 2017-08-30 Richard Biener <rguenther@suse.de>
15249 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
15250 (dwarf2out_early_finish): Move setting of AT_pubnames from
15251 early debug output to early finish.
15253 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15255 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
15256 and -mdata-region to the assembler.
15258 2017-08-30 Richard Biener <rguenther@suse.de>
15260 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
15262 (gen_subprogram_die): Add DW_AT_object_pointer only early.
15263 (dwarf2out_early_global_decl): Only generate a DIE for the
15264 abstract origin if it doesn't already exist or is a declaration DIE.
15265 (resolve_addr): Do not add the linkage name twice when
15266 generating a stub DIE for the DW_TAG_GNU_call_site target.
15268 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15270 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15271 Use machine_mode rather than int for arg1_mode.
15273 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
15276 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
15277 that the second argument of the built-in functions to unpack
15278 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
15279 switch statement instead a lot of if statements.
15280 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
15281 Allow 64-bit values to be in Altivec registers as well as
15282 traditional floating point registers.
15283 (pack<mode>, FMOVE128_VSX iterator): Likewise.
15285 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
15287 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
15288 MAX_REGS_PER_ADDRESS == 1.
15290 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
15292 * config/i386/i386.opt (flag_fentry): Do not init to -1.
15293 * config/i386/i386.c (ix86_option_override_internal): Simplify
15294 setting of opts->x_flag_entry.
15296 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15297 Jakub Jelinek <jakub@redhat.com>
15298 Richard Biener <rguenther@suse.de>
15300 PR tree-optimization/81503
15301 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
15302 folded constant fits in the target type; reorder tests for clarity.
15304 2017-08-29 Martin Liska <mliska@suse.cz>
15306 * passes.def: Include pass_lower_switch.
15307 * stmt.c (dump_case_nodes): Remove and move to
15308 tree-switch-conversion.
15309 (case_values_threshold): Likewise.
15310 (expand_switch_as_decision_tree_p): Likewise.
15311 (emit_case_decision_tree): Likewise.
15312 (expand_case): Likewise.
15313 (balance_case_nodes): Likewise.
15314 (node_has_low_bound): Likewise.
15315 (node_has_high_bound): Likewise.
15316 (node_is_bounded): Likewise.
15317 (emit_case_nodes): Likewise.
15318 (struct simple_case_node): New struct.
15319 (add_case_node): Remove.
15320 (emit_case_dispatch_table): Use vector instead of case_list.
15321 (reset_out_edges_aux): Remove.
15322 (compute_cases_per_edge): Likewise.
15323 (expand_case): Build list of simple_case_node.
15324 (expand_sjlj_dispatch_table): Use it.
15325 * tree-switch-conversion.c (struct case_node): Moved from
15326 stmt.c and adjusted.
15327 (emit_case_nodes): Likewise.
15328 (node_has_low_bound): Likewise.
15329 (node_has_high_bound): Likewise.
15330 (node_is_bounded): Likewise.
15331 (case_values_threshold): Likewise.
15332 (reset_out_edges_aux): Likewise.
15333 (compute_cases_per_edge): Likewise.
15334 (add_case_node): Likewise.
15335 (dump_case_nodes): Likewise.
15336 (balance_case_nodes): Likewise.
15337 (expand_switch_as_decision_tree_p): Likewise.
15338 (emit_jump): Likewise.
15339 (emit_case_decision_tree): Likewise.
15340 (try_switch_expansion): Likewise.
15341 (do_jump_if_equal): Likewise.
15342 (emit_cmp_and_jump_insns): Likewise.
15343 (fix_phi_operands_for_edge): New function.
15344 (record_phi_operand_mapping): Likewise.
15345 (class pass_lower_switch): New pass.
15346 (pass_lower_switch::execute): New function.
15347 (make_pass_lower_switch): Likewise.
15348 (conditional_probability):
15349 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
15350 * tree-pass.h: Add make_pass_lower_switch.
15352 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15355 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
15358 2017-08-29 Richard Biener <rguenther@suse.de>
15360 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
15361 we do not add a DW_AT_inline attribute twice.
15362 (gen_subprogram_die): Remove code setting DW_AT_inline on
15363 DECL_ABSTRACT_P nodes.
15365 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
15367 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
15368 calls to internal functions.
15369 (gimplify_modify_expr): Likewise.
15370 * tree-call-cdce.c (use_internal_fn): Likewise.
15371 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
15372 (convert_to_divmod): Set the nothrow flag.
15373 * tree-if-conv.c (predicate_mem_writes): Likewise.
15374 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15375 (vectorizable_call): Likewise.
15376 (vectorizable_store): Likewise.
15377 (vectorizable_load): Likewise.
15378 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
15379 (vect_recog_mask_conversion_pattern): Likewise.
15381 2017-08-29 Martin Liska <mliska@suse.cz>
15384 * gcc.c (driver_handle_option): Add new argument.
15385 * opts-common.c (handle_option): Pass
15386 target_option_override_hook.
15387 * opts-global.c (lang_handle_option): Add new option.
15388 (set_default_handlers): Add new argument.
15389 (decode_options): Likewise.
15390 * opts.c (target_handle_option): Likewise.
15391 (common_handle_option): Call target_option_override_hook.
15392 * opts.h (struct cl_option_handler_func): Add hook for
15393 target option override.
15394 (struct cl_option_handlers): Likewise.
15395 (set_default_handlers): Add new argument.
15396 (decode_options): Likewise.
15397 (common_handle_option): Likewise.
15398 (target_handle_option): Likewise.
15399 * toplev.c (toplev::main): Pass targetm.target_option.override
15402 2017-08-29 Richard Biener <rguenther@suse.de>
15403 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
15405 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
15406 life to the active subtree.
15408 2017-08-28 Jeff Law <law@redhat.com>
15410 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
15411 derive_equivalences.
15412 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
15413 Code moved into....
15414 (edge_info::derive_equivalences): New private member function
15416 * tree-ssa-dom.c (class edge_info): Changed from a struct
15417 to a class. Add ctor/dtor, methods and data members.
15418 (edge_info::edge_info): Renamed from allocate_edge_info.
15419 Initialize additional members.
15420 (edge_info::~edge_info): New.
15421 (free_dom_edge_info): Delete the edge info.
15422 (record_edge_info): Use new class & associated member functions.
15423 Tighten forms for testing for edge equivalences.
15424 (record_temporary_equivalences): Iterate over the simple
15425 equivalences rather than assuming there's only one per edge.
15426 (cprop_into_successor_phis): Iterate over the simple
15427 equivalences rather than assuming there's only one per edge.
15428 (optimize_stmt): Use operand_equal_p rather than pointer
15429 equality for mini-DSE code.
15431 2017-08-28 Nathan Sidwell <nathan@acm.org>
15433 * gcc.c (execute): Fold SIGPIPE handling into switch
15434 statement. Adjust internal error message.
15436 2017-08-28 Richard Biener <rguenther@suse.de>
15439 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
15440 Do nothing for removed DIEs.
15442 2017-08-28 Richard Biener <rguenther@suse.de>
15444 PR tree-optimization/81977
15445 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
15448 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
15451 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
15452 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
15453 using targetm.gen_mem_thread_fence.
15455 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
15458 * config/i386/i386.md (*<btsc><mode>): Change operand 2
15459 predicate to register_operand. Reorder operands.
15460 (*btr<mode>): Ditto.
15461 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
15462 (*btr<mode>_mask): Ditto.
15464 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
15466 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
15467 * config/rs6000/xmmintrin.h: New file.
15468 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
15470 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15473 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
15474 parameter and_insn and return it.
15475 (recombine_lvx_pattern): Insert a copy to ensure availability of
15476 the base register of the copied masking operation at the point of
15477 the instruction replacement.
15478 (recombine_stvx_pattern): Likewise.
15480 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
15482 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
15483 undocumented switches.
15484 (-mpower9-dform-vector): Likewise.
15485 (-mpower9-dform): Likewise.
15486 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
15487 comments to delete references to -mpower9-dform* switches.
15488 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
15489 Delete reference to -mpower9-dform* switches, test for
15490 -mpower9-vector instead.
15491 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
15492 (OTHER_P9_VECTOR_MASKS): Likewise.
15493 (POWERPC_MASKS): Likewise.
15494 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
15495 tests against -mpower9-dform* to -mpower9-vector. Delete code
15496 that checked for -mpower9-dform* consistancy with other options.
15497 Add test for -mpower9-misc to enable other power9 switches.
15498 (rs6000_init_hard_regno_mode_ok): Likewise.
15499 (rs6000_option_override_internal): Likewise.
15500 (rs6000_emit_prologue): Likewise.
15501 (rs6000_emit_epilogue): Likewise.
15502 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
15503 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
15504 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
15506 (emit_fusion_p9_store): Likewise.
15507 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
15508 resetting these macros if the assembler does not support ISA 3.0
15510 (TARGET_P9_DFORM_VECTOR): Likewise.
15511 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
15512 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
15514 2017-08-25 Alan Modra <amodra@gmail.com>
15516 PR rtl-optimization/81747
15517 * cse.c (cse_extended_basic_block): Don't attempt to record
15518 equivalences for degenerate conditional branches that branch
15519 to their fall-through.
15521 2017-08-24 Martin Sebor <msebor@redhat.com>
15523 PR middle-end/81908
15524 * gimple-fold.c (size_must_be_zero_p): New function.
15525 (gimple_fold_builtin_memory_op): Call it.
15527 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
15529 * config/rs6000/mm_malloc.h: New file.
15531 2017-08-24 Bin Cheng <bin.cheng@arm.com>
15533 PR tree-optimization/81913
15534 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
15535 analysis when either IVs in condition can wrap.
15537 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
15539 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
15540 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
15542 2017-08-24 Richard Biener <rguenther@suse.de>
15545 * targhooks.c (default_target_can_inline_p): Properly
15546 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
15547 is present and always compare.
15548 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
15549 infer -mfpmath=sse from TARGET_SSE_P.
15550 (ix86_can_inline_p): Properly use target_option_default_node when
15551 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
15553 2017-08-24 Richard Biener <rguenther@suse.de>
15556 * dwarf2out.c (output_die): Handle flag_generate_offload like
15558 (output_comp_unit): Likewise.
15559 (gen_array_type_die): Likewise.
15560 (dwarf2out_early_finish): Likewise.
15561 (note_variable_value_in_expr): Likewise.
15562 (dwarf2out_finish): Likewise. Adjust assert.
15563 * cgraphunit.c (symbol_table::compile): Move setting of
15564 flag_generate_offload earlier ...
15565 (symbol_table::finalize_compilation_unit): ... here, before
15566 early debug finalization.
15568 2017-08-24 Richard Biener <rguenther@suse.de>
15570 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
15571 and ipa-fnsummary.h.
15572 (ix86_can_inline_p): When ix86_fpmath flags do not match
15573 check whether the callee uses FP math at all.
15575 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
15577 PR middle-end/81931
15578 * tree-ssanames.c (get_nonzero_bits): Use element_precision
15579 instead of TYPE_PRECISION.
15581 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
15582 Alan Hayward <alan.hayward@arm.com>
15583 David Sherwood <david.sherwood@arm.com>
15585 * combine.c (make_extraction): Use subreg_offset_from_lsb.
15587 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
15589 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
15591 (ix86_frame::stack_realign_allocate): New field.
15592 (struct machine_frame_state): Modify comments.
15593 (machine_frame_state::sp_realigned_fp_end): New field.
15594 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
15595 layout calculation.
15596 (sp_valid_at): Add assertion to assure no attempt to access invalid
15597 offset of a realigned stack.
15598 (fp_valid_at): Likewise.
15599 (choose_baseaddr): Modify comments.
15600 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
15601 ix86_expand_prologue.
15602 (ix86_expand_prologue): Modify stack realignment and allocation.
15603 (ix86_expand_epilogue): Modify comments.
15604 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
15605 avx2_runtime, avx512f, and avx512f_runtime.
15607 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
15609 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
15610 (mstackrealign): Do not init to -1.
15611 * config/i386/i386.c (ix86_option_override_internal):
15612 Check opts_set, not opts when setting default value of
15613 opts->x_ix86_force_align_arg_pointer.
15615 2017-08-23 Richard Biener <rguenther@suse.de>
15617 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
15618 lang_hooks.decl_printable_name.
15619 * print-rtl-function.c (print_rtx_function): Likewise.
15620 * tree-pretty-print.c (dump_function_header): Likewise.
15622 2017-08-23 Richard Biener <rguenther@suse.de>
15625 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
15626 -g0 at compile-time.
15628 2017-08-23 Tamar Christina <tamar.christina@arm.com>
15630 PR middle-end/19706
15631 * doc/sourcebuild.texi (Other hardware attributes):
15634 2017-08-23 Tamar Christina <tamar.christina@arm.com>
15636 PR middle-end/19706
15637 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
15638 Fix single-use check.
15640 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
15642 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
15644 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
15646 * doc/install.texi: Modify to add more details on running selected
15649 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
15651 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
15652 is combined with -mabi=ms.
15653 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
15656 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15658 PR tree-optimization/81488
15659 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
15660 and cached_basis fields.
15661 (MAX_SPREAD): New constant.
15662 (alloc_cand_and_find_basis): Initialize new fields.
15663 (clear_visited): New function.
15664 (create_phi_basis_1): Rename from create_phi_basis, set visited
15665 and cached_basis fields.
15666 (create_phi_basis): New wrapper function.
15667 (phi_add_costs_1): Rename from phi_add_costs, add spread
15668 parameter, set visited field, short-circuit when limits reached.
15669 (phi_add_costs): New wrapper function.
15670 (record_phi_increments_1): Rename from record_phi_increments, set
15672 (record_phi_increments): New wrapper function.
15673 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
15674 (phi_incr_cost): New wrapper function.
15675 (all_phi_incrs_profitable_1): Rename from
15676 all_phi_incrs_profitable, set visited field.
15677 (all_phi_incrs_profitable): New wrapper function.
15679 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
15680 Alan Hayward <alan.hayward@arm.com>
15681 David Sherwood <david.sherwood@arm.com>
15683 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
15684 that takes the outer and inner modes.
15685 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
15686 comparison as the canonical test for a paradoxical subreg.
15687 * combine.c (simplify_set): Use paradoxical_subreg_p.
15688 (make_extraction): Likewise.
15689 (force_to_mode): Likewise.
15690 (rtx_equal_for_field_assignment_p): Likewise.
15691 (gen_lowpart_for_combine): Likewise.
15692 (simplify_comparison): Likewise.
15693 * cse.c (equiv_constant): Likewise.
15694 * expmed.c (store_bit_field_1): Likewise.
15695 * final.c (alter_subreg): Likewise.
15696 * fwprop.c (propagate_rtx): Likewise.
15697 (forward_propagate_subreg): Likewise.
15698 * ira-conflicts.c (ira_build_conflicts): Likewise.
15699 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
15700 * lra-constraints.c (curr_insn_transform): Likewise.
15701 (split_reg): Likewise.
15702 * lra-eliminations.c (move_plus_up): Likewise.
15703 (lra_eliminate_regs_1): Likewise.
15704 * recog.c (general_operand): Likewise.
15705 * ree.c (combine_reaching_defs): Likewise.
15706 * reload.c (push_reload): Likewise.
15707 (find_reloads): Likewise.
15708 * reload1.c (elimination_effects): Likewise.
15709 (compute_reload_subreg_offset): Likewise.
15710 (choose_reload_regs): Likewise.
15711 * rtlanal.c (subreg_lsb_1): Likewise.
15712 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15713 (simplify_subreg): Likewise.
15714 * var-tracking.c (track_loc_p): Likewise.
15715 * emit-rtl.c (byte_lowpart_offset): Likewise.
15716 (paradoxical_subreg_p): Delete out-of-line definition.
15718 2017-08-22 Jeff Law <law@redhat.com>
15720 PR tree-optimization/81741
15721 PR tree-optimization/71947
15722 * tree-ssa-dom.c: Include tree-inline.h.
15723 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
15724 equivalences if one is more expensive to compute than the other.
15725 * tree-ssa-scopedtables.h (class const_or_copies): Make
15726 record_const_or_copy_raw method private.
15727 (class avail_exprs_stack): New method simplify_binary_operation.
15728 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
15729 avail_exprs_stack::simplify_binary_operation as needed.
15730 (avail_exprs_stack::simplify_binary_operation): New function.
15732 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
15734 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
15735 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
15736 (DOT_SYMBOLS): Likewise.
15737 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
15738 (RELOCATABLE_NEEDS_FIXUP): Likewise.
15739 (RS6000_ABI_NAME): Likewise.
15740 (TARGET_CMODEL): Likewise.
15741 (TOC_SECTION_ASM_OP): Likewise.
15742 (SET_CMODEL): New macro.
15743 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
15745 2017-08-22 Richard Biener <rguenther@suse.de>
15747 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
15748 to allow for free-lang-data replacements similar to verify_type_variant.
15750 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
15751 Alan Hayward <alan.hayward@arm.com>
15752 David Sherwood <david.sherwood@arm.com>
15754 * config/aarch64/aarch64.md (casesi): Use DImode rather than
15755 VOIDmode for the LABEL_REF.
15757 2017-08-22 Richard Biener <rguenther@suse.de>
15759 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
15761 2017-08-22 Richard Biener <rguenther@suse.de>
15763 * common.opt (feliminate-dwarf2-dups): Ignore.
15764 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
15765 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
15766 same_die_p_wrap, compute_section_prefix,
15767 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
15768 (comdat_symbol_id, comdat_symbol_number): Likewise.
15769 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
15771 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
15772 (output_die): Mark unreachable path unreachable.
15773 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
15774 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
15775 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
15776 (dwarf2out_early_finish): Likewise.
15778 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
15780 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
15782 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
15785 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
15786 not VAR_P. Filter attribute warnings with OPT_Wattributes.
15787 (avr_attribute_table) <io, io_low, address>: Initialize
15788 .decl_required with true.
15790 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
15792 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
15793 undocumented debugging options.
15794 (-mvsx-scalar-double): Likewise.
15795 (-mallow-df-permute): Likewise.
15796 (-mvectorize-builtins): Likewise.
15797 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
15798 (rs6000_builtin_vectorized_function): Likewise.
15799 (rs6000_builtin_md_vectorized_function): Likewise.
15800 (rs6000_opt_vars): Likewise.
15802 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
15805 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
15806 (*btrq_imm): Rename from *btrq.
15807 (*btcq_imm): Rename from *btcq.
15808 (btsc): New code attribute.
15809 (*<btsc><mode>): New insn pattern.
15810 (*btr<mode>): Ditto.
15811 (*<btsc><mode>_mask): New insn_and_split pattern.
15812 (*btr<mode>_mask): Ditto.
15814 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15815 Alan Hayward <alan.hayward@arm.com>
15816 David Sherwood <david.sherwood@arm.com>
15818 * function.c (pad_below): Simplify padding calculation.
15820 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15821 Alan Hayward <alan.hayward@arm.com>
15822 David Sherwood <david.sherwood@arm.com>
15824 * target.def (function_prologue): Remove frame size argument.
15825 (function_epilogue): Likewise.
15826 * doc/tm.texi: Regenerate.
15827 * final.c (final_start_function): Update call to function_prologue.
15828 (final_end_function): Update call to function_epilogue.
15829 (default_function_pro_epilogue): Remove frame size argument.
15830 * output.h (default_function_pro_epilogue): Likewise.
15831 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
15832 (arm_output_function_prologue): Likewise.
15833 * config/frv/frv.c (frv_function_prologue): Likewise.
15834 (frv_function_epilogue): Likewise.
15835 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
15836 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
15837 (ia64_output_function_epilogue): Likewise.
15838 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
15839 (m32r_output_function_epilogue): Likewise.
15840 * config/microblaze/microblaze.c (microblaze_function_prologue)
15841 (microblaze_function_epilogue): Likewise.
15842 * config/mips/mips.c (mips_output_function_prologue): Likewise.
15843 (mips_output_function_epilogue): Likewise.
15844 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
15845 (mmix_target_asm_function_epilogue): Likewise.
15846 * config/msp430/msp430.c (msp430_start_function): Likewise.
15847 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
15848 (nds32_asm_function_epilogue): Likewise.
15849 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
15850 * config/pa/pa.c (pa_output_function_prologue): Likewise.
15851 (pa_output_function_epilogue): Likewise.
15852 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
15853 (rs6000_output_function_epilogue): Likewise.
15854 * config/rl78/rl78.c (rl78_start_function): Likewise.
15855 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
15856 (rs6000_output_function_epilogue): Likewise.
15857 * config/rx/rx.c (rx_output_function_prologue): Likewise.
15858 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
15859 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
15860 (sparc_asm_function_epilogue): Likewise.
15862 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15864 * tree.h (type_has_mode_precision_p): New function.
15865 * convert.c (convert_to_integer_1): Use it.
15866 * expr.c (expand_expr_real_2): Likewise.
15867 (expand_expr_real_1): Likewise.
15868 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
15869 * match.pd: Likewise.
15870 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
15871 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
15872 * tree-tailcall.c (process_assignment): Likewise.
15873 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15874 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
15875 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
15876 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15877 (vectorizable_assignment): Likewise.
15878 (vectorizable_shift): Likewise.
15879 (vectorizable_operation): Likewise.
15880 * tree-vrp.c (register_edge_assert_for_2): Likewise.
15882 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
15884 * match.pd: Add pow (C, x) simplification.
15886 2017-08-21 Richard Biener <rguenther@suse.de>
15888 PR tree-optimization/81900
15889 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
15890 for blocks with abnormal predecessors.
15891 (compute_antic): Do not set visited flag prematurely.
15893 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
15896 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
15898 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
15900 * stor-layout.h (vector_type_mode): Move to...
15901 * tree.h (vector_type_mode): ...here.
15902 * stor-layout.c (vector_type_mode): Move to...
15903 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
15905 2017-08-21 Richard Biener <rguenther@suse.de>
15907 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
15908 register_external_die hooks.
15909 (debug_false_tree_charstarstar_uhwistar): Declare.
15910 (debug_nothing_tree_charstar_uhwi): Likewise.
15911 * debug.c (do_nothing_debug_hooks): Adjust.
15912 (debug_false_tree_charstarstar_uhwistar): New do nothing.
15913 (debug_nothing_tree_charstar_uhwi): Likewise.
15914 * dbxout.c (dbx_debug_hooks): Adjust.
15915 (xcoff_debug_hooks): Likewise.
15916 * sdbout.c (sdb_debug_hooks): Likewise.
15917 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
15918 * dwarf2out.c (macinfo_label_base): New global.
15919 (dwarf2out_register_external_die): New function for the
15920 register_external_die hook.
15921 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
15922 (dwarf2_debug_hooks): Use them.
15923 (dwarf2_lineno_debug_hooks): Adjust.
15924 (struct die_struct): Add with_offset flag.
15925 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
15926 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
15927 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
15928 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
15929 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
15930 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
15931 defining section names for the early LTO debug variants.
15932 (reset_indirect_string): New helper.
15933 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
15934 (print_dw_val): Add support for offsetted symbol references.
15935 (get_ultimate_context): Split out from is_cxx.
15936 (is_cxx): Use get_ultimate_context.
15937 (is_fortran): Add decl overload.
15938 (compute_comp_unit_symbol): Split out worker from
15939 compute_section_prefix.
15940 (compute_section_prefix): Call compute_comp_unit_symbol and
15941 set comdat_type_p here.
15942 (output_die): Skip DIE symbol output for the LTO added one.
15943 Handle DIE symbol references with offset.
15944 (output_comp_unit): Guard section name mangling properly.
15945 For LTO debug sections emit a symbol at the section beginning
15946 which we use to refer to its DIEs.
15947 (add_abstract_origin_attribute): For DIEs registered via
15948 dwarf2out_register_external_die directly refer to the early
15949 DIE rather than indirectly through the shadow one we created.
15950 Remove obsolete call to dwarf2out_abstract_function for
15951 non-function/block origins.
15952 (gen_array_type_die): When generating early LTO debug do
15953 not emit DW_AT_string_length.
15954 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
15955 late when in LTO. As suggested place a gcc_unreachable for
15956 the DECL_ABSTRACT_P case.
15957 (gen_subprogram_die): Avoid another specification DIE
15958 for early built declarations/definitions for the late LTO case.
15959 (gen_variable_die): Add type references for late duplicated VLA dies
15961 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
15962 we have the abstract instance already.
15963 (process_scope_var): Adjust decl DIE contexts in LTO which
15964 first puts them in limbo.
15965 (gen_decl_die): Do not generate type DIEs late apart from
15966 types for VLAs or for decls we do not yet have a DIE. Do not
15967 call dwarf2out_abstract_function late.
15968 (dwarf2out_early_global_decl): Make sure to create DIEs
15969 for abstract instances of a decl first.
15970 (dwarf2out_late_global_decl): Adjust comment.
15971 (output_macinfo_op): With multiple macro sections use
15972 macinfo_label_base to distinguish labels.
15973 (output_macinfo): Likewise. Update macinfo_label_base.
15974 Pass in the line info label.
15975 (note_variable_value_in_expr): When generating LTO resolve
15976 all variable values here by generating DIEs as needed.
15977 (init_sections_and_labels): Add early LTO debug flag parameter
15978 and generate different sections and names if set. Add generation
15979 counter for the labels so we can have multiple of them.
15980 (reset_dies): Helper to allow DIEs to be output multiple times.
15981 (dwarf2out_finish): When outputting DIEs to the fat part of an
15982 LTO object first reset DIEs.
15983 (dwarf2out_early_finish): Output early DIEs when generating LTO.
15984 (modified_type_die): Check for decl_ultimate_origin being self
15986 (gen_type_die_with_usage): Likewise.
15987 (gen_typedef_die): Allow decl_ultimate_origin being self.
15988 (set_decl_abstract_flags): Remove.
15989 (set_block_abstract_flags): Likewise.
15990 (dwarf2out_abstract_function): Treat the early generated DIEs
15991 as the abstract copy and only add DW_AT_inline and
15992 DW_AT_artificial here and call set_decl_origin_self.
15993 If the DIE has an abstract origin don't do anything.
15994 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
15995 if we have none yet (Go fails to build one, PR78628).
15996 (variably_modified_type_p): Prevent endless recursion for Ada
15997 cyclic pointer types.
15998 * lto-streamer-in.c: Include debug.h.
15999 (dref_queue): New global.
16000 (lto_read_tree_1): Stream in DIE references.
16001 (lto_input_tree): Register DIE references.
16002 (input_function): Stream DECL_DEBUG_ARGS.
16003 * lto-streamer-out.c: Include debug.h.
16004 (lto_write_tree_1): Output DIE references.
16005 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
16006 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
16007 (output_function): Stream DECL_DEBUG_ARGS.
16008 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
16009 Stream DECL_ABSTRACT_ORIGIN.
16010 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
16011 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
16012 DECL_CONTEXT for file-scope decls.
16013 * lto-streamer.h (struct dref_entry): Declare.
16014 (dref_queue): Likewise.
16015 * cfgexpand.c (pass_expand::execute): Do not call the
16016 outlining_inline_function hook here.
16017 * lto-wrapper.c (debug_obj): New global.
16018 (tool_cleanup): Unlink it if required.
16019 (debug_objcopy): New function.
16020 (run_gcc): Handle early debug sections in the IL files by
16021 extracting them to separate files, partially linkin them and
16022 feeding the result back as result to the linker.
16023 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
16024 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
16025 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
16026 sections into a separate segment.
16027 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
16029 (darwin_asm_dwarf_section): Likewise.
16030 (darwin_asm_output_dwarf_offset): Likewise.
16031 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
16033 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16034 Alan Hayward <alan.hayward@arm.com>
16035 David Sherwood <david.sherwood@arm.com>
16037 * read-md.h (md_reader::record_potential_iterator_use): Replace
16038 pointer argument with an rtx and an index.
16039 * read-rtl.c (iterator_group::apply_iterator): Likewise.
16040 (apply_mode_iterator): Likewise.
16041 (apply_code_iterator): Likewise.
16042 (apply_int_iterator): Likewise.
16043 (apply_subst_iterator): Likewise.
16044 (record_iterator_use): Likewise.
16045 (record_attribute_use): Likewise.
16046 (md_reader::record_potential_iterator_use): Likewise. Update calls
16047 to record_iterator_use and apply_iterator.
16048 (iterator_use): Replace ptr with x and index.
16049 (attribute_use): Likewise.
16050 (apply_attribute_uses): Update calls to apply_iterator.
16051 (apply_iterators): Likewise. Update initialization of iterator_use.
16052 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
16053 and record_potential_iterator_use.
16054 (rtx_reader::read_rtx_operand): Likewise.
16056 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16057 Alan Hayward <alan.hayward@arm.com>
16058 David Sherwood <david.sherwood@arm.com>
16060 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
16063 2017-08-21 Richard Biener <rguenther@suse.de>
16065 PR middle-end/81884
16066 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
16067 at struct end conservatively when comparing common bases.
16069 2017-08-21 Richard Biener <rguenther@suse.de>
16071 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
16072 (mem_ref_in_stmt): Remove.
16073 (determine_max_movement): Use ref index to get at the reference.
16074 (invariantness_dom_walker::before_dom_children): Deal with
16075 lim data already initialized.
16076 (gather_mem_refs_stmt): Initialize lim data and record ref index.
16078 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
16080 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
16081 (TARGET_ISA_ROUND): Ditto.
16082 (TARGET_ROUND): Ditto.
16083 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
16084 * config/i386/i386.md: Ditto.
16085 * config/i386/sse.md: Ditto.
16086 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
16087 with OPTION_MASK_ISA_SSE4_1.
16089 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
16092 * doc/extend.texi (x86 Built-in Functions): Correct the name of
16093 __builtin_ia32_lzcnt_u16.
16095 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
16098 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
16099 (rs6000_set_current_function): Rewrite function to use it.
16101 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
16104 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
16105 and TYPE_WARN_IF_NOT_ALIGN.
16106 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
16107 (handle_warn_if_not_align): New.
16108 (place_union_field): Call handle_warn_if_not_align.
16109 (place_field): Call handle_warn_if_not_align.
16110 Copy TYPE_WARN_IF_NOT_ALIGN.
16111 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
16112 (layout_type): Likewise.
16113 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
16115 (tree_decl_common): Add warn_if_not_align.
16116 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
16117 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
16118 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
16119 (DECL_WARN_IF_NOT_ALIGN): Likewise.
16120 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
16121 * doc/extend.texi: Document warn_if_not_aligned attribute.
16122 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
16124 2017-08-17 Martin Liska <mliska@suse.cz>
16127 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
16128 (get_data_dependence): Use it as pointer type.
16129 (distribute_loop): Likewise.
16131 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16133 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
16134 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
16135 (p8_vmrgow_<mode>_direct): New define_insn.
16136 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
16137 handle endianness for vmrgew and vmrgow permute patterns.
16139 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
16141 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
16142 * config/rs6000/rs6000-cpus.def: Remove comment.
16143 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
16144 (POWERPC_MASKS): Likewise.
16145 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
16146 use of TARGET_VSX_TIMODE.
16147 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
16148 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
16149 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
16150 (rs6000_option_override_internal): Remove dead code.
16151 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
16152 (rs6000_legitimize_reload_address): Likewise.
16153 (rs6000_legitimate_address_p): Likewise.
16154 (rs6000_opt_masks): Delete "vsx-timode".
16155 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
16156 from function comment.
16157 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
16158 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
16159 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
16161 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
16162 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
16163 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
16164 (VSX_TI): Likewise.
16166 (define_peephole2): Likewise.
16168 2017-08-17 Martin Sebor <msebor@redhat.com>
16171 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
16172 past the end of an array.
16173 (test_pp_format): Add test cases.
16175 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
16177 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
16178 missing ECF_NOTHROW flags.
16180 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
16183 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
16184 operands residing in integer registers.
16185 (*vsx_le_perm_load_<mode>): Likewise.
16186 (*vsx_le_perm_store_<mode>): Likewise.
16187 (define_peephole2): Add peepholes to optimize the above.
16189 2017-08-17 Marek Polacek <polacek@redhat.com>
16191 PR middle-end/81814
16192 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
16193 to mimic what shorten_compare did. Change the return type to bool.
16194 (fold_cond_expr_with_comparison): Update call to
16195 operand_equal_for_comparison_p.
16196 (fold_ternary_loc): Likewise.
16198 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
16200 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
16202 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
16204 2017-08-17 Richard Biener <rguenther@suse.de>
16206 * tree-ssa-structalias.c (solve_graph): When propagating
16207 to successors update the graphs succ edges and avoid duplicate work.
16209 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
16212 * config/i386/i386.c (ix86_option_override_internal): Save target
16213 specific options after ix86_stack_protector_guard_reg was changed.
16215 2017-08-17 Richard Biener <rguenther@suse.de>
16217 PR tree-optimization/81827
16218 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
16219 (new_var_info): Initialize it conservatively.
16220 (get_call_vi): Mark register vars.
16221 (new_scalar_tmp_constraint_exp): Likewise.
16222 (handle_rhs_call): Likewise.
16223 (handle_const_call): Likewise.
16224 (create_function_info_for): Likewise.
16225 (solve_constraints): Sort varinfos to separate register from
16226 non-register vars to pack points-to solution bitmaps during
16229 2017-08-17 Marek Polacek <polacek@redhat.com>
16231 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
16233 2017-08-17 Richard Biener <rguenther@suse.de>
16235 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
16236 to true when overflow is undefined and we saturated the result.
16238 2017-08-17 Alan Modra <amodra@gmail.com>
16241 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
16242 Don't use store multiple if only one reg needs saving.
16243 (interesting_frame_related_regno): New function.
16244 (rs6000_frame_related): Don't emit frame info for regs that
16246 (rs6000_emit_epilogue): Likewise.
16248 2017-08-16 Nathan Sidwell <nathan@acm.org>
16250 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
16251 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
16252 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
16253 * tree.c (free_lang_data_in_type): Use else-if chain. Always
16254 clear TYPE_LANG_1. Remove obsolete member-function stripping.
16255 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
16256 (verify_type): Adjust for TYPE_BINFO move.
16257 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
16258 process TYPE_BINFO directly.
16259 (hash_tree): Likewise.
16260 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
16262 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
16265 2017-08-16 David Malcolm <dmalcolm@redhat.com>
16267 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
16269 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
16272 * config/i386/i386.md (*anddi_1_btr): Change predicates of
16273 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
16274 Add ix86_binary_operator_ok to insn constraint.
16275 (*iordi_1_bts): Ditto.
16276 (*xordi_1_btc): Ditto.
16277 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
16278 Update corresponding peephole2 pattern.
16282 2017-08-16 Bin Cheng <bin.cheng@arm.com>
16284 PR tree-optimization/81832
16285 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
16286 copy loop header which has IFN_LOOP_DIST_ALIAS call.
16288 2017-08-16 Marek Polacek <polacek@redhat.com>
16291 * fold-const.c (fold_indirect_ref_1): Restore original behavior
16292 regarding size_zero_node.
16294 2017-08-16 Martin Liska <mliska@suse.cz>
16297 * config.gcc: Respect previously set extra_objs in case
16300 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
16302 PR tree-optimization/81835
16303 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
16304 the phi SSA_NAME. Check that the condition in a COND_EXPR does
16305 not depend on the phi.
16307 2017-08-16 Alan Modra <amodra@gmail.com>
16309 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
16312 2017-08-16 Alan Modra <amodra@gmail.com>
16314 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
16315 (save_reg_p): ..into this. Update all callers.
16316 (first_reg_to_save): Simplify.
16318 2017-08-16 Alan Modra <amodra@gmail.com>
16320 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
16323 2017-08-15 Joseph Myers <joseph@codesourcery.com>
16327 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
16328 constant count if that count is less than 32.
16330 2017-08-15 Nathan Sidwell <nathan@acm.org>
16332 * gcc.c (execute): Emit friendlier message if inferior is killed
16333 by an external cause.
16335 2017-08-15 Richard Biener <rguenther@suse.de>
16337 PR tree-optimization/81790
16338 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
16339 CONSTRUCTORs from simplifying and VN.
16341 2017-08-14 Martin Sebor <msebor@redhat.com>
16343 * builtin-attrs.def: Add comments.
16345 2017-08-14 Martin Sebor <msebor@redhat.com>
16348 * doc/extend.texi (attribute nonstring): Document new attribute.
16350 2017-08-14 Martin Sebor <msebor@redhat.com>
16353 * tree-diagnostic.c (default_tree_printer): Handle %G.
16354 * gimple-pretty-print.h (percent_G_format): Declare new function.
16355 * gimple-pretty-print.c (percent_G_format): Define.
16356 * tree-pretty-print.c (percent_K_format): Add argument.
16358 2017-08-14 Martin Sebor <msebor@redhat.com>
16360 PR translation/79998
16361 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
16362 Remove a stray space.
16364 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
16367 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
16368 (*iordi_1_bts): Ditto.
16369 (*xordi_1_btc): Ditto.
16371 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16374 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
16375 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16377 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
16378 quoted strings, and make more translator-friendly.
16379 (darwin_rs6000_override_options): Likewise.
16380 (rs6000_option_override_internal): Likewise.
16381 (rs6000_return_in_memory): Fix overlong line.
16382 (init_cmulative_args): Use quoted strings, and make more
16383 translator-friendly.
16384 (rs6000_pass_by_reference): Fix overlong line.
16385 (def_builtin): Use quoted strings.
16386 (altivec_expand_predicate_builtin): Use quoted strings, and make
16387 more translator-friendly.
16388 (htm_expand_builtin): Use quoted strings.
16389 (cpu_expand_builtin): Use quoted strings, and make more
16390 translator-friendly.
16391 (altivec_expand_builtin): Likewise.
16392 (paired_expand_predicate_builtin): Likewise.
16393 (rs6000_invalid_builtin): Likewise.
16394 (builtin_function_type): Use quoted strings.
16395 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
16396 more translator-friendly.
16397 (rs6000_trampoline_init): Likewise.
16398 (rs6000_handle_altivec_attribute): Likewise.
16399 (rs6000_inner_target_options): Use quoted strings.
16400 (rs6000_disable_incompatible_switches): Likewise.
16401 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
16402 strings, and make more translator-friendly.
16403 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
16405 2017-08-14 Bin Cheng <bin.cheng@arm.com>
16407 PR tree-optimization/81799
16408 * tree-loop-distribution.c (version_loop_by_alias_check): Force
16409 cond_expr to simple gimple operand.
16411 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
16413 PR middle-end/46932
16414 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
16416 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
16420 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
16421 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
16422 TARGET_GASISR_PROLOGUES.
16423 * config/avr/avr.c (avr_option_override): Same.
16424 (avr_pass_pre_proep::execute): Same.
16426 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
16429 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
16430 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
16432 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
16434 * config/i386/i386.md (*load_tp_<mode>): Redefine as
16435 define_insn_and_split. Split to a memory load from 0 in
16436 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
16437 using PTR mode iterator.
16438 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
16439 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
16440 (*add_tp_<mode>): Redefine as define_insn_and_split.
16441 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
16442 address space. Merge with *add_tp_x32 using PTR mode iterator.
16443 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
16444 Split to an add with a memory load from 0 in
16445 DEFAULT_TLS_SEG_REG address space.
16447 2017-08-12 Andrew Pinski <apinski@cavium.com>
16449 * config/aarch64/aarch64-option-extensions.def (rdma):
16450 Fix feature string to what Linux prints out in /proc/cpuinfo.
16452 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
16455 * dwarf2out.c (modified_type_die): For C typedef types that have
16456 an ultimate origin, process the ultimate origin instead of the
16458 (gen_typedef_die): Assert that input DECLs have no ultimate
16460 (gen_type_die_with_usage): For typedef variants that have an
16461 ultimate origin, just call gen_decl_die on the original DECL.
16462 (process_scope_var): Avoid creating DIEs for local typedefs and
16463 concrete static variables.
16465 2017-08-12 Alan Modra <amodra@gmail.com>
16469 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
16470 match gnu-user.h startfile.
16471 (ENDFILE_LINUX_SPEC): Similarly.
16473 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
16476 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
16478 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
16480 2017-08-11 Tamar Christina <tamar.christina@arm.com>
16481 * config/aarch64/aarch64.md (mov<mode>): Change.
16482 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
16483 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
16484 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
16486 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
16488 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
16489 for storage order barriers.
16491 2017-08-11 Martin Liska <mliska@suse.cz>
16494 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
16495 variables of void type.
16497 2017-08-11 Martin Liska <mliska@suse.cz>
16499 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
16500 TARGET_SUPPORTS_ALIASES.
16501 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
16502 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
16503 (optimize_weakref): Likewise.
16504 * symtab.c (symtab_node::noninterposable_alias): Likewise.
16505 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
16506 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
16508 2017-08-11 Martin Liska <mliska@suse.cz>
16511 * config/i386/i386.c (make_resolver_func): Do complete
16512 refactoring of the function.
16514 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
16517 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
16518 * config/i386/i386.c (ix86_stack_protect_guard): Use
16519 ix86_stack_protect_guard_symbol_str to generate varible declaration.
16520 * doc/invoke.texi (x86 Options): Document
16521 -mstack-protector-guard-symbol= option.
16523 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
16525 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
16526 * config/i386/i386.c (ix86_split_stack_guard): New function.
16527 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
16528 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
16529 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
16530 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
16531 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
16532 (split_stack_space_check): Call ix86_split_stack_guard.
16534 2017-08-10 Martin Sebor <msebor@redhat.com>
16536 * print-tree.c (print_node): Print location using the established
16538 Replace spaces with colons.
16539 (debug_raw, debug): Ditto.
16541 2017-08-10 Martin Sebor <msebor@redhat.com>
16544 * pretty-print.c (pp_format): Correct the handling of %s precision.
16546 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
16549 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
16551 (ix86_finalize_stack_frame_flags): This. Also clear
16552 frame_pointer_needed if -fno-omit-frame-pointer is used without
16554 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
16555 with ix86_finalize_stack_frame_flags.
16556 (ix86_expand_epilogue): Likewise.
16557 (ix86_expand_split_stack_prologue): Likewise.
16558 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
16560 2017-08-10 Martin Liska <mliska@suse.cz>
16563 * c-attribs.c (handle_target_attribute):
16564 Report warning for an empty string argument of target attribute.
16566 2017-08-09 Jakub Jelinek <jakub@redhat.com>
16569 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
16571 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
16572 or DECL_NONLOCAL labels.
16573 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
16574 or DECL_NONLOCAL labels here.
16576 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
16578 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
16579 to indicate when early gimple folding has been disabled.
16580 (rs6000_gimple_fold_builtin): Add debug content.
16581 (rs6000_invalid_builtin): Fix whitespace.
16582 (rs6000_expand_builtin): Fix whitespace.
16583 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
16585 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
16588 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
16589 SAVE_MULTIPLE if not all the registers that saves, should be saved.
16591 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
16593 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
16594 (qdf24xx): Likewise.
16595 * config/aarch64/aarch64.md: Include falkor.md.
16596 * config/aarch64/falkor.md: New.
16598 2017-08-09 Marek Polacek <polacek@redhat.com>
16601 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
16602 * diagnostic.c (emit_diagnostic): Add a comment.
16603 (emit_diagnostic_valist): New function.
16605 2017-08-09 Marek Polacek <polacek@redhat.com>
16608 * input.c (make_location): New overload.
16609 * input.h (make_location): Declare.
16611 2017-08-08 Alan Modra <amodra@gmail.com>
16612 H.J. Lu <hongjiu.lu@intel.com>
16615 * gcc.c (NO_PIE_SPEC): Delete.
16616 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
16618 (LINK_PIE_SPEC): ..to here.
16619 (LINK_COMMAND_SPEC): Support -no-pie.
16620 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
16621 chain of crtbegin*.o selection, update for PIE_SPEC changes and
16623 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
16624 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
16625 (ENDFILE_CRTEND_SPEC): Similarly.
16627 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
16630 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
16631 (mstack-protector-guard-offset=): Ditto.
16632 * config/i386/i386.c (ix86_option_override): Handle
16633 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
16635 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
16636 ix86_stack_protect_guard_offset variables.
16637 (TARGET_STACK_PROTECT_GUARD): Always define.
16638 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
16639 and -mstack-protector-guard-offset= options.
16641 2017-08-08 Bin Cheng <bin.cheng@arm.com>
16643 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
16644 boundary case for the last candidate.
16646 2017-08-08 Bin Cheng <bin.cheng@arm.com>
16648 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
16649 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
16651 2017-08-08 Tamar Christina <tamar.christina@arm.com>
16653 PR middle-end/19706
16654 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
16655 * config/aarch64/aarch64-builtins.c
16656 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
16657 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
16658 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
16660 2017-08-08 Tamar Christina <tamar.christina@arm.com>
16661 Andrew Pinski <pinskia@gmail.com>
16663 PR middle-end/19706
16664 * internal-fn.def (XORSIGN): New.
16665 * optabs.def (xorsign_optab): New.
16666 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
16667 (convert_expand_mult_copysign): New.
16668 (pass_optimize_widening_mul::execute): Call
16669 convert_expand_mult_copysign.
16671 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16673 PR tree-optimization/81354
16674 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
16675 Insert on edges rather than explicitly creating landing pads.
16676 (analyze_candidates_and_replace): Commit edge inserts.
16678 2017-08-08 Richard Biener <rguenther@suse.de>
16680 PR middle-end/81719
16681 * tree-ssa-loop-niter.c: Include tree-dfa.h.
16682 (expand_simple_operations): Also look through ADDR_EXPRs with
16683 MEM_REF bases treating them as POINTER_PLUS_EXPR.
16685 2017-08-08 Richard Biener <rguenther@suse.de>
16687 PR tree-optimization/81723
16688 * tree-vect-slp.c (struct bst_traits): New hash traits.
16689 (bst_fail): New global.
16690 (vect_build_slp_tree_2): New worker, split out from ...
16691 (vect_build_slp_tree): ... this now wrapping it with using
16692 bst_fail set to cache SLP tree build fails. Properly handle
16694 (vect_analyze_slp_instance): Allocate and free bst_fail.
16696 2017-08-08 Martin Liska <mliska@suse.cz>
16699 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
16700 LABEL_DECLs that can be from a different function.
16702 2017-08-08 Bin Cheng <bin.cheng@arm.com>
16704 PR tree-optimization/81744
16705 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
16706 loop's number of iterations.
16708 2017-08-08 Martin Liska <mliska@suse.cz>
16710 * asan.c: Include header files.
16711 * attribs.c (build_decl_attribute_variant): New function moved
16713 (build_type_attribute_qual_variant): Likewise.
16714 (cmp_attrib_identifiers): Likewise.
16715 (simple_cst_list_equal): Likewise.
16716 (omp_declare_simd_clauses_equal): Likewise.
16717 (attribute_value_equal): Likewise.
16718 (comp_type_attributes): Likewise.
16719 (build_type_attribute_variant): Likewise.
16720 (lookup_ident_attribute): Likewise.
16721 (remove_attribute): Likewise.
16722 (merge_attributes): Likewise.
16723 (merge_type_attributes): Likewise.
16724 (merge_decl_attributes): Likewise.
16725 (merge_dllimport_decl_attributes): Likewise.
16726 (handle_dll_attribute): Likewise.
16727 (attribute_list_equal): Likewise.
16728 (attribute_list_contained): Likewise.
16729 * attribs.h (lookup_attribute): New function moved from tree.[ch].
16730 (lookup_attribute_by_prefix): Likewise.
16731 * bb-reorder.c: Include header files.
16732 * builtins.c: Likewise.
16733 * calls.c: Likewise.
16734 * cfgexpand.c: Likewise.
16735 * cgraph.c: Likewise.
16736 * cgraphunit.c: Likewise.
16737 * convert.c: Likewise.
16738 * dwarf2out.c: Likewise.
16739 * final.c: Likewise.
16740 * fold-const.c: Likewise.
16741 * function.c: Likewise.
16742 * gimple-expr.c: Likewise.
16743 * gimple-fold.c: Likewise.
16744 * gimple-pretty-print.c: Likewise.
16745 * gimple.c: Likewise.
16746 * gimplify.c: Likewise.
16747 * hsa-common.c: Likewise.
16748 * hsa-gen.c: Likewise.
16749 * internal-fn.c: Likewise.
16750 * ipa-chkp.c: Likewise.
16751 * ipa-cp.c: Likewise.
16752 * ipa-devirt.c: Likewise.
16753 * ipa-fnsummary.c: Likewise.
16754 * ipa-inline.c: Likewise.
16755 * ipa-visibility.c: Likewise.
16757 * lto-cgraph.c: Likewise.
16758 * omp-expand.c: Likewise.
16759 * omp-general.c: Likewise.
16760 * omp-low.c: Likewise.
16761 * omp-offload.c: Likewise.
16762 * omp-simd-clone.c: Likewise.
16763 * opts-global.c: Likewise.
16764 * passes.c: Likewise.
16765 * predict.c: Likewise.
16766 * sancov.c: Likewise.
16767 * sanopt.c: Likewise.
16768 * symtab.c: Likewise.
16769 * toplev.c: Likewise.
16770 * trans-mem.c: Likewise.
16771 * tree-chkp.c: Likewise.
16772 * tree-eh.c: Likewise.
16773 * tree-into-ssa.c: Likewise.
16774 * tree-object-size.c: Likewise.
16775 * tree-parloops.c: Likewise.
16776 * tree-profile.c: Likewise.
16777 * tree-ssa-ccp.c: Likewise.
16778 * tree-ssa-live.c: Likewise.
16779 * tree-ssa-loop.c: Likewise.
16780 * tree-ssa-sccvn.c: Likewise.
16781 * tree-ssa-structalias.c: Likewise.
16782 * tree-ssa.c: Likewise.
16783 * tree-streamer-in.c: Likewise.
16784 * tree-vectorizer.c: Likewise.
16785 * tree-vrp.c: Likewise.
16786 * tsan.c: Likewise.
16787 * ubsan.c: Likewise.
16788 * varasm.c: Likewise.
16789 * varpool.c: Likewise.
16790 * tree.c: Remove functions moved to attribs.[ch].
16791 * tree.h: Likewise.
16792 * config/aarch64/aarch64.c: Add attrs.h header file.
16793 * config/alpha/alpha.c: Likewise.
16794 * config/arc/arc.c: Likewise.
16795 * config/arm/arm.c: Likewise.
16796 * config/avr/avr.c: Likewise.
16797 * config/bfin/bfin.c: Likewise.
16798 * config/c6x/c6x.c: Likewise.
16799 * config/cr16/cr16.c: Likewise.
16800 * config/cris/cris.c: Likewise.
16801 * config/darwin.c: Likewise.
16802 * config/epiphany/epiphany.c: Likewise.
16803 * config/fr30/fr30.c: Likewise.
16804 * config/frv/frv.c: Likewise.
16805 * config/ft32/ft32.c: Likewise.
16806 * config/h8300/h8300.c: Likewise.
16807 * config/i386/winnt.c: Likewise.
16808 * config/ia64/ia64.c: Likewise.
16809 * config/iq2000/iq2000.c: Likewise.
16810 * config/lm32/lm32.c: Likewise.
16811 * config/m32c/m32c.c: Likewise.
16812 * config/m32r/m32r.c: Likewise.
16813 * config/m68k/m68k.c: Likewise.
16814 * config/mcore/mcore.c: Likewise.
16815 * config/microblaze/microblaze.c: Likewise.
16816 * config/mips/mips.c: Likewise.
16817 * config/mmix/mmix.c: Likewise.
16818 * config/mn10300/mn10300.c: Likewise.
16819 * config/moxie/moxie.c: Likewise.
16820 * config/msp430/msp430.c: Likewise.
16821 * config/nds32/nds32-isr.c: Likewise.
16822 * config/nds32/nds32.c: Likewise.
16823 * config/nios2/nios2.c: Likewise.
16824 * config/nvptx/nvptx.c: Likewise.
16825 * config/pa/pa.c: Likewise.
16826 * config/pdp11/pdp11.c: Likewise.
16827 * config/powerpcspe/powerpcspe.c: Likewise.
16828 * config/riscv/riscv.c: Likewise.
16829 * config/rl78/rl78.c: Likewise.
16830 * config/rx/rx.c: Likewise.
16831 * config/s390/s390.c: Likewise.
16832 * config/sh/sh.c: Likewise.
16833 * config/sol2.c: Likewise.
16834 * config/sparc/sparc.c: Likewise.
16835 * config/spu/spu.c: Likewise.
16836 * config/stormy16/stormy16.c: Likewise.
16837 * config/tilegx/tilegx.c: Likewise.
16838 * config/tilepro/tilepro.c: Likewise.
16839 * config/v850/v850.c: Likewise.
16840 * config/vax/vax.c: Likewise.
16841 * config/visium/visium.c: Likewise.
16842 * config/xtensa/xtensa.c: Likewise.
16844 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16847 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
16848 constraints since the -mupper-regs-* switches have been
16850 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
16851 into a vector from a double word element that was extracted from
16852 another vector, and eliminate extra XXPERMDI instructions.
16853 (vsx_concat_<mode>_2): Likewise.
16854 (vsx_concat_<mode>_3): Likewise.
16855 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
16856 concat to allow optimizing inserts from previous extracts.
16858 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
16860 * config/i386/i386.c (ix86_stack_protect_guard): Generate
16861 memory reference to a SSP offset in TLS address space.
16862 (ix86_print_operand) <case '@'>: Remove.
16863 (ix86_print_operand_punct_valid_p): Remove '@' code.
16864 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
16865 UNSPEC_SP_TLS_TEST.
16866 (stack_tls_protect_set_<mode>): Remove.
16867 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
16868 (stack_tls_protect_test_<mode>): Remove.
16869 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
16871 2017-08-07 Olivier Hainque <hainque@adacore.com>
16874 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
16876 2017-08-07 Douglas Rupp <rupp@adacore.com>
16878 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
16879 variable was referenced as multidir in command.
16881 2017-08-07 Jakub Jelinek <jakub@redhat.com>
16884 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
16887 2017-08-07 Martin Liska <mliska@suse.cz>
16889 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
16890 * config/rl78/rl78.c: Add include of attribs.h.
16891 * config/sh/sh.c: Likewise.
16892 * config/v850/v850.c: Likewise.
16894 2017-08-07 Tom de Vries <tom@codesourcery.com>
16896 PR middle-end/78266
16897 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
16899 2017-08-07 Martin Liska <mliska@suse.cz>
16901 * config/mips/mips.c: Include attribs.h.
16903 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
16906 * doc/invoke.texi: Document change in behvaior for -Ofast for
16909 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
16911 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
16913 (aarch64_pop_regs): Likewise.
16914 (aarch64_gen_load_pair): Likewise.
16915 (aarch64_save_callee_saves): Likewise.
16916 (aarch64_restore_callee_saves): Likewise.
16918 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
16920 * config/i386/i386.c: Revert the last change.
16922 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
16925 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
16927 (ix86_finalize_stack_frame_flags): This. Also clear
16928 frame_pointer_needed if -fno-omit-frame-pointer is used without
16930 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
16931 with ix86_finalize_stack_frame_flags.
16932 (ix86_expand_epilogue): Likewise.
16933 (ix86_expand_split_stack_prologue): Likewise.
16935 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
16938 * config/i386/i386.c (get_builtin_code_for_version): Set priority
16939 to P_AES for Westmere.
16941 2017-08-07 Jonathan Yong <10walls@gmail.com>
16943 * config/i386/mingw.opt (fset-stack-executable): Removed.
16944 * config/i386/cygming.opt (fset-stack-executable): Moved
16946 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
16948 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
16950 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
16952 2017-08-07 Marek Polacek <polacek@redhat.com>
16954 PR middle-end/81737
16955 * fold-const.c (fold_indirect_ref_1): Check type_domain.
16957 2017-08-07 Martin Liska <mliska@suse.cz>
16959 * attribs.h (canonicalize_attr_name): New function.
16960 (cmp_attribs): Move from c-format.c and adjusted.
16961 (is_attribute_p): Moved from tree.h.
16962 * tree-inline.c: Add new includes.
16963 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
16964 (private_is_attribute_p): Remove.
16965 (private_lookup_attribute): Likewise.
16966 (private_lookup_attribute_by_prefix): Simplify.
16967 (remove_attribute): Use is_attribute_p.
16968 * tree.h: Remove removed declarations.
16970 2017-08-07 Jakub Jelinek <jakub@redhat.com>
16972 PR middle-end/81698
16973 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
16974 instead of computing it in the function. Formatting fix.
16975 (expand_case): Don't rely on default_edge being the first edge,
16976 clear it if removing it, pass default_edge to
16977 emit_case_dispatch_table.
16978 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
16981 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
16983 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
16984 insn in the function, emit NOP after the insn.
16986 2017-08-06 Tom de Vries <tom@codesourcery.com>
16988 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
16991 2017-08-06 Tom de Vries <tom@codesourcery.com>
16993 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
16996 2017-08-04 Yury Gribov <tetra2005@gmail.com>
16998 PR tree-optimization/57371
16999 * match.pd: New pattern.
17001 2017-08-04 Marek Polacek <polacek@redhat.com>
17003 PR middle-end/81695
17004 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
17005 perform the computation in offset_int.
17007 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
17009 PR tree-optimization/81136
17010 * tree-vectorizer.h: Include tree-hash-traits.h.
17011 (vec_base_alignments): New typedef.
17012 (vec_info): Add a base_alignments field.
17013 (vect_record_base_alignments): Declare.
17014 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
17016 (DR_IS_CONDITIONAL_IN_STMT): New macro.
17017 (create_data_ref): Add an is_conditional_in_stmt argument.
17018 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
17019 the is_conditional_in_stmt field.
17020 (data_ref_loc): Add an is_conditional_in_stmt field.
17021 (get_references_in_stmt): Set the is_conditional_in_stmt field.
17022 (find_data_references_in_stmt): Update call to create_data_ref.
17023 (graphite_find_data_references_in_stmt): Likewise.
17024 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
17025 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
17026 (vect_record_base_alignment): New function.
17027 (vect_record_base_alignments): Likewise.
17028 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
17029 for nested statements even if we fail to compute a misalignment.
17030 Use pooled base alignments for unconditional references.
17031 (vect_find_same_alignment_drs): Compare base addresses instead
17033 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
17034 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
17036 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
17038 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
17039 Add an explicit name for the enum. Use auto_vec for slp_instances
17040 and grouped_stores.
17041 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
17043 (_bb_vec_info): Add a constructor and destructor.
17044 (vinfo_for_stmt): Return NULL for uids of -1 as well.
17045 (destroy_loop_vec_info): Delete.
17046 (vect_destroy_datarefs): Likewise.
17047 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
17048 (vec_info::vec_info): New function.
17049 (vec_info::~vec_info): Likewise.
17050 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
17051 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
17052 destroy_loop_vec_info.
17053 * tree-vect-loop.c (new_loop_vec_info): Replace with...
17054 (_loop_vec_info::_loop_vec_info): ...this.
17055 (destroy_loop_vec_info): Replace with...
17056 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
17057 the stmt_vec_infos. Leave handling of vec_info information to its
17058 destructor. Remove explicit vector releases.
17059 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
17060 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
17061 * tree-vect-slp.c (new_bb_vec_info): Replace with...
17062 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
17063 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
17064 (destroy_bb_vec_info): Replace with...
17065 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
17066 information to its destructor.
17067 (vect_slp_analyze_bb_1): Use new and delete instead of
17068 new_bb_vec_info and destroy_bb_vec_info.
17069 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
17072 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
17074 * tree-data-ref.h (subscript): Add access_fn field.
17075 (data_dependence_relation): Add could_be_independent_p.
17076 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
17077 (same_access_functions): Move to tree-data-ref.c.
17078 * tree-data-ref.c (ref_contains_union_access_p): New function.
17079 (access_fn_component_p): Likewise.
17080 (access_fn_components_comparable_p): Likewise.
17081 (dr_analyze_indices): Add a reference to access_fn_component_p.
17082 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
17084 (constant_access_functions): Likewise.
17085 (add_other_self_distances): Likewise.
17086 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
17087 (initialize_data_dependence_relation): Use XCNEW and remove
17088 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
17089 of access functions that have the same type. Allow the
17090 subsequence to end with different bases in some circumstances.
17091 Record the chosen access functions in SUB_ACCESS_FN.
17092 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
17093 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
17094 (subscript_dependence_tester_1): Likewise dra and drb.
17095 (build_classic_dist_vector): Update calls accordingly.
17096 (subscript_dependence_tester): Likewise.
17097 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
17098 DDR_COULD_BE_INDEPENDENT_P.
17099 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
17100 comp_alias_ddrs instead of may_alias_ddrs.
17101 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
17103 (vect_analyze_data_ref_dependence): Use it if
17104 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
17105 distance vectors if that fails.
17106 (dependence_distance_ge_vf): New function.
17107 (vect_prune_runtime_alias_test_list): Use it. Don't clear
17108 LOOP_VINFO_MAY_ALIAS_DDRS.
17110 2017-08-04 Richard Biener <rguenther@suse.de>
17112 PR middle-end/81705
17113 * fold-const.c (fold_binary_loc): Properly restrict
17114 minus_var0 && minus_var1 case when associating undefined overflow
17117 2017-08-04 Tom de Vries <tom@codesourcery.com>
17119 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
17121 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17123 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17124 Don't start diagnostic messages with a capital letter.
17125 * config/rs6000/rs6000.c (rs6000_option_override_internal):
17127 (rs6000_invalid_builtin): Likewise.
17128 (rs6000_trampoline_init): Likewise.
17130 2017-08-03 Jakub Jelinek <jakub@redhat.com>
17133 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
17134 after setting changeable df flags.
17136 2017-08-03 Richard Biener <rguenther@suse.de>
17138 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
17139 up if the use is in USE - X.
17141 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
17143 * toplev.c (dumpfile.h): New include.
17144 (internal_error_reentered): New static function. Use it...
17145 (internal_error_function): ...here to handle reentered internal_error.
17147 2017-08-03 Richard Biener <rguenther@suse.de>
17149 PR middle-end/81148
17150 * fold-const.c (split_tree): Add minus_var and minus_con
17151 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
17152 here but always use minus_*.
17153 (associate_trees): Assert we never associate with MINUS_EXPR
17154 and NULL first operand. Do not recurse for PLUS_EXPR operands
17155 when associating as MINUS_EXPR either.
17156 (fold_binary_loc): Track minus_var and minus_con.
17158 2017-08-03 Tom de Vries <tom@codesourcery.com>
17161 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
17162 ACCEL_COMPILER, apply finish_options on
17163 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
17165 2017-08-03 Tom de Vries <tom@codesourcery.com>
17168 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
17169 function_entry_patch_area_size > 0.
17171 2017-08-03 Jakub Jelinek <jakub@redhat.com>
17174 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
17175 instead of 10??LU, perform unit multiplication in wide_int,
17176 don't change alloc_object_size_limit if the limit is larger
17179 PR tree-optimization/81655
17180 PR tree-optimization/81588
17181 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
17182 the case when ranges[i].low and high are 1 for unsigned type with
17185 PR middle-end/81052
17186 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
17187 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
17189 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17191 * tree-vrp.h: Add include guard.
17193 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
17196 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
17197 (ud2): New insn pattern.
17198 * config/i386/i386.c (ix86_expand_epilogue):
17199 For naked functions, generate ud2 instead of trap insn.
17201 2017-08-02 Marek Polacek <polacek@redhat.com>
17204 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
17206 2017-08-02 Tom de Vries <tom@codesourcery.com>
17207 Cesar Philippidis <cesar@codesourcery.com>
17209 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
17210 Add missing edge probabilities.
17212 2017-08-02 Tamar Christina <tamar.christina@arm.com>
17214 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
17215 Correct endianness.
17217 2017-08-02 Jakub Jelinek <jakub@redhat.com>
17219 PR middle-end/79499
17220 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
17221 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
17222 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
17223 prologue_seq sequences - if any.
17225 2017-08-02 Richard Biener <rguenther@suse.de>
17227 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
17228 via vectors if supported, integer extracts via punning if supported
17229 or otherwise vector extracts.
17231 2017-08-02 Richard Biener <rguenther@suse.de>
17233 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
17235 (bitmap_insert_into_set): ... this.
17237 2017-08-02 Richard Biener <rguenther@suse.de>
17239 PR tree-optimization/81633
17241 2015-08-17 Alan Hayward <alan.hayward@arm.com>
17243 PR tree-optimization/71752
17244 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
17246 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
17248 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
17249 (machine_function::call_ms2sysv_pad_out): Remove field.
17250 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
17251 (ix86_compute_frame_layout): Likewise.
17253 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
17256 * config/i386/i386.c (ix86_set_func_type): Disallow naked
17257 attribute with interrupt attribute.
17259 2017-08-01 Andrew Pinski <apinski@cavium.com>
17261 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
17262 BIT_INSERT_EXPR's operand 1
17263 to see if the types precision matches.
17265 2017-08-01 Martin Liska <mliska@suse.cz>
17267 PR middle-end/70140
17268 * builtins.c (expand_builtin_memcpy_args): Remove.
17269 (expand_builtin_memcpy): Call newly added function
17270 expand_builtin_memory_copy_args.
17271 (expand_builtin_memcpy_with_bounds): Likewise.
17272 (expand_builtin_mempcpy): Remove last argument.
17273 (expand_builtin_mempcpy_with_bounds): Likewise.
17274 (expand_builtin_memory_copy_args): New function created from
17275 expand_builtin_mempcpy_args with small modifications.
17276 (expand_builtin_mempcpy_args): Remove.
17277 (expand_builtin_stpcpy): Remove unused argument.
17278 (expand_builtin): Likewise.
17279 (expand_builtin_with_bounds): Likewise.
17281 2017-08-01 Martin Liska <mliska@suse.cz>
17284 Make mempcpy more optimal (PR middle-end/70140).
17286 2017-08-01 Jakub Jelinek <jakub@redhat.com>
17289 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
17290 __builtin_vec_cmpne verify both arguments are compatible vectors
17291 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
17292 verify arg1_type is a pointer or array type. For __builtin_vec_st,
17293 move computation of aligned to after checking the argument types.
17297 * config/rs6000/vsx.md (vextract_fp_from_shorth,
17298 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
17301 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
17302 Doug Rupp <rupp@adacore.com>
17303 Olivier Hainque <hainque@adacore.com>
17305 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
17306 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
17308 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
17309 for TARGET_OS_CPP_BUILTIN.
17310 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
17311 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
17313 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
17314 passing required abi options to the assembler for EABI configurations.
17315 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
17316 of .text.hot and .text.unlikely sections for kernel modules when
17317 using ARM style exceptions.
17318 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
17319 options. Add EXTRA_CC1_SPEC.
17320 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
17322 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
17324 (ARM_TARGET2_DWARF_FORMAT): Define.
17325 * config/arm/t-vxworks: Adjust multilib control to removal of the
17326 Diab command line options.
17328 2017-08-01 Martin Liska <mliska@suse.cz>
17330 PR gcov-profile/81561
17331 * gcov.c (unblock): Make unblocking safe as we need to preserve
17332 index correspondence of blocks and block_lists.
17334 2017-08-01 Richard Biener <rguenther@suse.de>
17336 PR tree-optimization/81181
17337 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
17338 (compute_antic): ... end of iteration here.
17340 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
17342 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
17343 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
17344 (ftree-slp-vectorize): Likewise.
17345 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
17346 can no longer be set independent of flag_tree_loop_vectorize.
17347 * omp-general.c (emp_max_vf): Likewise.
17348 * opts.c (enable_fdo_optimizations): Remove references to
17349 flag_tree_vectorize, these are now implicit.
17350 (common_handle_option): Remove handling for OPT_ftree_vectorize,
17351 and leave it for the options machinery.
17353 2017-08-01 Martin Liska <mliska@suse.cz>
17355 PR middle-end/70140
17356 * builtins.c (expand_builtin_memcpy_args): Remove.
17357 (expand_builtin_memcpy): Call newly added function
17358 expand_builtin_memory_copy_args.
17359 (expand_builtin_memcpy_with_bounds): Likewise.
17360 (expand_builtin_mempcpy): Remove last argument.
17361 (expand_builtin_mempcpy_with_bounds): Likewise.
17362 (expand_builtin_memory_copy_args): New function created from
17363 expand_builtin_mempcpy_args with small modifications.
17364 (expand_builtin_mempcpy_args): Remove.
17365 (expand_builtin_stpcpy): Remove unused argument.
17366 (expand_builtin): Likewise.
17367 (expand_builtin_with_bounds): Likewise.
17369 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
17372 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
17373 print "ds:" only for immediates in generic address space.
17375 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
17378 * config/i386/i386.c (ix86_funciton_naked): New prototype.
17379 (ix86_function_ok_for_sibcall): Return false for naked functions.
17381 2017-08-01 Richard Biener <rguenther@suse.de>
17383 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
17384 (compute_antic): Seed worklist with exit block predecessors.
17385 * cfganal.c (dfs_find_deadend): For a cycle return the source
17386 of the edge closing it.
17388 2017-08-01 Tamar Christina <tamar.christina@arm.com>
17390 * config/aarch64/aarch64.c
17391 (aarch64_can_const_movi_rtx_p): Move 0 check.
17393 2017-08-01 Bin Cheng <bin.cheng@arm.com>
17395 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
17396 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
17398 * match.pd: Ditto in address comparison pattern.
17400 2017-08-01 Bin Cheng <bin.cheng@arm.com>
17402 PR tree-optimization/81627
17403 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
17404 closed ssa form for store-store chain.
17406 2017-08-01 Bin Cheng <bin.cheng@arm.com>
17408 PR tree-optimization/81620
17409 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
17410 for store-store chain.
17412 2017-08-01 Jakub Jelinek <jakub@redhat.com>
17414 PR tree-optimization/81588
17415 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
17416 ranges[i].in_p, invert comparison code ccode. For >/>=,
17417 swap rhs1 and rhs2 and comparison code unconditionally,
17418 for </<= don't do that. Don't swap rhs1/rhs2 again if
17419 ranges[i].in_p, instead invert comparison code ccode if
17420 opcode or oe->rank is BIT_IOR_EXPR.
17423 * optabs.def (vec_extract_optab, vec_init_optab): Change from
17424 a direct optab to conversion optab.
17425 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
17426 with GET_MODE_INNER as last argument instead of optab_handler.
17427 * expmed.c (extract_bit_field_1): Likewise. Use vector from
17428 vector extraction if possible and optab is available.
17429 * expr.c (store_constructor): Use convert_optab_handler instead
17430 of optab_handler. Use vector initialization from smaller
17431 vectors if possible and optab is available.
17432 * tree-vect-stmts.c (vectorizable_load): Likewise.
17433 * doc/md.texi (vec_extract, vec_init): Document that the optabs
17434 now have two modes.
17435 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
17436 of vec_init from half-sized vectors with the same element mode.
17437 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
17438 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
17439 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
17440 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
17441 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
17442 after mode in gen_vec_extract* calls.
17443 (vec_extract<mode>): Renamed to ...
17444 (vec_extract<mode><ssescalarmodelower>): ... this.
17445 (vec_extract<mode><ssehalfvecmodelower>): New expander.
17446 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
17447 element mode after mode in gen_vec_init* calls.
17448 (VEC_INIT_HALF_MODE): New mode iterator.
17449 (vec_init<mode>): Renamed to ...
17450 (vec_init<mode><ssescalarmodelower>): ... this.
17451 (vec_init<mode><ssehalfvecmodelower>): New expander.
17452 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
17453 (vec_extractv2sfsf): ... this.
17454 (vec_initv2sf): Renamed to ...
17455 (vec_initv2sfsf): ... this.
17456 (vec_extractv2si): Renamed to ...
17457 (vec_extractv2sisi): ... this.
17458 (vec_initv2si): Renamed to ...
17459 (vec_initv2sisi): ... this.
17460 (vec_extractv4hi): Renamed to ...
17461 (vec_extractv4hihi): ... this.
17462 (vec_initv4hi): Renamed to ...
17463 (vec_initv4hihi): ... this.
17464 (vec_extractv8qi): Renamed to ...
17465 (vec_extractv8qiqi): ... this.
17466 (vec_initv8qi): Renamed to ...
17467 (vec_initv8qiqi): ... this.
17468 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
17469 (vec_init<mode>): Renamed to ...
17470 (vec_init<mode><VEC_base_l>): ... this.
17471 (vec_extract<mode>): Renamed to ...
17472 (vec_extract<mode><VEC_base_l>): ... this.
17473 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
17474 (vec_initv2sfsf): ... this.
17475 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
17476 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
17477 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
17478 element mode after mode in gen_vec_init* calls.
17479 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
17480 (vec_init<mode><Vel>): ... this.
17481 (vec_extract<mode>): Renamed to ...
17482 (vec_extract<mode><Vel>): ... this.
17483 * config/aarch64/iterators.md (Vel): New mode attribute.
17484 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
17485 Add element mode after mode in gen_vec_extract* calls.
17486 * config/s390/vector.md (non_vec_l): New mode attribute.
17487 (vec_extract<mode>): Renamed to ...
17488 (vec_extract<mode><non_vec_l>): ... this.
17489 (vec_init<mode>): Renamed to ...
17490 (vec_init<mode><non_vec_l>): ... this.
17491 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
17492 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
17494 * config/arm/iterators.md (V_elem_l): New mode attribute.
17495 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
17496 (vec_extract<mode><V_elem_l>): ... this.
17497 (vec_extractv2di): Renamed to ...
17498 (vec_extractv2didi): ... this.
17499 (vec_init<mode>): Renamed to ...
17500 (vec_init<mode><V_elem_l>): ... this.
17501 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
17502 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
17503 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
17504 Add element mode after gen_vec_extract* calls.
17505 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
17506 (vec_init<mode><unitmode>): ... this.
17507 (vec_extract<mode>): Renamed to ...
17508 (vec_extract<mode><unitmode>): ... this.
17509 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
17510 (vec_init<mode><unitmode>): ... this.
17511 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
17512 (vec_initv2sfsf): ... this.
17513 (vec_extractv2sf): Renamed to ...
17514 (vec_extractv2sfsf): ... this.
17515 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
17516 Add element mode after gen_vec_extract* calls.
17517 * config/mips/mips.md (unitmode): New mode iterator.
17518 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
17519 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
17520 * config/spu/spu.md (inner_l): New mode attribute.
17521 (vec_init<mode>): Renamed to ...
17522 (vec_init<mode><inner_l>): ... this.
17523 (vec_extract<mode>): Renamed to ...
17524 (vec_extract<mode><inner_l>): ... this.
17525 * config/sparc/sparc.md (veltmode): New mode iterator.
17526 (vec_init<VMALL:mode>): Renamed to ...
17527 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
17528 * config/ia64/vect.md (vec_initv2si): Renamed to ...
17529 (vec_initv2sisi): ... this.
17530 (vec_initv2sf): Renamed to ...
17531 (vec_initv2sfsf): ... this.
17532 (vec_extractv2sf): Renamed to ...
17533 (vec_extractv2sfsf): ... this.
17534 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
17535 (vec_init<mode>): Renamed to ...
17536 (vec_init<mode><VEC_base_l>): ... this.
17537 (vec_extract<mode>): Renamed to ...
17538 (vec_extract<mode><VEC_base_l>): ... this.
17539 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
17540 (vec_initv2sfsf): ... this.
17541 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
17542 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
17543 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
17544 gen_vec_init* calls.
17546 2017-08-01 Richard Biener <rguenther@suse.de>
17548 PR tree-optimization/81297
17549 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
17550 TREE_OVERFLOW from INTEGER_CSTs.
17552 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
17554 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
17556 2017-07-31 Carl Love <cel@us.ibm.com>
17558 * config/rs6000/rs6000-c: Add support for built-in functions
17559 vector signed char vec_xl_be (signed long long, signed char *);
17560 vector unsigned char vec_xl_be (signed long long, unsigned char *);
17561 vector signed int vec_xl_be (signed long long, signed int *);
17562 vector unsigned int vec_xl_be (signed long long, unsigned int *);
17563 vector signed long long vec_xl_be (signed long long, signed long long *);
17564 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
17565 vector signed short vec_xl_be (signed long long, signed short *);
17566 vector unsigned short vec_xl_be (signed long long, unsigned short *);
17567 vector double vec_xl_be (signed long long, double *);
17568 vector float vec_xl_be (signed long long, float *);
17569 * config/rs6000/altivec.h (vec_xl_be): Add #define.
17570 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
17571 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
17573 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
17574 (altivec_expand_builtin): Add switch statement to call
17575 altivec_expand_xl_be for each builtin.
17576 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
17577 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
17578 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
17579 __builtin_vsx_le_be_v16qi.
17580 * doc/extend.texi: Update the built-in documentation file for the
17581 new built-in functions.
17583 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
17586 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
17588 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17590 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17592 * config.gcc: Add z14.
17593 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
17594 CPU model numbers for z13s and z14.
17595 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
17597 * config/s390/s390-opts.h (enum processor_type): Rename
17598 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
17599 * config/s390/s390.c (processor_table): Add field for CPU name to
17600 be passed to Binutils.
17601 (s390_asm_output_machine_for_arch): Use the new field in
17602 processor_table for Binutils.
17603 (s390_expand_builtin): Replace arch12 with z14.
17604 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
17605 (s390_get_sched_attrmask): Likewise.
17606 (s390_get_unit_mask): Likewise.
17607 * config/s390/s390.opt: Add z14 to processor_type enum.
17609 2017-07-31 Martin Jambor <mjambor@suse.cz>
17612 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
17613 regardless of optimization level.
17615 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
17616 Martin Liska <mliska@suse.cz>
17618 * predict.def: Remove old comment and adjust probability.
17619 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
17620 PREDICT statements.
17622 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
17625 * config/i386/i386.c (ix86_function_naked): New function.
17626 (ix86_can_use_return_insn_p): Return false for naked functions.
17627 (ix86_expand_prologue): Skip prologue for naked functions.
17628 (ix86_expand_epilogue): Skip epilogue for naked functions
17629 and emit trap instruction.
17630 (ix86_warn_func_return): New function.
17631 (ix86_attribute_table): Add "naked" attribute specification.
17632 (TARGET_WARN_FUNC_RETURN): Define.
17633 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
17635 2017-07-31 Martin Liska <mliska@suse.cz>
17637 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
17638 (dump_gimple_bb_header): Always dump BB info.
17639 (pp_cfg_jump): Do not append info about BB when dumping a jump.
17641 2017-07-31 Martin Liska <mliska@suse.cz>
17644 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
17645 also with current_function_decl non-null equality.
17647 2017-07-31 Jakub Jelinek <jakub@redhat.com>
17650 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
17651 change type to the element type, instead add eltype variable and
17652 use it where we are interested in the element type.
17654 PR tree-optimization/81603
17655 * ipa-polymorphic-call.c
17656 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
17657 offset arithmetic in offset_int, bail out if the resulting bit offset
17658 doesn't fit into shwi.
17660 2017-07-31 Martin Liska <mliska@suse.cz>
17662 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
17663 (gimplify_save_expr): Fix comment.
17665 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
17668 * config/i386/i386.c (ix86_function_arg): Update arguments for
17670 (ix86_compute_frame_layout): Set the initial stack offset to
17671 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
17672 INCOMING_FRAME_SP_OFFSET.
17673 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
17674 stack before exception handler returns.
17675 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
17676 the 'ERROR_CODE' for exception handler.
17678 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
17680 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
17681 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
17682 (ASM_OUTPUT_REG_POP): Ditto.
17683 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
17684 instead of asm_fprintf to output pure string.
17686 2017-07-29 Jakub Jelinek <jakub@redhat.com>
17688 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
17689 to imported_module_or_decl hook.
17690 (debug_nothing_tree_tree_tree_bool): Remove.
17691 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
17692 * debug.c (do_nothing_debug_hooks): Use
17693 debug_nothing_tree_tree_tree_bool_bool instead of
17694 debug_nothing_tree_tree_tree_bool.
17695 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
17696 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
17697 * sdbout.c (sdb_debug_hooks): Likewise.
17698 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
17699 (gen_namespace_die): Add DW_AT_export_symbols attribute if
17701 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
17702 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
17703 attribute, don't add anything.
17705 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17707 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
17708 (fold_build2_stat_loc): Likewise.
17709 (fold_build3_stat_loc): Likewise.
17710 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
17711 (fold_build1_loc): Remove macro.
17712 (fold_build2_loc): Likewise.
17713 (fold_build3_loc): Likewise.
17715 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17717 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
17718 (gimple_build_debug_bind_source_stat): Likewise.
17719 * gimple.h (gimple_build_debug_bind): Remove macro.
17720 (gimple_build_debug_bind_source): Likewise.
17722 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17724 * bitmap.c (bitmap_alloc): Adjust.
17725 (bitmap_gc_alloc): Likewise.
17726 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
17728 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17730 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
17731 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
17732 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
17733 (bitmap_gc_alloc_stat): Likewise.
17734 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
17736 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17738 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
17739 * rtl.h (shallow_copy_rtx): Remove macro.
17741 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17743 * emit-rtl.c (gen_raw_REG): Adjust.
17744 * gengenrtl.c (gendef): Likewise.
17745 * rtl.c (rtx_alloc_stat): Remove _stat from name.
17746 * rtl.h (rtx_alloc): Remove macro.
17748 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17750 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
17751 (build_tree_list_stat): Likewise.
17752 * tree.h (build_tree_list): Remove macro.
17753 (build_tree_list_vec): Likewise.
17755 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17757 * tree.c (make_vector_stat): Remove _stat from name.
17758 (build_vector_stat): Likewise.
17759 * tree.h (make_vector_stat): Remove macro.
17760 (build_vector_stat): Likewise.
17762 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17764 * tree.h (build_var_debug_value): Remove prototype.
17766 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17768 * tree.c (tree_cons_stat): Remove _stat from name.
17769 * tree.h (tree_cons): Remove macro.
17771 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17773 * tree.c (build_vl_exp_stat): Remove _stat from name.
17774 * tree.h (build_vl_exp): Remove macro.
17776 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17778 * tree.c (build_decl_stat): Remove _stat from name.
17779 * tree.h (build_decl): Remove macro.
17781 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17783 * gimple.c (gimple_build_with_ops_stat): Adjust.
17784 (gimple_alloc_stat): Remove _stat from name.
17785 * gimple.h (gimple_alloc): Remove macro.
17787 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17789 * tree.c (make_tree_vec_stat): Remove _stat from name.
17790 (grow_tree_vec_stat): Likewise.
17791 * tree.h (make_tree_vec_stat): Adjust prototype.
17792 (grow_tree_vec_stat): Likewise.
17793 (make_tree_vec): Remove macro.
17794 (grow_tree_vec): Likewise.
17796 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17798 * fold-const.c (fold_build1_stat_loc): Adjust.
17799 (fold_build2_stat_loc): Likewise.
17800 (fold_build3_stat_loc): Likewise.
17801 * tree.c (build0_stat): Remove _stat from name.
17802 (build1_stat): Likewise.
17803 (build2_stat): Likewise.
17804 (build3_stat): Likewise.
17805 (build4_stat): Likewise.
17806 (build5_stat): Likewise.
17807 * tree.h (build1_loc): Remove macro, and rename _stat function
17809 (build2_loc): Likewise.
17810 (build3_loc): Likewise.
17811 (build4_loc): Likewise.
17812 (build5_loc): Likewise.
17814 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17816 * tree.c (make_int_cst_stat): Remove _stat from name.
17817 * tree.h (make_int_cst_stat): Adjust prototype.
17818 (make_int_cst): Remove macro.
17820 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17822 * tree.c (make_tre_binfo_stat): Remove _stat from name.
17823 * tree.h (make_tree_binfo_stat): Adjust prototype.
17824 (make_tree_binfo): Remove.
17826 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17828 * tree.c (copy_node_stat): Rename to copy_node.
17829 (build_distinct_type_copy): Adjust.
17830 * tree.h (copy_node_stat): Adjust prototype.
17831 (copy_node): Remove macro.
17833 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17835 * tree.c (make_node_stat): rename to make_node.
17836 (build_tree_list_stat): Adjust.
17837 (build0_stat): Likewise.
17838 (build2_stat): Likewise.
17839 (build3_stat): Likewise.
17840 (build4_stat): Likewise.
17841 (build5_stat): Likewise.
17842 (build_decl_stat): Likewise.
17843 * tree.h (make_node_stat): Adjust prototype.
17844 (make_node): remove macro.
17846 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
17848 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
17849 (PPC_FEATURE2_SCV): Likewise.
17850 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
17852 2017-07-28 Tamar Christina <tamar.christina@arm.com>
17854 * config/aarch64/aarch64.c
17855 (aarch64_internal_mov_immediate): Add new special pattern.
17856 * config/aarch64/aarch64.md (*movdi_aarch64):
17857 Add reg/32bit const mov case.
17859 2017-07-28 Tamar Christina <tamar.christina@arm.com>
17860 Richard Sandiford <richard.sandiford@linaro.org>
17862 * config/aarch64/aarch64.md (mov<mode>): Generalize.
17863 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
17864 Add integer and movi cases.
17865 (movi-split-hf-df-sf split, fp16): New.
17866 (enabled): Added TARGET_FP_F16INST.
17867 * config/aarch64/iterators.md (GPF_HF): New.
17868 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
17870 2017-07-28 Tamar Christina <tamar.christina@arm.com>
17872 * config/aarch64/aarch64.c
17873 (aarch64_simd_container_mode): Add prototype.
17874 (aarch64_expand_mov_immediate): Add HI support.
17875 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
17876 (aarch64_can_const_movi_rtx_p): New.
17877 (aarch64_preferred_reload_class):
17878 Remove restrictions of using FP registers for certain SIMD operations.
17879 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
17880 (aarch64_valid_floating_const): Add integer move validation.
17881 (aarch64_simd_imm_scalar_p): Remove.
17882 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
17883 (aarch64_legitimate_constant_p): Expand list of supported cases.
17884 * config/aarch64/aarch64-protos.h
17885 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
17886 (aarch64_reinterpret_float_as_int): New.
17887 (aarch64_simd_imm_scalar_p): Remove.
17888 * config/aarch64/constraints.md (Uvi): New.
17889 (Dd): Split into Ds and new Dd.
17890 * config/aarch64/aarch64.md (*movsi_aarch64):
17892 (*movdi_aarch64): Add SIMD mov case.
17894 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17896 * tree-predcom.c: (struct chain): Handle store-store chain in which
17897 stores for elimination only store loop invariant values.
17898 (execute_pred_commoning_chain): Ditto.
17899 (prepare_initializers_chain_store_elim): Ditto.
17900 (prepare_finalizers): Ditto.
17901 (is_inv_store_elimination_chain): New function.
17902 (initialize_root_vars_store_elim_1): New function.
17904 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17906 * tree-predcom.c: Revise general description of the pass.
17907 (enum chain_type): New enum type for store elimination.
17908 (struct chain): New field supporting store elimination.
17909 (struct component): Ditto.
17910 (dump_chain): Dump store-stores chain.
17911 (release_chain): Release resources.
17912 (split_data_refs_to_components): Compute and create component
17913 contains only stores for elimination.
17914 (get_chain_last_ref_at): New function.
17915 (make_invariant_chain): Initialization.
17916 (make_rooted_chain): Specify chain type in parameter and record it.
17917 (add_looparound_copies): Skip for store-stores chain.
17918 (determine_roots_comp): Compute type of chain and pass it to
17920 (initialize_root_vars_store_elim_2): New function.
17921 (finalize_eliminated_stores): New function.
17922 (remove_stmt): Handle store for elimination.
17923 (execute_pred_commoning_chain): Execute predictive commoning on
17924 store-store chains.
17925 (determine_unroll_factor): Skip unroll for store-stores chain.
17926 (prepare_initializers_chain_store_elim): New function.
17927 (prepare_initializers_chain): Hanlde store-store chain.
17928 (prepare_finalizers_chain, prepare_finalizers): New function.
17929 (tree_predictive_commoning_loop): Return integer value indicating
17930 if loop is unrolled or lcssa form is corrupted.
17931 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
17933 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17935 * tree-predcom.c (initialize_root): Delete.
17936 (execute_pred_commoning_chain): Initialize root vars and replace
17937 reference of non-combined chain directly, rather than call above
17940 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17942 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
17943 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
17945 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17947 * tree-predcom.c (struct chain): New field init_seq.
17948 (release_chain): Release init_seq.
17949 (prepare_initializers_chain): Record intialization stmts in above
17951 (insert_init_seqs): New function.
17952 (tree_predictive_commoning_loop): Call insert_init_seqs.
17954 2017-07-28 Bin Cheng <bin.cheng@arm.com>
17956 * tree-predcom.c (determine_roots_comp): Skip trivial components.
17958 2017-07-28 Richard Biener <rguenther@suse.de>
17960 * match.pd: Remove superfluous :c.
17961 * genmatch.c (simplify::id): Add member.
17962 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
17964 (current_id): New global.
17965 (dt_node::parent): Move from ...
17966 (dt_operand::parent): ... here. Add for_id member.
17967 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
17968 (decision_tree::find_node): Relax order requirement when
17969 merging DT_TRUE nodes to ones inbetween the current simplify
17970 and the one we try to merge with. Add diagnostic whenever
17971 we need to enforce pattern order by not merging.
17972 (decision_tree::insert): Set current_id.
17973 (decision_tree::print_node): Dump parent node and for_id.
17974 (parser::last_id): Add member.
17975 (parser::push_simplify): Assign unique id.
17976 (parser::parser): Initialize last_id.
17978 2017-07-28 Martin Liska <mliska@suse.cz>
17981 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
17982 gimple_build_debug_bind.
17984 2017-07-28 Richard Biener <rguenther@suse.de>
17986 PR tree-optimization/81502
17987 * match.pd: Add pattern combining BIT_INSERT_EXPR with
17989 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
17991 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
17992 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
17993 for BIT_FIELD_REF args.
17994 * fold-const.c (make_bit_field_ref): Likewise.
17995 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
17997 2017-07-28 Jakub Jelinek <jakub@redhat.com>
18000 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
18001 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
18002 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
18003 Or it into SANITIZER_UNDEFINED.
18004 * ubsan.c: Include gimple-fold.h and varasm.h.
18005 (ubsan_expand_ptr_ifn): New function.
18006 (instrument_pointer_overflow): New function.
18007 (maybe_instrument_pointer_overflow): New function.
18008 (instrument_object_size): Formatting fix.
18009 (pass_ubsan::execute): Call instrument_pointer_overflow
18010 and maybe_instrument_pointer_overflow.
18011 * internal-fn.c (expand_UBSAN_PTR): New function.
18012 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
18013 * sanitizer.def (__ubsan_handle_pointer_overflow,
18014 __ubsan_handle_pointer_overflow_abort): New builtins.
18015 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
18016 * internal-fn.def (UBSAN_PTR): New internal function.
18017 * opts.c (sanitizer_opts): Add pointer-overflow.
18018 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
18019 * fold-const.c (build_range_check): Compute pointer range check in
18020 integral type if pointer arithmetics would be needed. Formatting
18023 2017-07-28 Martin Liska <mliska@suse.cz>
18026 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
18027 parameters that are of a variable-length.
18029 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
18031 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
18033 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
18034 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
18035 (CRT_CALL_STATIC_FUNCTION): Likewise.
18036 (ASM_DEFAULT_SPEC): New define.
18037 (ASM_SPEC32): Likewise.
18038 (ASM_SPEC64): Likewise.
18039 (ASM_SPEC_COMMON): Likewise.
18040 (ASM_SPEC): Likewise.
18041 (INVALID_64BIT): Likewise.
18042 (LINK_OS_DEFAULT_SPEC): Likewise.
18043 (LINK_OS_SPEC32): Likewise.
18044 (LINK_OS_SPEC64): Likewise.
18045 (POWERPC_LINUX): Likewise.
18046 (PTRDIFF_TYPE): Likewise.
18047 (RESTORE_FP_PREFIX): Likewise.
18048 (RESTORE_FP_SUFFIX): Likewise.
18049 (SAVE_FP_PREFIX): Likewise.
18050 (SAVE_FP_SUFFIX): Likewise.
18051 (SIZE_TYPE): Likewise.
18052 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
18053 (TARGET_64BIT): Likewise.
18054 (TARGET_64BIT): Likewise.
18055 (TARGET_AIX): Likewise.
18056 (WCHAR_TYPE_SIZE): Likewise.
18057 (WCHAR_TYPE): Undefine.
18058 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
18059 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
18060 (CPP_OS_RTEMS_SPEC): Delete.
18061 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
18062 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
18064 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
18066 2017-07-28 Jakub Jelinek <jakub@redhat.com>
18068 PR tree-optimization/81578
18069 * tree-parloops.c (build_new_reduction): Bail out if
18070 reduction_code isn't one of the standard OpenMP reductions.
18071 Move the details printing after that decision.
18073 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18075 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
18076 related to reload_in_progress.
18077 (splat_input_operand): Likewise.
18078 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
18080 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
18082 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
18083 (TARGET_INSTANTIATE_DECLS): Likewise.
18084 (legitimate_indexed_address_p): Delete reload_in_progress code.
18085 (rs6000_debug_legitimate_address_p): Likewise.
18086 (rs6000_eliminate_indexed_memrefs): Likewise.
18087 (rs6000_emit_le_vsx_store): Likewise.
18088 (rs6000_emit_move_si_sf_subreg): Likewise.
18089 (rs6000_emit_move): Likewise.
18090 (register_to_reg_type): Likewise.
18091 (rs6000_pre_atomic_barrier): Likewise.
18092 (rs6000_machopic_legitimize_pic_address): Likewise.
18093 (rs6000_allocate_stack_temp): Likewise.
18094 (rs6000_address_for_fpconvert): Likewise.
18095 (rs6000_address_for_altivec): Likewise.
18096 (rs6000_secondary_memory_needed_rtx): Delete function.
18097 (rs6000_check_sdmode): Likewise.
18098 (rs6000_alloc_sdmode_stack_slot): Likewise.
18099 (rs6000_instantiate_decls): Likewise.
18100 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
18101 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
18102 Delete reload_in_progress.
18103 (*vec_reload_and_plus_<mptrsize>): Likewise.
18104 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
18105 (vsx_div_v2di): Likewise.
18106 (vsx_udiv_v2di): Likewise.
18108 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18110 * config/rs6000/rs6000.opt (mlra): Replace with stub.
18111 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
18112 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
18113 (rs6000_debug_reg_global): Delete print of LRA status.
18114 (rs6000_option_override_internal): Delete dead LRA related code.
18115 (rs6000_lra_p): Delete function.
18116 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
18118 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
18120 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
18121 * config/riscv/rtems.h: New file.
18123 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18124 Sudakshina Das <sudi.das@arm.com>
18126 * config/aarch64/aarch64.md
18127 (define_split for and<mode>3nr_compare): Move
18128 non aarch64_logical_operand to a register.
18129 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
18130 register immediate operand to a register.
18131 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
18133 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18135 PR middle-end/81564
18136 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
18138 2017-07-27 Richard Biener <rguenther@suse.de>
18140 PR tree-optimization/81573
18141 PR tree-optimization/81494
18142 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
18143 multi defuse cycle case.
18145 2017-07-27 Richard Biener <rguenther@suse.de>
18147 PR tree-optimization/81571
18148 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
18151 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
18153 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
18154 earlier and only if MASK_FPU is set. Adjust formatting.
18156 2017-07-27 Martin Liska <mliska@suse.cz>
18158 * opt-functions.awk: Add validation of value of Init.
18159 * optc-gen.awk: Pass new argument.
18161 2017-07-27 Martin Liska <mliska@suse.cz>
18163 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
18164 Fix wrong condition.
18166 2017-07-27 Martin Liska <mliska@suse.cz>
18168 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
18169 BBs and edges seen by autoFDO.
18171 2017-07-27 Richard Biener <rguenther@suse.de>
18173 PR tree-optimization/81502
18174 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
18175 with incompatible but same sized type.
18176 (execute_update_addresses_taken): Likewise.
18178 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
18180 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
18181 flag_tree_loop_vectorize rather than flag_tree_vectorize.
18183 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18186 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
18187 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
18188 Change s_operand to memory_operand.
18190 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
18192 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
18193 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
18194 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
18195 Emit instructions rather than returning an expression. Handle TFmode
18196 and KFmode by casting to TImode.
18197 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
18198 (rs6000_emit_le_vsx_store): Likewise.
18199 * config/rs6000/vsx.md (VSX_TI): New iterator.
18200 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
18201 (*vsx_le_undo_permute_<mode>): Likewise.
18202 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
18203 emit the split sequence.
18204 (*vsx_le_perm_store_<mode>): Likewise.
18206 2017-07-27 Jakub Jelinek <jakub@redhat.com>
18208 PR tree-optimization/81555
18209 PR tree-optimization/81556
18210 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
18211 if true, force CHANGED for the recursive invocation.
18212 (reassociate_bb): Remember original length of ops array, pass
18213 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
18215 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
18216 attributes for noipa attribute. For naked attribute use
18217 lookup_attribute first before lookup_attribute_spec.
18218 * final.c (rest_of_handle_final): Disable IPA RA for functions with
18220 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
18221 for functions with noipa attribute.
18222 (cgraph_externally_visible_p): Return true for functions with noipa
18224 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
18225 for functions with noipa attribute.
18226 * doc/extend.texi: Document noipa function attribute.
18227 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
18228 also for functions with noipa attribute.
18229 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
18231 2017-07-26 Andrew Pinski <apinski@cavium.com>
18233 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
18234 vec_unalign_load_cost and vec_unalign_store_cost.
18236 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
18238 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
18239 -mvsx-small-integer option.
18240 (ISA_3_0_MASKS_IEEE): Likewise.
18241 (OTHER_VSX_VECTOR_MASKS): Likewise.
18242 (POWERPC_MASKS): Likewise.
18243 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
18244 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
18245 code, only testing for DImode being allowed in non-VSX floating
18247 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
18248 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
18250 (rs6000_option_override_internal): Delete -mvsx-small-integer.
18251 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
18252 TARGET_P8_VECTOR test.
18253 (rs6000_secondary_reload_simple_move): Likewise.
18254 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
18255 since TARGET_P9_VECTOR was already tested.
18256 (rs6000_opt_masks): Remove -mvsx-small-integer.
18257 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
18258 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18260 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
18261 test for TARGET_VEXTRACTUB was used, and that uses
18263 (p9 extract splitter): Likewise.
18264 (vsx_extract_<mode>_di_p9): Likewise.
18265 (vsx_extract_<mode>_store_p9): Likewise.
18266 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
18267 for TARGET_P9_VECTOR was used. Delete code that is now dead with
18268 the elimination of TARGET_VSX_SMALL_INTEGER.
18269 (vsx_extract_<mode>_p8): Likewise.
18270 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
18271 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
18272 (vsx_set_<mode>_p9): Likewise.
18273 (vsx_set_v4sf_p9): Likewise.
18274 (vsx_set_v4sf_p9_zero): Likewise.
18275 (vsx_insert_extract_v4sf_p9): Likewise.
18276 (vsx_insert_extract_v4sf_p9_2): Likewise.
18277 * config/rs6000/rs6000.md (sign extend splitter): Change
18278 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
18279 (floatsi<mode>2_lfiwax_mem): Likewise.
18280 (floatunssi<mode>2_lfiwzx_mem): Likewise.
18281 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
18282 since a test for TARGET_P9_VECTOR was used.
18283 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18284 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
18285 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18286 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
18287 TARGET_P8_VECTOR test.
18288 (fix_trunc<mode>si2_stfiwx): Likewise.
18289 (fix_trunc<mode>si2_internal): Likewise.
18290 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
18291 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18293 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
18294 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
18295 TARGET_P8_VECTOR test.
18296 (fixuns_trunc<mode>si2_stfiwx): Likewise.
18297 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
18298 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18300 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
18301 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
18302 since a test for TARGET_P9_VECTOR was used.
18303 (splitter for loading small constants): Likewise.
18305 2017-07-26 Andrew Pinski <apinski@cavium.com>
18307 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
18310 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
18313 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
18314 (fp_valid_at): Likewise.
18316 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
18318 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
18319 (qdf24xx_addrcost_table): Likewise.
18320 (cortexa57_tunings): Update to use generic_branch_cost.
18321 (cortexa72_tunings): Likewise.
18322 (cortexa73_tunings): Likewise.
18323 (qdf24xx_tunings): Likewise.
18325 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
18327 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
18328 (thunderx2t99_branch_cost): Likewise.
18329 (cortexa35_tunings): Update to use generic_branch_cost.
18330 (cortexa53_tunings): Likewise.
18331 (cortexa57_tunings): Likewise.
18332 (cortexa72_tunings): Likewise.
18333 (cortexa73_tunings): Likewise.
18334 (thunderx2t99_tunings): Likewise.
18336 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
18338 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
18339 (sparc_option_override): Honour MASK_FSMULD.
18340 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
18341 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
18342 * config/sparc/sparc.opt (mfsmuld): New option.
18343 * doc/invoke.texi (mfsmuld): Document option.
18345 2017-07-26 Marek Polacek <polacek@redhat.com>
18347 PR middle-end/70992
18348 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
18350 2017-07-26 Richard Biener <rguenther@suse.de>
18352 * gimple-match-head.c (do_valueize): Return OP if valueize
18354 (get_def): New helper to get at the def stmt of a SSA name
18355 if valueize allows.
18356 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
18357 do_valueize to get at the def stmt.
18358 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
18360 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
18362 PR middle-end/46932
18363 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
18365 2017-07-26 Martin Liska <mliska@suse.cz>
18368 * function.c (expand_function_start): Make expansion of
18369 nonlocal_goto_save_area after parm_birth_insn.
18371 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
18373 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
18374 from all CPU target flags enable members.
18376 2017-07-26 Richard Biener <rguenther@suse.de>
18378 * genmatch.c (dt_simplify::gen): Make iterator vars const.
18379 (decision_tree::gen): Make 'type' const.
18380 (write_predicate): Likewise.
18382 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
18384 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
18385 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
18386 (rs6000_option_override_internal): Likewise.
18387 (rs6000_expand_vector_set): Likewise.
18388 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
18389 (TARGET_UPPER_REGS_SF): Likewise.
18390 (TARGET_UPPER_REGS_DI): Likewise.
18391 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
18392 (TARGET_DIRECT_MOVE_64BIT): Likewise.
18393 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
18394 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18395 (Splitters for DI constants in Altivec registers): Likewise.
18396 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
18397 (vsx_set_v4sf_p9): Likewise.
18398 (vsx_set_v4sf_p9_zero): Likewise.
18399 (vsx_insert_extract_v4sf_p9): Likewise.
18400 (vsx_insert_extract_v4sf_p9_2): Likewise.
18402 2017-07-25 Carl Love <cel@us.ibm.com>
18404 * doc/extend.texi: Update the built-in documentation file for the
18405 existing built-in functions
18406 vector signed char vec_cnttz (vector signed char);
18407 vector unsigned char vec_cnttz (vector unsigned char);
18408 vector signed short vec_cnttz (vector signed short);
18409 vector unsigned short vec_cnttz (vector unsigned short);
18410 vector signed int vec_cnttz (vector signed int);
18411 vector unsigned int vec_cnttz (vector unsigned int);
18412 vector signed long long vec_cnttz (vector signed long long);
18413 vector unsigned long long vec_cnttz (vector unsigned long long);
18415 2017-07-25 Andrew Pinski <apinski@cavium.com>
18417 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
18418 accesses where the use is for the first operand of a BIT_INSERT.
18420 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
18423 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
18424 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
18426 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
18428 * config/i386/gstabs.h: Delete.
18429 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
18431 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
18433 * config/i386/i386.c (ix86_decompose_address): Do not check for
18434 register RTX when looking at index_reg or base_reg.
18435 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
18437 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
18439 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
18440 to update EH info here.
18442 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18444 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
18446 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18448 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
18450 2017-07-25 Torsten Duwe <duwe@suse.de>
18452 * common.opt: Introduce -fpatchable-function-entry
18453 command line option, and its variables function_entry_patch_area_size
18454 and function_entry_patch_area_start.
18455 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
18456 including a two-value parser.
18457 * target.def (print_patchable_function_entry): New target hook.
18458 * targhooks.h (default_print_patchable_function_entry): New function.
18459 * targhooks.c (default_print_patchable_function_entry): Likewise.
18460 * toplev.c (process_options): Switch off IPA-RA if
18461 patchable function entries are being generated.
18462 * varasm.c (assemble_start_function): Look at the
18463 patchable-function-entry command line switch and current
18464 function attributes and maybe generate NOP instructions by
18465 calling the print_patchable_function_entry hook.
18466 * doc/extend.texi: Document patchable_function_entry attribute.
18467 * doc/invoke.texi: Document -fpatchable_function_entry
18468 command line option.
18469 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
18471 * doc/tm.texi: Re-generate.
18473 2017-07-25 Jakub Jelinek <jakub@redhat.com>
18476 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
18477 TARGET_AVX512DQ rather than TARGET_AVX512BW.
18479 2017-07-25 Tamar Christina <tamar.christina@arm.com>
18481 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
18483 2017-07-25 Richard Biener <rguenther@suse.de>
18485 PR tree-optimization/81455
18486 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
18487 not walk in cycles when looking for guards.
18489 2017-07-25 Richard Biener <rguenther@suse.de>
18491 PR tree-optimization/81529
18492 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
18493 when optimizing backedge uses.
18495 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
18497 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
18499 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
18500 to dl_section_ref. On AIX, append an expression to subtract
18501 the size of the section length to dl_section_ref.
18503 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
18505 * configure.ac: If any of the config.* scripts fail, exit 1.
18506 * configure: Regenerate.
18508 2017-07-25 Richard Biener <rguenther@suse.de>
18510 PR middle-end/81546
18511 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
18512 of immediate uses, be more verbose on errors.
18514 2017-07-25 Richard Biener <rguenther@suse.de>
18516 PR tree-optimization/81510
18517 * tree-vect-loop.c (vect_is_simple_reduction): When the
18518 reduction stmt is not inside the loop bail out.
18520 2017-07-25 Richard Biener <rguenther@suse.de>
18522 PR tree-optimization/81303
18523 * tree-vect-loop-manip.c (vect_loop_versioning): Build
18524 profitability check against LOOP_VINFO_NITERSM1.
18526 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
18528 * domwalk.c (cmp_bb_postorder): Simplify.
18529 (sort_bbs_postorder): New function. Use it...
18530 (dom_walker::walk): ...here to optimize common cases.
18532 2017-07-25 Martin Liska <mliska@suse.cz>
18535 * ipa-visibility.c (function_and_variable_visibility): Make the
18536 redirection just on target that supports aliasing.
18537 Fix GNU coding style.
18539 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
18542 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
18544 * config/arm/rtems.h: Likewise.
18545 * config/bfin/rtems.h: Likewise.
18546 * config/i386/rtemself.h: Likewise.
18547 * config/lm32/rtems.h: Likewise.
18548 * config/m32c/rtems.h: Likewise.
18549 * config/m68k/rtemself.h: Likewise.
18550 * config/microblaze/rtems.h: Likewise.
18551 * config/mips/rtems.h: Likewise.
18552 * config/moxie/rtems.h: Likewise.
18553 * config/nios2/rtems.h: Likewise.
18554 * config/powerpcspe/rtems.h: Likewise.
18555 * config/rs6000/rtems.h: Likewise.
18556 * config/rtems.h: Likewise.
18557 * config/sh/rtems.h: Likewise.
18558 * config/sh/rtemself.h: Likewise.
18559 * config/sparc/rtemself.h: Likewise.
18561 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
18564 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
18565 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
18566 * tree-ssa-structalias.c (alias_get_name): Same.
18568 2017-07-25 Bin Cheng <bin.cheng@arm.com>
18571 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
18572 instructions if no du chain is found.
18574 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
18576 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
18578 2017-07-25 Richard Biener <rguenther@suse.de>
18580 PR middle-end/81505
18581 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
18584 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
18586 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
18587 upper-regs options.
18588 (ISA_2_7_MASKS_SERVER): Likewise.
18589 (ISA_3_0_MASKS_IEEE): Likewise.
18590 (OTHER_P8_VECTOR_MASKS): Likewise.
18591 (OTHER_VSX_VECTOR_MASKS): Likewise.
18592 (POWERPC_MASKS): Likewise.
18593 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
18594 duplicate list of options.
18595 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
18596 explicit -mupper-regs options.
18597 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
18598 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
18599 alias for -mupper-regs-df.
18600 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
18601 (rs6000_init_hard_regno_mode_ok): Likewise.
18602 (rs6000_option_override_internal): Likewise.
18603 (rs6000_opt_masks): Likewise.
18604 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
18605 options in terms of whether -mvsx or -mpower8-vector was used.
18606 (TARGET_UPPER_REGS_DI): Likewise.
18607 (TARGET_UPPER_REGS_SF): Likewise.
18608 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
18609 -mupper-regs-* options.
18611 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
18613 * passes.c (emergency_dump_function): Print some empty lines and a
18614 header before the RTL dump.
18616 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
18618 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
18620 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
18623 * config/aarch64/aarch64.c (aarch64_classify_symbol):
18624 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
18626 2017-07-24 Carl Love <cel@us.ibm.com>
18628 * config/rs6000/rs6000-c.c: Add support for built-in functions
18629 vector float vec_extract_fp32_from_shorth (vector unsigned short);
18630 vector float vec_extract_fp32_from_shortl (vector unsigned short);
18631 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
18632 vec_extract_fp_from_shortl): Add defines for the two builtins.
18633 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
18634 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
18636 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
18637 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
18638 * doc/extend.texi: Update the built-in documentation file for the
18639 new built-in function.
18641 2017-07-24 Jakub Jelinek <jakub@redhat.com>
18644 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
18646 * doc/generic.texi: Likewise.
18647 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
18648 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
18650 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
18652 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
18653 (aarch64_mls_elt_merge<mode>): Likewise.
18655 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
18657 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
18658 having __cxa_atexit.
18660 2017-07-23 Michael Collison <michael.collison@arm.com>
18662 * config/arm/arm.c (arm_option_override): Deprecate
18663 use of -mstructure-size-boundary.
18664 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
18665 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
18667 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18670 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
18671 Reduce cost estimate for direct moves.
18673 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
18676 * config/i386/i386.c (ix86_option_override_internal): Disable
18677 BMI, BMI2 and TBM instructions for -m16.
18679 2017-07-21 Carl Love <cel@us.ibm.com>
18681 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18682 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
18683 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
18684 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
18685 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
18686 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
18687 VMULOSW): New enum "unspec" values.
18688 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
18689 altivec_vmulosw): New patterns.
18690 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
18691 VMULOSW): Add definitions.
18693 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
18695 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
18696 (qdf24xx): Likewise.
18697 * config/aarch64/aarch64-options-extensions.def (rdma); New.
18698 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
18699 (AARCH64_FL_V8_1): Renumber.
18700 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
18701 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
18702 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
18703 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
18704 rdma to feature modifiers list.
18706 2017-07-21 Yury Gribov <tetra2005@gmail.com>
18708 PR middle-end/56727
18709 * ipa-visibility (function_and_variable_visibility): Convert
18710 recursive PLT call to direct call if appropriate.
18712 2017-07-21 Andrew Pinski <apinski@cavium.com>
18714 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
18715 operand 1 to see if the types precision matches.
18716 * fold-const.c (operand_equal_p): Likewise.
18718 2017-07-21 Richard Biener <rguenther@suse.de>
18720 PR tree-optimization/81303
18721 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
18722 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
18723 (vect_peeling_hash_get_lowest_cost): Adjust.
18724 (vect_enhance_data_refs_alignment): Likewise. Use
18725 vect_get_peeling_costs_all_drs to compute the penalty for no
18726 peeling to match up costs.
18728 2017-07-21 Richard Biener <rguenther@suse.de>
18730 PR tree-optimization/81500
18731 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
18732 we didn't identify a reduction path.
18734 2017-07-21 Tom de Vries <tom@codesourcery.com>
18735 Cesar Philippidis <cesar@codesourcery.com>
18737 PR gcov-profile/81442
18738 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
18741 2017-07-21 Tom de Vries <tom@codesourcery.com>
18744 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
18746 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
18747 nvptx_override_options_after_change.
18749 2017-07-21 Ulrich Drepper <drepper@redhat.com>
18751 * dwarf2out.c (output_file_names): Avoid double testing for
18752 dwarf_version >= 5.
18754 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
18756 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
18758 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
18760 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
18762 (try_crossjump_to_edge): Do not punt on partitioned functions.
18764 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
18766 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
18767 Put all BBs reachable only via paths crossing cold region to cold
18769 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
18771 2016-07-21 Richard Biener <rguenther@suse.de>
18773 PR tree-optimization/81303
18774 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
18775 into account prologue and epilogue iterations when raising
18776 min_profitable_iters to sth at least covering one vector iteration.
18778 2017-07-21 Tamar Christina <tamar.christina@arm.com>
18780 * config/arm/arm.c (arm_test_cpu_arch_dat):
18783 2017-07-20 Nathan Sidwell <nathan@acm.org>
18785 Remove TYPE_METHODS.
18786 * tree.h (TYPE_METHODS): Delete.
18787 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
18788 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
18789 (dbxout_type_methods): Scan TYPE_FIELDS.
18790 (dbxout_type): Don't check TYPE_METHODS here.
18791 * function.c (use_register_for_decl): Always ignore register for
18792 class types when not optimizing.
18793 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
18794 * tree.c (free_lang_data_in_type): Stitch out member functions and
18795 templates from TYPE_FIELDS.
18796 (build_distinct_type_copy, verify_type_variant,
18797 verify_type): Member fns are on TYPE_FIELDS.
18798 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
18799 * tree-pretty-print.c (dump_generic_node): Likewise.
18801 2017-07-20 Jakub Jelinek <jakub@redhat.com>
18804 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
18805 V2TImode and V4TImode.
18806 (ix86_expand_vector_extract): Likewise.
18807 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
18808 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
18809 (ssescalarmode): Handle V4TImode and V2TImode.
18810 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
18811 (*vec_extractv2ti, *vec_extractv4ti): New insns.
18812 (VEXTRACTI128_MODE): New mode iterator.
18813 (splitter for *vec_extractv?ti first element): New.
18814 (VEC_INIT_MODE): New mode iterator.
18815 (vec_init<mode>): Consolidate 3 expanders into one using
18816 VEC_INIT_MODE mode iterator.
18818 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
18820 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
18821 non_spilled_static_chain_regno_p.
18823 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
18825 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
18827 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
18829 * bb-reorder.c (connect_traces): Allow copying of blocks within
18832 2017-07-20 Richard Biener <rguenther@suse.de>
18834 * gimple.h (gimple_phi_result): Add gphi * overload.
18835 (gimple_phi_result_ptr): Likewise.
18836 (gimple_phi_arg): Likewise. Adjust index assert to only
18837 allow actual argument accesses rather than all slots available
18839 (gimple_phi_arg_def): Add gphi * overload.
18840 * tree-phinodes.c (make_phi_node): Initialize only actual
18842 (resize_phi_node): Clear memory not covered by old node,
18843 do not initialize excess argument slots.
18844 (reserve_phi_args_for_new_edge): Initialize new argument slot
18847 2017-07-20 Bin Cheng <bin.cheng@arm.com>
18849 PR tree-optimization/81388
18851 2016-07-21 Bin Cheng <bin.cheng@arm.com>
18853 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
18854 by removing computation of may_be_zero.
18856 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
18857 Tom de Vries <tom@codesourcery.com>
18859 PR middle-end/81030
18860 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
18861 when gimple level profile disagrees with what RTL expander did.
18863 2017-07-20 Richard Biener <rguenther@suse.de>
18865 PR tree-optimization/61171
18866 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
18867 (vect_analyze_stmt): Add slp instance parameter.
18868 (vectorizable_reduction): Likewise.
18869 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
18870 (vect_is_simple_reduction): Deal with chains not detected
18871 as SLP reduction chain, specifically not properly associated
18872 chains containing a mix of plus/minus.
18873 (get_reduction_op): Remove.
18874 (get_initial_defs_for_reduction): Simplify, pass in whether
18875 this is a reduction chain, pass in the SLP node for the PHIs.
18876 (vect_create_epilog_for_reduction): Get the SLP instance as
18878 (vectorizable_reduction): Get the SLP instance as arg.
18879 During analysis remember the SLP node with the PHIs in the
18880 instance. Simplify getting at the vectorized reduction PHIs.
18881 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
18882 through SLP instance.
18883 (vect_slp_analyze_operations): Likewise.
18884 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
18885 (vect_transform_stmt): Likewise.
18887 2017-07-20 Tom de Vries <tom@codesourcery.com>
18889 PR tree-optimization/81489
18890 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
18891 read of phi arg location to before loop that modifies phi.
18893 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18895 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
18898 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
18900 PR middle-end/81331
18901 * except.c (execute): Fix ordering issue.
18903 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
18905 PR rtl-optimization/81423
18906 * combine.c (make_compound_operation_int): Don't try to optimize
18907 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
18909 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
18911 PR rtl-optimization/81423
18912 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
18913 with a constant that is -1 in the truncated to mode.
18915 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
18917 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
18918 (determine_unlikely_bbs): ... here.
18919 * predict.h (propagate_unlikely_bbs_forward): Declare.
18920 * cfgexpand.c (pass_expand::execute): Use it.
18921 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
18923 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
18924 propagate_unlikely_bbs_forward.
18926 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
18928 PR middle-end/81331
18929 * except.c (maybe_add_nop_after_section_switch): New function.
18932 2017-07-19 Tom de Vries <tom@codesourcery.com>
18934 * gimple.h (gimple_phi_set_arg): Make assert more strict.
18936 2017-07-19 Tom de Vries <tom@codesourcery.com>
18938 * gimple.h (gimple_phi_arg): Make assert more strict.
18940 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
18942 * config.gcc (powerpc*-*-*): Add mmintrin.h.
18943 * config/rs6000/mmintrin.h: New file.
18944 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
18946 2017-07-19 Jakub Jelinek <jakub@redhat.com>
18948 PR tree-optimization/81346
18949 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
18951 2017-07-19 Tom de Vries <tom@codesourcery.com>
18953 * config/nvptx/nvptx.md (VECIM): Add V2DI.
18955 2017-07-19 Tom de Vries <tom@codesourcery.com>
18957 * config/nvptx/nvptx-modes.def: Add V2DImode.
18958 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
18959 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
18960 (nvptx_output_mov_insn): Handle lack of mov.b128.
18961 (nvptx_print_operand): Handle 'H' and 'L' codes.
18962 (nvptx_vector_mode_supported): Allow V2DImode.
18963 (nvptx_preferred_simd_mode): New function.
18964 (nvptx_data_alignment): New function.
18965 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
18966 nvptx_preferred_simd_mode.
18967 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
18969 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
18971 2017-07-19 Tom de Vries <tom@codesourcery.com>
18973 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
18974 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
18975 (nvptx_vector_mode_supported): New function. Allow V2SImode.
18976 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
18977 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
18978 (mov<VECIM>_insn): New define_insn.
18979 (define_expand "mov<VECIM>): New define_expand.
18981 2017-07-19 Tom de Vries <tom@codesourcery.com>
18983 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
18985 2017-07-19 Jakub Jelinek <jakub@redhat.com>
18987 PR tree-optimization/81346
18988 * fold-const.h (fold_div_compare, range_check_type): Declare.
18989 * fold-const.c (range_check_type): New function.
18990 (build_range_check): Use range_check_type.
18991 (fold_div_compare): No longer static, rewritten into
18992 a match.pd helper function.
18993 (fold_comparison): Don't call fold_div_compare here.
18994 * match.pd (X / C1 op C2): New optimization using fold_div_compare
18995 as helper function.
18997 2017-07-19 Nathan Sidwell <nathan@acm.org>
18999 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
19000 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
19001 * tree.c (find_decls_types_r, verify_type): Use
19002 TYPE_{MIN,MAX}_VALUE_RAW.
19003 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
19004 (hash_tree): Likewise.
19005 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
19007 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
19010 2017-07-18 Tom de Vries <tom@codesourcery.com>
19012 PR middle-end/81464
19013 * omp-expand.c (expand_omp_for_static_chunk): Handle
19014 equal-argument loop exit phi.
19016 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
19019 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
19020 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
19021 operand 2 predicate.
19022 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
19023 operand 2 predicate.
19024 (ror,rol -> rorx splitters): Use const_int_operand as
19025 operand 2 predicate.
19027 2017-06-18 Richard Biener <rguenther@suse.de>
19029 PR tree-optimization/81410
19030 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
19031 the gap in the ! slp_perm SLP case after each group.
19033 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
19035 PR middle-end/81463
19036 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
19039 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
19041 PR middle-end/81462
19042 * predict.c (set_even_probabilities): Cleanup; do not affect
19043 probabilities that are already known.
19044 (combine_predictions_for_bb): Call even when count is set.
19046 2017-07-18 Nathan Sidwell <nathan@acm.org>
19048 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
19051 2017-07-18 Bin Cheng <bin.cheng@arm.com>
19054 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
19055 optimization for loop niter analysis.
19057 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
19060 * config/avr/avr.c (avr_optimize_casesi): Don't use
19061 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
19063 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
19065 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
19066 body_cost_vec from _vect_peel_extended_info.
19067 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
19068 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
19071 2017-07-18 Bin Cheng <bin.cheng@arm.com>
19073 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
19075 2017-07-18 Richard Biener <rguenther@suse.de>
19077 PR tree-optimization/80620
19078 PR tree-optimization/81403
19079 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
19080 info when re-using a VN table entry.
19082 2017-07-18 Richard Biener <rguenther@suse.de>
19084 PR tree-optimization/81418
19085 * tree-vect-loop.c (vectorizable_reduction): Properly compute
19086 vectype_in. Verify that with lane-reducing reduction operations
19087 we have a single def-use cycle.
19089 2017-07-17 Carl Love <cel@us.ibm.com>
19091 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
19093 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19094 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19095 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19096 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19097 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19098 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19099 VMULOSW): New enum "unspec" values.
19100 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19101 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19102 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19103 altivec_vmulosw): New patterns.
19104 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19105 VMULOSW): Add definitions.
19106 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
19108 * config/alpha/alpha.c: Include predict.h.
19110 2017-07-17 Yury Gribov <tetra2005@gmail.com>
19112 * tree-vrp.c (compare_assert_loc): Fix comparison function
19113 to return predictable results.
19115 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19117 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
19119 (subdi3): Likewise.
19120 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
19121 * doc/invoke.texi (mexpand-adddi): Update text.
19123 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19125 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
19126 that also clobbers the CC register. The old expand code is moved
19128 (*arc_clzsi2): ... here.
19129 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
19130 the CC register. The old expand code is moved to ...
19131 (arc_ctzsi2): ... here.
19133 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19135 * config/arc/arc.opt (mindexed-loads): Use initial value
19136 TARGET_INDEXED_LOADS_DEFAULT.
19137 (mauto-modify-reg): Use initial value
19138 TARGET_AUTO_MODIFY_REG_DEFAULT.
19139 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
19140 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
19141 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
19142 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
19144 2017-07-17 Martin Liska <mliska@suse.cz>
19147 * opts.c (finish_options): Do not allow -fgnu-tm
19148 w/ -fsanitize={kernel-,}address. Say sorry.
19150 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19153 * tree-loop-distribution.c (classify_partition): Only assert on
19154 numer of iterations.
19155 (merge_dep_scc_partitions): Delete prameter. Update function call.
19156 (distribute_loop): Remove code handling loop with unknown niters.
19157 (pass_loop_distribution::execute): Skip loop with unknown niters.
19159 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19162 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
19163 function sort_partitions_by_post_order.
19165 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19167 PR tree-optimization/81374
19168 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
19169 the max index of basic blocks, rather than number of basic blocks.
19171 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19173 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
19175 (arc_legitimate_pic_operand_p): Likewise.
19176 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
19178 (arc_needs_pcl_p): Likewise.
19179 (arc_legitimate_pc_offset_p): Likewise.
19180 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
19181 function is also used in constrains.md.
19182 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
19183 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
19184 PLUS. Only return true/false in known cases, otherwise assert.
19185 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
19186 is already called in arc_legitimate_constant_p.
19187 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
19189 (LEGITIMATE_PIC_OPERAND_P): Use
19190 arc_raw_symbolic_reference_mentioned_p function.
19191 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
19196 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19197 Andrew Burgess <andrew.burgess@embecosm.com>
19199 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
19200 (arc_return_address_register): New function.
19201 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
19202 (arc_handle_fndecl_attribute): Add naked attribute.
19203 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
19204 (TARGET_WARN_FUNC_RETURN): Likewise.
19205 (arc_allocate_stack_slots_for_args): New function.
19206 (arc_warn_func_return): Likewise.
19207 (machine_function): Change type fn_type.
19208 (arc_compute_function_type): Consider new naked function type,
19209 change function return type.
19210 (arc_must_save_register): Adapt to handle new
19211 arc_compute_function_type's return type.
19212 (arc_expand_prologue): Likewise.
19213 (arc_expand_epilogue): Likewise.
19214 (arc_return_address_regs): Delete.
19215 (arc_return_address_register): New function.
19216 (arc_epilogue_uses): Use above function.
19217 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
19218 (arc_function_type): Change encoding, add naked type.
19219 (ARC_INTERRUPT_P): Change to handle the new encoding.
19220 (ARC_FAST_INTERRUPT_P): Likewise.
19221 (ARC_NORMAL_P): Define.
19222 (ARC_NAKED_P): Likewise.
19223 (arc_compute_function_type): Delete prototype.
19224 * config/arc/arc.md (in_ret_delay_slot): Use
19225 arc_return_address_register function.
19226 (simple_return): Likewise.
19227 (p_return_i): Likewise.
19229 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19231 PR tree-optimization/81428
19232 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
19233 can't be built for those types.
19235 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
19237 Remove stuff dead since r239246.
19239 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
19240 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
19241 (avr_inform_devices): Remove dead stuff.
19243 2017-07-17 Tamar Christina <tamar.christina@arm.com>
19245 * config/arm/arm_neon.h: Fix softp typo.
19247 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19249 PR tree-optimization/81365
19250 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
19251 aggregate moves onto bb predecessor edges, make sure there are no
19252 loads that could alias the lhs in between the start of bb and the
19255 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
19258 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
19259 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
19260 [LSHIFTRT, outer_code = TRUNCATE]: Same.
19262 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19264 PR tree-optimization/81396
19265 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
19266 (init_symbolic_number): Initialize it to 1.
19267 (perform_symbolic_merge): Add n_ops from both operands into the new
19269 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
19270 without base_addr as useless if they need more than one operation.
19271 (bswap_replace): Handle !bswap case for NULL base_addr.
19273 2017-07-17 Tom de Vries <tom@codesourcery.com>
19276 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
19279 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
19281 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
19282 conditional builtin define __FIX_LEON3FT_B2BST.
19284 2017-07-17 Daniel Cederman <cederman@gaisler.com>
19286 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
19287 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
19290 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
19292 PR rtl-optimization/81424
19293 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
19294 to remove potential trapping from operands if -fnon-call-exceptions.
19296 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19298 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
19299 profile_proability for scalling.
19300 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
19302 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19304 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
19306 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19308 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
19309 fixpoint arithmetics.
19311 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19313 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
19314 fixpoint arithmetics.
19316 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19318 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
19319 fixpoint arithmetics.
19321 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19323 * profile-count.h (profile_probability::from_reg_br_prob_note,
19324 profile_probability::to_reg_br_prob_note): New functions.
19325 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
19326 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
19327 * predict.c (probability_reliable_p): Update.
19328 (edge_probability_reliable_p): Update.
19329 (br_prob_note_reliable_p): Update.
19330 (invert_br_probabilities): Update.
19331 (add_reg_br_prob_note): New function.
19332 (combine_predictions_for_insn): Update.
19333 * asan.c (asan_clear_shadow): Update.
19334 * cfgbuild.c (compute_outgoing_frequencies): Update.
19335 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
19336 (update_br_prob_note): Update.
19337 (rtl_verify_edges): Update.
19338 (purge_dead_edges): Update.
19339 (fixup_reorder_chain): Update.
19340 * emit-rtl.c (try_split): Update.
19341 * ifcvt.c (cond_exec_process_insns): Update.
19342 (cond_exec_process_if_block): Update.
19343 (dead_or_predicable): Update.
19344 * internal-fn.c (expand_addsub_overflow): Update.
19345 (expand_neg_overflow): Update.
19346 (expand_mul_overflow): Update.
19347 * loop-doloop.c (doloop_modify): Update.
19348 * loop-unroll.c (compare_and_jump_seq): Update.
19349 * optabs.c (emit_cmp_and_jump_insn_1): Update.
19350 * predict.h: Update.
19351 * reorg.c (mostly_true_jump): Update.
19353 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
19354 * config/alpha/alpha.c (emit_unlikely_jump): Update.
19355 * config/arc/arc.c: (emit_unlikely_jump): Update.
19356 * config/arm/arm.c: (emit_unlikely_jump): Update.
19357 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
19358 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
19359 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
19360 (ix86_print_operand): Update.
19361 (ix86_split_fp_branch): Update.
19362 (predict_jump): Update.
19363 * config/ia64/ia64.c (ia64_print_operand): Update.
19364 * config/mmix/mmix.c (mmix_print_operand): Update.
19365 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
19366 (rs6000_expand_split_stack_prologue): Update.
19367 * config/rs6000/rs6000.c: Update.
19368 * config/s390/s390.c (s390_expand_vec_strlen): Update.
19369 (s390_expand_vec_movstr): Update.
19370 (s390_expand_cs_tdsi): Update.
19371 (s390_expand_split_stack_prologue): Update.
19372 * config/sh/sh.c (sh_print_operand): Update.
19373 (expand_cbranchsi4): Update.
19374 (expand_cbranchdi4): Update.
19375 * config/sparc/sparc.c (output_v9branch): Update.
19376 * config/spu/spu.c (get_branch_target): Update.
19377 (ea_load_store_inline): Update.
19378 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
19379 * config/tilepro/tilepro.c: Update.
19381 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
19383 * gimplify.c (mostly_copy_tree_r): Revert latest change.
19384 (gimplify_save_expr): Likewise.
19386 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19388 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
19390 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19392 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
19394 * timevar.def (TV_IPA_FNSUMMARY): Define.
19396 2017-07-16 Daniel Cederman <cederman@gaisler.com>
19398 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
19399 to back store errata sensitive sequence from being generated.
19400 (sqrtdf2_fix): Likewise.
19402 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19404 * tree-ssa-threadupdate.c (compute_path_counts,
19405 update_joiner_offpath_counts): Use profile_probability.
19407 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
19410 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19412 * config/arm/arm-c.c (arm_cpu_builtins): Define
19413 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
19415 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
19417 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19418 array entries to represent __ieee128 versions of the
19419 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
19420 scalar_extract_sig, and scalar_insert_exp built-in functions.
19421 (altivec_resolve_overloaded_builtin): Add special case handling
19422 for the __builtin_scalar_insert_exp function, as represented by
19423 the P9V_BUILTIN_VEC_VSIEDP constant.
19424 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
19425 exponent support for __ieee128 argument.
19426 (VSESQP): Add scalar extract signature support for __ieee128
19428 (VSTDCNQP): Add scalar test negative support for __ieee128
19430 (VSIEQP): Add scalar insert exponent support for __int128 argument
19431 with __ieee128 result.
19432 (VSIEQPF): Add scalar insert exponent support for __ieee128
19433 argument with __ieee128 result.
19434 (VSTDCQP): Add scalar test data class support for __ieee128
19436 (VSTDCNQP): Add overload support for scalar test negative with
19437 __ieee128 argument.
19438 (VSTDCQP): Add overload support for scalar test data class
19439 __ieee128 argument.
19440 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
19441 UNSPEC_VSX_SXSIGDP.
19442 (UNSPEC_VSX_SIEXPQP): New constant.
19443 (xsxexpqp): New insn for VSX scalar extract exponent quad
19445 (xsxsigqp): New insn for VSX scalar extract significand quad
19447 (xsiexpqpf): New insn for VSX scalar insert exponent quad
19448 precision with floating point argument.
19449 (xststdcqp): New expand for VSX scalar test data class quad
19451 (xststdcnegqp): New expand for VSX scalar test negative quad
19453 (xststdcqp): New insn to match expansions for VSX scalar test data
19454 class quad precision and VSX scalar test negative quad precision.
19455 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
19456 special case operand checking to enforce that second operand of
19457 VSX scalar test data class with quad precision argument is a 7-bit
19459 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19460 prototypes and descriptions of __ieee128 versions of
19461 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
19462 scalar_test_data_class, and scalar_test_neg built-in functions.
19464 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19466 PR tree-optimization/81162
19467 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
19468 replace a negate with an add.
19470 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
19472 * doc/invoke.texi (arm/-mcpu): Document +crypto.
19474 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19476 * config/arm/arm-c.c (arm_cpu_builtins): Define
19477 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
19479 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19481 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
19482 (armv8-r): Set ARM Cortex-R52 as default CPU.
19483 * config/arm/arm-tables.opt: Regenerate.
19484 * config/arm/arm-tune.md: Regenerate.
19485 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
19487 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
19488 extension for -mcpu=cortex-r52.
19490 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
19492 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
19493 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
19494 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
19495 (fp-armv8): Define it as FP_ARMv8 only.
19496 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
19497 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
19499 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
19500 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
19501 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
19502 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
19503 than TARGET_FPU_ARMV8.
19504 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
19505 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
19506 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
19508 * config/arm/neon.md (neon_vrint): Likewise.
19509 (neon_vcvt): Likewise.
19510 (neon_<fmaxmin_op><mode>): Likewise.
19511 (<fmaxmin><mode>3): Likewise.
19512 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
19513 * config/arm/predicates.md (arm_cond_move_operator): Check against
19514 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
19516 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
19518 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
19519 to top of function.
19521 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19523 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
19524 loop in comment with memset.
19526 2017-07-14 Martin Liska <mliska@suse.cz>
19528 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
19529 * dwarf2out.c (is_java): Remove the function.
19530 (output_pubname): Remove usage of the function.
19531 (lower_bound_default): Remove usage of DW_LANG_Java.
19532 (gen_compile_unit_die): Likewise.
19533 * gcc.c: Remove compiler defaults for .java and .zip files.
19534 * gimple-expr.c (remove_suffix): Change as there's no longer
19535 extension than 4-letter one.
19536 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
19537 (gimplify_save_expr): Likewise.
19538 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
19539 as it's possible even for other languages than Java.
19540 * langhooks.h (struct lang_hooks): Remove Java from a comment.
19541 * lto-opts.c (lto_write_options): Remove reference to Java.
19542 * opts.c (strip_off_ending): Update file extension handling.
19543 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
19544 * tree-eh.c (lower_resx): Likewise.
19545 * tree.c (free_lang_data_in_type): Remove dead code.
19546 (find_decls_types_r): Likewise.
19547 (build_common_builtin_nodes): Remove Java from a comment.
19548 (verify_type): Remove dead code.
19549 * varasm.c (assemble_external): Remove Java from a comment.
19551 2017-07-14 Martin Liska <mliska@suse.cz>
19553 * opts.c (finish_options): Add quotes.
19554 (common_handle_option): Likewise.
19556 2017-07-14 Martin Liska <mliska@suse.cz>
19558 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
19559 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
19560 Remove N_SO_PASCAL.
19561 * dwarf2out.c (lower_bound_default): Do not handle
19563 (gen_compile_unit_die): Likewise.
19564 * gcc.c: Remove default extension binding for GNU Pascal.
19565 * stmt.c: Remove Pascal language from a comment.
19566 * xcoffout.c: Likewise.
19568 2017-07-13 David Malcolm <dmalcolm@redhat.com>
19571 * diagnostic-show-locus.c (fixit_cmp): New function.
19572 (layout::layout): Sort m_fixit_hints.
19573 (column_range::column_range): Assert that the values are valid.
19574 (struct char_span): New struct.
19575 (correction::overwrite): New method.
19576 (struct source_line): New struct.
19577 (line_corrections::add_hint): Add assertions. Reimplement memcpy
19578 calls in terms of classes source_line and char_span, and
19579 correction::overwrite.
19580 (selftest::test_overlapped_fixit_printing_2): New function.
19581 (selftest::diagnostic_show_locus_c_tests): Call it.
19583 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
19585 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
19586 early if there is no lhs.
19588 2017-07-13 Martin Liska <mliska@suse.cz>
19590 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
19591 (gen_reference_type_die): Likewise.
19592 * stor-layout.c: Remove Pascal-related comment.
19594 2017-07-13 Martin Liska <mliska@suse.cz>
19596 * opts.c (finish_options): Add quotes to error messages.
19597 (parse_sanitizer_options): Likewise.
19599 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
19601 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
19603 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
19605 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
19607 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
19609 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
19611 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
19613 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
19616 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
19617 provides the hardware capability bits, define the macro
19618 __BUILTIN_CPU_SUPPORTS__.
19619 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
19620 if GLIBC does not provide the hardware capability bits. Add a
19621 gcc_unreachable call if the built-in cpu function is neither
19622 __builtin_cpu_is nor __builtin_cpu_supports.
19623 (rs6000_get_function_versions_dispatcher): Change the warning
19624 that an old GLIBC is used which does not export the capability
19625 bits to be an error.
19626 * doc/extend.texi (target_clones attribute): Document the
19627 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
19628 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
19629 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
19630 the macros defined by GCC if the newer GLIBC is available.
19632 2017-07-12 Jeff Law <law@redhat.com>
19634 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
19635 remaining includes slightly.
19636 * config/riscv/riscv-builtins.c: Include profile-count.h.
19638 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
19641 * config/avr/avr.c (avr_set_current_function): In diagnostic
19642 messages: Quote keywords and (parts of) identifiers.
19643 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
19646 2017-07-12 Carl Love <cel@us.ibm.com>
19648 * config/rs6000/rs6000-c.c: Add support for built-in functions
19649 vector bool char vec_revb (vector bool char);
19650 vector bool short vec_revb (vector short char);
19651 vector bool int vec_revb (vector bool int);
19652 vector bool long long vec_revb (vector bool long long);
19653 * doc/extend.texi: Update the built-in documentation file for the
19654 new built-in functions.
19656 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19658 * config/s390/s390.md: Remove movcc splitter.
19660 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19662 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
19663 load/store on condition.
19665 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
19668 * config/avr/avr.c (avr_encode_section_info)
19669 [progmem && !TREE_READONLY]: Error if progmem object needs
19672 2017-07-11 Michael Collison <michael.collison@arm.com>
19674 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
19677 2017-07-11 Carl Love <cel@us.ibm.com>
19679 * config/rs6000/rs6000-c.c: Add support for builtins
19680 vector unsigned int vec_parity_lsbb (vector signed int);
19681 vector unsigned int vec_parity_lsbb (vector unsigned int);
19682 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
19683 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
19684 vector unsigned long long vec_parity_lsbb (vector signed long long);
19685 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
19686 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
19687 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
19688 * doc/extend.texi: Update the built-in documentation file for the
19689 new built-in functions.
19691 2017-07-11 David Malcolm <dmalcolm@redhat.com>
19693 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
19694 (layout::m_primary_loc): New field.
19695 (layout::layout): Initialize new field. Move location filtering
19696 logic from here to...
19697 (layout::maybe_add_location_range): ...this new method. Add
19698 support for filtering to just the lines already specified by other
19700 (layout::will_show_line_p): New method.
19701 (gcc_rich_location::add_location_if_nearby): New method.
19702 (selftest::test_add_location_if_nearby): New test function.
19703 (selftest::diagnostic_show_locus_c_tests): Call it.
19704 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
19707 2017-07-11 Tom de Vries <tom@codesourcery.com>
19709 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
19710 (bb_first_real_insn): New function.
19711 (nvptx_single): Add extra initialization of broadcasted condition
19714 2017-07-11 Nathan Sidwell <nathan@acm.org>
19716 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
19718 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
19720 * doc/extend.texi (AVR Function Attributes): Remove weblink to
19721 Binutils doc as TEXI will mess them up.
19722 * doc/invoke.texi (AVR Options): Same here.
19724 2017-07-11 Daniel Cederman <cederman@gaisler.com>
19726 * config/sparc/sparc.opt (mfix-ut700): New option.
19727 (mfix-gr712rc): Likewise.
19728 (sparc_fix_b2bst): New variable.
19729 * doc/invoke.texi (SPARC options): Document them.
19730 (ARM options): Fix warnings.
19731 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
19732 instructions to prevent sequences that can trigger the store-store
19733 errata for certain LEON3FT processors.
19734 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
19735 (sparc_option_override): Set sparc_fix_b2bst appropriately.
19736 * config/sparc/sparc.md (fix_b2bst): New attribute.
19737 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
19739 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
19742 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
19744 (*rsqrtsf2_sse): Ditto.
19746 (div<mode>3): Macroize insn from divdf3 and divsf3
19747 using MODEF mode iterator.
19749 2017-07-10 Martin Sebor <msebor@redhat.com>
19751 PR tree-optimization/80397
19752 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
19753 instead of testing for equality to INTEGER_TYPE.
19755 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
19757 * config.gcc: Remove uclibc from arc target spec.
19759 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
19761 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
19763 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
19766 * lto-wrapper.c (remove_option): New function.
19767 (merge_and_complain): Merge PIC/PIE options more realistically.
19769 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
19771 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
19775 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
19776 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
19777 * config.in: Regenerate.
19778 * configure: Regenerate.
19779 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
19780 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
19781 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
19782 (TARGET_GASISR_PROLOGUES): ...target mask.
19783 * common/config/avr/avr-common.c
19784 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
19785 Set -mgas-isr-prologues.
19786 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
19787 INSERT_PASS_BEFORE for it.
19788 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
19789 * config/avr/avr.c (avr_option_override)
19790 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
19791 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
19792 (avr_attribute_table) <no_gccisr>: Add new function attribute.
19793 (avr_set_current_function) <is_no_gccisr>: Init machine field.
19794 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
19796 (make_avr_pass_pre_proep): New function.
19797 (emit_push_sfr) <treg>: Add argument to function and use it
19798 instead of TMP_REG.
19799 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
19800 and set machine->gasisr.yes.
19801 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
19802 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
19803 __gcc_isr.n_pushed to .L__stack_usage.
19804 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
19805 (avr_asm_final_postscan_insn): ...this new static function.
19806 * config/avr/avr.h (machine_function)
19807 <is_no_gccisr, use_L__stack_usage>: New fields.
19808 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
19809 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
19810 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
19811 (gasisr, *gasisr): New expander and insn.
19812 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
19813 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
19814 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
19816 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
19818 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
19821 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
19823 Move jump-tables out of .text again.
19826 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
19827 (ASM_OUTPUT_ADDR_VEC): New function.
19828 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
19829 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
19830 INSN_ADDRESSes as asm comment.
19831 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
19832 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
19833 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
19834 * config/avr/avr.md (*tablejump): Adjust comment.
19835 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
19836 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
19838 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
19839 (avr_output_addr_vec): New proto.
19840 (avr_log_t) <insn_addresses>: New field.
19842 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
19845 * config/i386/i386.c (ix86_function_arg_advance): Set
19846 outgoing_args_on_stack to true if there are outgoing arguments
19848 (ix86_function_arg): Likewise.
19849 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
19850 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
19851 * config/i386/i386.h (machine_function): Add
19852 outgoing_args_on_stack.
19854 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
19856 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
19857 supporting pthreds.
19858 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
19860 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
19862 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
19863 (REAL_H): Remove $(MACHMODE_H).
19864 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
19866 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
19867 $(MACHMODE_H) and double-int.h.
19868 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
19870 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
19873 2017-07-07 Andrew Pinski <apinski@cavium.com>
19875 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
19876 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
19878 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
19880 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
19881 Add warning if GCC was not configured to link against a GLIBC that
19882 exports the hardware capability bits.
19883 (make_resolver_func): Make resolver function private and not a
19884 COMDAT function. Create the name with clone_function_name instead
19885 of make_unique_name.
19888 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
19889 correct operand in doing the split.
19891 2017-07-07 Carl Love <cel@us.ibm.com>
19893 * config/rs6000/rs6000-c: Add support for built-in function
19894 vector unsigned short vec_pack_to_short_fp32 (vector float,
19896 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
19897 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
19898 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
19899 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
19900 (convert_4f32_8i16): Add define_expand.
19901 * doc/extend.texi: Update the built-in documentation file for the
19902 new built-in function.
19904 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19906 * config/sparc/m8.md: New file.
19907 * config/sparc/sparc.md: Include m8.md.
19909 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19911 * config/sparc/sparc.opt: New option -mvis4b.
19912 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
19913 (sparc_option_override): Handle VIS4B.
19914 (enum sparc_builtins): Define
19915 SPARC_BUILTIN_DICTUNPACK{8,16,32},
19916 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
19917 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
19918 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
19919 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
19920 (check_constant_argument): New function.
19921 (sparc_vis_init_builtins): Define builtins
19922 __builtin_vis_dictunpack{8,16,32},
19923 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
19924 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
19925 __builtin_vis_fpcmpde{8,16,32}shl and
19926 __builtin_vis_fpcmpur{8,16,32}shl.
19927 (sparc_expand_builtin): Check that the constant operands to
19928 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
19929 constant and in range.
19930 * config/sparc/sparc-c.c (sparc_target_macros): Handle
19932 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
19933 (SPARC_IMM5_P): Likewise.
19934 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
19935 (enabled): Handle vis4b.
19936 (UNSPEC_DICTUNPACK): New unspec.
19937 (UNSPEC_FPCMPSHL): Likewise.
19938 (UNSPEC_FPUCMPSHL): Likewise.
19939 (UNSPEC_FPCMPDESHL): Likewise.
19940 (UNSPEC_FPCMPURSHL): Likewise.
19941 (cpu_feature): New CPU feature `vis4b'.
19942 (dictunpack{8,16,32}): New insns.
19943 (FPCSMODE): New mode iterator.
19944 (fpcscond): New code iterator.
19945 (fpcsucond): Likewise.
19946 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
19947 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
19948 (fpcmpde{8,16,32}{si,di}shl): Likewise.
19949 (fpcmpur{8,16,32}{si,di}shl): Likewise.
19950 * config/sparc/constraints.md: Define constraints `q' for unsigned
19951 2-bit integer constants and `t' for unsigned 5-bit integer
19953 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
19955 (imm5_operand_dictunpack16): Likewise.
19956 (imm5_operand_dictunpack32): Likewise.
19957 (imm2_operand): Likewise.
19958 * doc/invoke.texi (SPARC Options): Document -mvis4b.
19959 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
19960 ditunpack* and fpcmp*shl builtins.
19962 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19964 * config.gcc: Handle m8 in --with-{cpu,tune} options.
19965 * config.in: Add HAVE_AS_SPARC6 define.
19966 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
19968 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
19970 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
19971 (CPP_CPU_SPEC): Handle m8.
19972 (ASM_CPU_SPEC): Likewise.
19973 * config/sparc/sparc-opts.h (enum processor_type): Add
19975 * config/sparc/sparc.c (m8_costs): New struct.
19976 (sparc_option_override): Handle TARGET_CPU_m8.
19977 (sparc32_initialize_trampoline): Likewise.
19978 (sparc64_initialize_trampoline): Likewise.
19979 (sparc_issue_rate): Likewise.
19980 (sparc_register_move_cost): Likewise.
19981 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
19982 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
19983 (ASM_CPU64_DEFAULT_SPEC): Likewise.
19984 (CPP_CPU_SPEC): Handle M8.
19985 (ASM_CPU_SPEC): Likewise.
19986 (AS_M8_FLAG): Define.
19987 * config/sparc/sparc.md: Add m8 to the cpu attribute.
19988 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
19989 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
19991 * configure: Regenerate.
19992 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
19995 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
19997 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
19999 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
20000 ("*movdi_insn_sp32"): Do not set v3pipe.
20001 ("*movsi_insn"): Likewise.
20002 ("*movdi_insn_sp64"): Likewise.
20003 ("*movsf_insn"): Likewise.
20004 ("*movdf_insn_sp32"): Likewise.
20005 ("*movdf_insn_sp64"): Likewise.
20006 ("*zero_extendsidi2_insn_sp64"): Likewise.
20007 ("*sign_extendsidi2_insn"): Likewise.
20008 ("*mov<VM32:mode>_insn"): Likewise.
20009 ("*mov<VM64:mode>_insn_sp64"): Likewise.
20010 ("*mov<VM64:mode>_insn_sp32"): Likewise.
20011 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
20012 ("<vlop:code><VL:mode>3"): Likewise.
20013 ("*not_<vlop:code><VL:mode>3"): Likewise.
20014 ("*nand<VL:mode>_vis"): Likewise.
20015 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
20016 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
20017 ("one_cmpl<VL:mode>2"): Likewise.
20018 ("faligndata<VM64:mode>_vis"): Likewise.
20019 ("alignaddrsi_vis"): Likewise.
20020 ("alignaddrdi_vis"): Likweise.
20021 ("alignaddrlsi_vis"): Likewise.
20022 ("alignaddrldi_vis"): Likewise.
20023 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
20024 ("bmaskdi_vis"): Likewise.
20025 ("bmasksi_vis"): Likewise.
20026 ("bshuffle<VM64:mode>_vis"): Likewise.
20027 ("cmask8<P:mode>_vis"): Likewise.
20028 ("cmask16<P:mode>_vis"): Likewise.
20029 ("cmask32<P:mode>_vis"): Likewise.
20030 ("pdistn<P:mode>_vis"): Likewise.
20031 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
20033 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20035 * config/sparc/sparc.md ("subtype"): New insn attribute.
20036 ("*wrgsr_sp64"): Set insn subtype.
20037 ("*rdgsr_sp64"): Likewise.
20038 ("alignaddrsi_vis"): Likewise.
20039 ("alignaddrdi_vis"): Likewise.
20040 ("alignaddrlsi_vis"): Likewise.
20041 ("alignaddrldi_vis"): Likewise.
20042 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
20043 ("fexpand_vis"): Likewise.
20044 ("fpmerge_vis"): Likewise.
20045 ("faligndata<VM64:mode>_vis"): Likewise.
20046 ("bshuffle<VM64:mode>_vis"): Likewise.
20047 ("cmask8<P:mode>_vis"): Likewise.
20048 ("cmask16<P:mode>_vis"): Likewise.
20049 ("cmask32<P:mode>_vis"): Likewise.
20050 ("fchksm16_vis"): Likewise.
20051 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
20052 ("fmean16_vis"): Likewise.
20053 ("fp<plusminus_insn>64_vis"): Likewise.
20054 ("<plusminus_insn>v8qi3"): Likewise.
20055 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
20056 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
20057 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
20058 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
20059 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
20060 ("*movqi_insn"): Likewise.
20061 ("*movhi_insn"): Likewise.
20062 ("*movsi_insn"): Likewise.
20063 ("movsi_pic_gotdata_op"): Likewise.
20064 ("*movdi_insn_sp32"): Likewise.
20065 ("*movdi_insn_sp64"): Likewise.
20066 ("movdi_pic_gotdata_op"): Likewise.
20067 ("*movsf_insn"): Likewise.
20068 ("*movdf_insn_sp32"): Likewise.
20069 ("*movdf_insn_sp64"): Likewise.
20070 ("*zero_extendhisi2_insn"): Likewise.
20071 ("*zero_extendqihi2_insn"): Likewise.
20072 ("*zero_extendqisi2_insn"): Likewise.
20073 ("*zero_extendqidi2_insn"): Likewise.
20074 ("*zero_extendhidi2_insn"): Likewise.
20075 ("*zero_extendsidi2_insn_sp64"): Likewise.
20076 ("ldfsr"): Likewise.
20077 ("prefetch_64"): Likewise.
20078 ("prefetch_32"): Likewise.
20079 ("tie_ld32"): Likewise.
20080 ("tie_ld64"): Likewise.
20081 ("*tldo_ldub_sp32"): Likewise.
20082 ("*tldo_ldub1_sp32"): Likewise.
20083 ("*tldo_ldub2_sp32"): Likewise.
20084 ("*tldo_ldub_sp64"): Likewise.
20085 ("*tldo_ldub1_sp64"): Likewise.
20086 ("*tldo_ldub2_sp64"): Likewise.
20087 ("*tldo_ldub3_sp64"): Likewise.
20088 ("*tldo_lduh_sp32"): Likewise.
20089 ("*tldo_lduh1_sp32"): Likewise.
20090 ("*tldo_lduh_sp64"): Likewise.
20091 ("*tldo_lduh1_sp64"): Likewise.
20092 ("*tldo_lduh2_sp64"): Likewise.
20093 ("*tldo_lduw_sp32"): Likewise.
20094 ("*tldo_lduw_sp64"): Likewise.
20095 ("*tldo_lduw1_sp64"): Likewise.
20096 ("*tldo_ldx_sp64"): Likewise.
20097 ("*mov<VM32:mode>_insn"): Likewise.
20098 ("*mov<VM64:mode>_insn_sp64"): Likewise.
20099 ("*mov<VM64:mode>_insn_sp32"): Likewise.
20101 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20103 * config/sparc/sparc.md ("type"): New insn type viscmp.
20104 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
20106 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
20107 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
20108 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
20109 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
20111 ("n7_vis_logical_11cycle"): Likewise.
20112 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
20113 * config/sparc/niagara2.md ("niag3_vis": Likewise.
20114 * config/sparc/niagara.md ("niag_vis"): Likewise.
20115 * config/sparc/ultra3.md ("us3_fga"): Likewise.
20116 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
20118 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20120 * config/sparc/sparc.md: New instruction type `bmask'.
20121 (bmaskdi_vis): Use the `bmask' type.
20122 (bmasksi_vis): Likewise.
20123 * config/sparc/ultra3.md (us3_array): Likewise.
20124 * config/sparc/niagara7.md (n7_array): Likewise.
20125 * config/sparc/niagara4.md (n4_array): Likewise.
20126 * config/sparc/niagara2.md (niag2_vis): Likewise.
20127 (niag3_vis): Likewise.
20128 * config/sparc/niagara.md (niag_vis): Likewise.
20130 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20132 * ipa-comdats.c: Remove optimize check from gate.
20133 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
20134 for functions not optimized.
20135 (ipa_fn_summary_read): Skip optimize check.
20136 (ipa_fn_summary_write): Likewise.
20137 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
20139 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
20141 (can_inline_edge_p): Check flag_pcc_struct_return for match.
20142 (check_callers): Give up on caller which is not optimized.
20143 (inline_small_functions): Likewise.
20144 (ipa_inline): Do not give up when not optimizing.
20145 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
20146 away unoptimizes cdtors.
20147 (whole_program_function_and_variable_visibility): Do
20148 ipa_discover_readonly_nonaddressable_vars in LTO mode.
20149 * ipa.c (process_references): Do not check optimize.
20150 (symbol_table::remove_unreachable_nodes): Update optimize check.
20151 (set_writeonly_bit): Update optimize check.
20152 (pass_ipa_cdtor_merge::gate): Do not check optimize.
20153 (pass_ipa_single_use::gate): Remove.
20155 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20157 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
20158 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
20159 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
20160 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
20161 permute_load, permute_store, adjust_extract, adjust_splat,
20162 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
20163 replace_swap_with_copy, dump_swap_insn_table,
20164 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
20165 recombine_lvx_pattern, recombine_stvx_pattern,
20166 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
20167 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
20168 to file rs6000-p8swap.c.
20169 * config/rs6000/rs6000-p8swap.c: New file.
20170 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
20171 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
20172 and rs6000*-*-* targets.
20174 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20176 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
20178 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20180 * lto-wrapper.c (merge_and_complain): Do not merge
20181 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
20182 fsigned_zeros, ftrapping_math, fwrapv.
20183 (append_compiler_options): Do not track these options.
20184 (append_linker_options): Likewie
20186 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20188 * cgraphunit.c (cgraph_node::finalize_function): When
20189 !flag_toplevel_reorde set no_reorder flag.
20190 (varpool_node::finalize_decl): Likewise.
20191 (symbol_table::compile): Drop no toplevel reorder path.
20193 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20195 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
20196 edges; zero probability is not better than uninitialized.
20198 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
20200 * asan.h (asan_sanitize_allocas_p): Declare.
20201 * asan.c (asan_sanitize_allocas_p): New function.
20202 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
20203 (handle_builtin_alloca): Likewise.
20204 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
20205 if !asan_sanitize_allocas_p.
20206 * params.def (asan-instrument-allocas): Add new option.
20207 * params.h (ASAN_PROTECT_ALLOCAS): Define.
20208 * opts.c (common_handle_option): Disable allocas sanitization for
20211 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
20213 * asan.c: Include gimple-fold.h.
20214 (get_last_alloca_addr): New function.
20215 (handle_builtin_stackrestore): Likewise.
20216 (handle_builtin_alloca): Likewise.
20217 (asan_emit_allocas_unpoison): Likewise.
20218 (get_mem_refs_of_builtin_call): Add new parameter, remove const
20219 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
20220 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
20221 (instrument_builtin_call): Pass gimple iterator to
20222 get_mem_refs_of_builtin_call.
20223 (last_alloca_addr): New global.
20224 * asan.h (asan_emit_allocas_unpoison): Declare.
20225 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
20226 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
20227 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
20228 if function calls alloca.
20229 * gimple-fold.c (replace_call_with_value): Remove static keyword.
20230 * gimple-fold.h (replace_call_with_value): Declare.
20231 * internal-fn.c: Include asan.h.
20232 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
20233 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
20235 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20237 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
20238 (C_SELFTEST_FLAGS): New.
20239 (CPP_SELFTEST_FLAGS): New.
20240 (SELFTEST_DEPS): New, from deps of s-selftest.
20241 (C_SELFTEST_DEPS): New, from deps of s-selftest.
20242 (CPP_SELFTEST_DEPS): New.
20243 (selftest): Add dependency on s-selftest-c++.
20244 (s-selftest): Rename to...
20245 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
20246 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
20247 than SELFTEST_FLAGS.
20248 (selftest-gdb): Rename to...
20249 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
20251 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
20252 (selftest-valgrind): Rename to...
20253 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
20255 (selftest-valgrind): Reintroduce as an alias for
20256 selftest-c-valgrind.
20257 (s-selftest-c++): New.
20258 (selftest-c++-gdb): New.
20259 (selftest-c++-valgrind): New.
20261 2017-07-06 Olivier Hainque <hainque@adacore.com>
20263 * gcc.c (process_command): When deciding if undefined variables
20264 should be ignored when processing specs, accept "gcc -v" as well.
20266 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20268 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
20269 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
20271 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
20273 * config/arm/arm-cpus.in (armv8-r): Add new entry.
20274 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
20275 * config/arm/arm-tables.opt: Regenerate.
20276 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
20278 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
20280 2017-07-06 Carl Love <cel@us.ibm.com>
20282 * ChangeLog: Clean up from mid air collision
20284 2017-07-06 Carl Love <cel@us.ibm.com>
20286 * config/rs6000/rs6000-c.c: Add support for built-in functions
20287 vector signed int vec_subc (vector signed int, vector signed int);
20288 vector signed __int128 vec_subc (vector signed __int128,
20289 vector signed __int128);
20290 vector unsigned __int128 vec_subc (vector unsigned __int128,
20291 vector unsigned __int128);
20292 vector signed int vec_sube (vector signed int, vector signed int,
20293 vector signed int);
20294 vector unsigned int vec_sube (vector unsigned int,
20295 vector unsigned int,
20296 vector unsigned int);
20297 vector signed __int128 vec_sube (vector signed __int128,
20298 vector signed __int128,
20299 vector signed__int128);
20300 vector unsigned __int128 vec_sube (vector unsigned __int128,
20301 vector unsigned __int128,
20302 vector unsigned __int128);
20303 vector signed int vec_subec (vector signed int, vector signed int,
20304 vector signed int);
20305 vector unsigned int vec_subec (vector unsigned int,
20306 vector unsigned int,
20307 vector unsigned int);
20308 vector signed __int128 vec_subec (vector signed __int128,
20309 vector signed __int128,
20310 vector signed__int128);
20311 vector unsigned __int128 vec_subec (vector unsigned __int128,
20312 vector unsigned __int128,
20313 vector unsigned __int128);
20314 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
20315 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
20316 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
20317 BU_ALTIVEC_OVERLOAD_X definitions.
20318 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
20319 * doc/extend.texi: Update the built-in documentation file for the new
20320 built-in functions.
20322 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20325 * diagnostic-show-locus.c (layout::layout): Use start and finish
20326 spelling location for the start and finish of each range.
20327 * genmatch.c (linemap_client_expand_location_to_spelling_point):
20328 Add unused aspect param.
20329 * input.c (expand_location_1): Add "aspect" param, and use it
20330 to access the correct part of the location.
20331 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
20333 (expand_location_to_spelling_point): Likewise.
20334 (linemap_client_expand_location_to_spelling_point): Add "aspect"
20335 param, and pass it to expand_location_1.
20337 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
20339 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
20340 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
20341 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
20342 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
20343 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
20344 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
20345 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
20346 _mm_maskz_getmant_ss): New intrinsics.
20347 (__builtin_ia32_getexpss128_mask): Changed to ...
20348 __builtin_ia32_getexpss128_round ... this.
20349 (__builtin_ia32_getexpsd128_mask): Changed to ...
20350 __builtin_ia32_getexpsd128_round ... this.
20351 * config/i386/i386-builtin-types.def
20352 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
20353 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
20354 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
20355 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
20356 __builtin_ia32_getmantss_mask_round): New builtins.
20357 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
20358 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
20359 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
20360 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
20361 * config/i386/sse.md
20362 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
20363 avx512f_sgetexp<mode><mask_scalar_name>
20364 <round_saeonly_scalar_name> ... this.
20365 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
20366 %0, %1, %2<round_saeonly_op3>}): Changed to ...
20367 vgetexp<ssescalarmodesuffix>
20368 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20369 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
20370 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
20371 avx512f_vgetmant<mode><mask_scalar_name>
20372 <round_saeonly_scalar_name> ... this.
20373 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
20374 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
20375 vgetmant<ssescalarmodesuffix>
20376 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
20377 %0<mask_scalar_operand4>, %1, %2
20378 <round_saeonly_scalar_mask_op4>, %3} ... this.
20379 * config/i386/subst.md (mask_scalar_operand4,
20380 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
20381 round_saeonly_scalar_nimm_predicate): New subst attributes.
20383 2017-07-06 Julia Koval <julia.koval@intel.com>
20385 * config/i386/i386.c (ix86_erase_embedded_rounding):
20386 Remove code for old rounding pattern.
20388 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
20390 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
20392 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
20394 * doc/sourcebuild.texi (Test Directives, Variants of
20395 dg-require-support): Add documentation for dg-require-stack-check.
20397 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
20399 * config/i386/subst.md (mask_scalar, round_scalar,
20400 round_saeonly_scalar): New meta-templates.
20401 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
20402 round_scalar_mask_operand3, round_scalar_mask_op3,
20403 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
20404 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
20405 round_saeonly_scalar_constraint,
20406 round_saeonly_scalar_prefix): New subst attribute.
20407 * config/i386/sse.md
20408 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
20409 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
20410 <round_scalar_name> ... this.
20411 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
20412 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
20413 <round_scalar_name> ... this.
20414 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
20415 <sse>_vm<code><mode>3<mask_scalar_name>
20416 <round_saeonly_scalar_name> ... this.
20417 (v<plusminus_mnemonic><ssescalarmodesuffix>
20418 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
20419 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
20420 v<plusminus_mnemonic><ssescalarmodesuffix>
20421 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20422 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
20423 (v<multdiv_mnemonic><ssescalarmodesuffix>
20424 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
20425 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
20426 v<multdiv_mnemonic><ssescalarmodesuffix>
20427 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20428 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
20429 (v<maxmin_float><ssescalarmodesuffix>
20430 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
20431 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
20432 v<maxmin_float><ssescalarmodesuffix>
20433 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20434 %0<mask_scalar_operand3>, %1, %<iptr>2
20435 <round_saeonly_scalar_mask_op3>} ... this.
20437 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
20439 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
20440 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
20442 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
20443 Alan Hayward <alan.hayward@arm.com>
20444 David Sherwood <david.sherwood@arm.com>
20446 * combine.c (simplify_if_then_else): Remove "enum" before
20448 * compare-elim.c (can_eliminate_compare): Likewise.
20449 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
20451 (aarch64_lookup_simd_builtin_type): Likewise.
20452 (aarch64_simd_builtin_type): Likewise.
20453 (aarch64_init_simd_builtin_types): Likewise.
20454 (aarch64_simd_expand_args): Likewise.
20455 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
20457 (aarch64_reverse_mask): Likewise.
20458 (aarch64_simd_emit_reg_reg_move): Likewise.
20459 (aarch64_gen_adjusted_ldpstp): Likewise.
20460 (aarch64_ccmp_mode_to_code): Likewise.
20461 (aarch64_operands_ok_for_ldpstp): Likewise.
20462 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
20463 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
20465 (aarch64_min_divisions_for_recip_mul): Likewise.
20466 (aarch64_reassociation_width): Likewise.
20467 (aarch64_get_condition_code_1): Likewise.
20468 (aarch64_simd_emit_reg_reg_move): Likewise.
20469 (aarch64_simd_attr_length_rglist): Likewise.
20470 (aarch64_reverse_mask): Likewise.
20471 (aarch64_operands_ok_for_ldpstp): Likewise.
20472 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
20473 (aarch64_gen_adjusted_ldpstp): Likewise.
20474 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
20476 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
20477 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
20478 (arm_lookup_simd_builtin_type): Likewise.
20479 (arm_simd_builtin_type): Likewise.
20480 (arm_init_simd_builtin_types): Likewise.
20481 (arm_expand_builtin_args): Likewise.
20482 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
20483 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
20484 (ft32_setup_incoming_varargs): Likewise.
20485 (ft32_function_arg): Likewise.
20486 (ft32_function_arg_advance): Likewise.
20487 (ft32_pass_by_reference): Likewise.
20488 (ft32_arg_partial_bytes): Likewise.
20489 (ft32_valid_pointer_mode): Likewise.
20490 (ft32_addr_space_pointer_mode): Likewise.
20491 (ft32_addr_space_legitimate_address_p): Likewise.
20492 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
20494 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
20495 (ix86_emit_outlined_ms2sysv_restore): Likewise.
20496 (iamcu_alignment): Likewise.
20497 (canonicalize_vector_int_perm): Likewise.
20498 (ix86_noce_conversion_profitable_p): Likewise.
20499 (ix86_mpx_bound_mode): Likewise.
20500 (ix86_operands_ok_for_move_multiple): Likewise.
20501 * config/microblaze/microblaze-protos.h
20502 (microblaze_expand_conditional_branch_reg): Likewise.
20503 * config/microblaze/microblaze.c
20504 (microblaze_expand_conditional_branch_reg): Likewise.
20505 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
20507 (rs6000_reassociation_width): Likewise.
20508 (rs6000_invalid_binary_op): Likewise.
20509 (fusion_p9_p): Likewise.
20510 (emit_fusion_p9_load): Likewise.
20511 (emit_fusion_p9_store): Likewise.
20512 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
20514 (riscv_hard_regno_mode_ok_p): Likewise.
20515 (riscv_address_insns): Likewise.
20516 (riscv_split_symbol): Likewise.
20517 (riscv_legitimize_move): Likewise.
20518 (riscv_function_value): Likewise.
20519 (riscv_hard_regno_nregs): Likewise.
20520 (riscv_expand_builtin): Likewise.
20521 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
20522 (riscv_build_integer): Likewise.
20523 (riscv_split_integer): Likewise.
20524 (riscv_legitimate_constant_p): Likewise.
20525 (riscv_cannot_force_const_mem): Likewise.
20526 (riscv_regno_mode_ok_for_base_p): Likewise.
20527 (riscv_valid_base_register_p): Likewise.
20528 (riscv_valid_offset_p): Likewise.
20529 (riscv_valid_lo_sum_p): Likewise.
20530 (riscv_classify_address): Likewise.
20531 (riscv_legitimate_address_p): Likewise.
20532 (riscv_address_insns): Likewise.
20533 (riscv_load_store_insns): Likewise.
20534 (riscv_force_binary): Likewise.
20535 (riscv_split_symbol): Likewise.
20536 (riscv_force_address): Likewise.
20537 (riscv_legitimize_address): Likewise.
20538 (riscv_move_integer): Likewise.
20539 (riscv_legitimize_const_move): Likewise.
20540 (riscv_legitimize_move): Likewise.
20541 (riscv_address_cost): Likewise.
20542 (riscv_subword): Likewise.
20543 (riscv_output_move): Likewise.
20544 (riscv_canonicalize_int_order_test): Likewise.
20545 (riscv_emit_int_order_test): Likewise.
20546 (riscv_function_arg_boundary): Likewise.
20547 (riscv_pass_mode_in_fpr_p): Likewise.
20548 (riscv_pass_fpr_single): Likewise.
20549 (riscv_pass_fpr_pair): Likewise.
20550 (riscv_get_arg_info): Likewise.
20551 (riscv_function_arg): Likewise.
20552 (riscv_function_arg_advance): Likewise.
20553 (riscv_arg_partial_bytes): Likewise.
20554 (riscv_function_value): Likewise.
20555 (riscv_pass_by_reference): Likewise.
20556 (riscv_setup_incoming_varargs): Likewise.
20557 (riscv_print_operand): Likewise.
20558 (riscv_elf_select_rtx_section): Likewise.
20559 (riscv_save_restore_reg): Likewise.
20560 (riscv_for_each_saved_reg): Likewise.
20561 (riscv_register_move_cost): Likewise.
20562 (riscv_hard_regno_mode_ok_p): Likewise.
20563 (riscv_hard_regno_nregs): Likewise.
20564 (riscv_class_max_nregs): Likewise.
20565 (riscv_memory_move_cost): Likewise.
20566 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
20567 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
20568 (rl78_addr_space_address_mode): Likewise.
20569 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20571 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20572 (rs6000_reassociation_width): Likewise.
20573 (rs6000_invalid_binary_op): Likewise.
20574 (fusion_p9_p): Likewise.
20575 (emit_fusion_p9_load): Likewise.
20576 (emit_fusion_p9_store): Likewise.
20577 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
20578 (ok_for_simple_move_operands): Likewise.
20579 (ok_for_simple_move_strict_operands): Likewise.
20580 (ok_for_simple_arith_logic_operands): Likewise.
20581 (visium_legitimize_reload_address): Likewise.
20582 (visium_select_cc_mode): Likewise.
20583 (output_cbranch): Likewise.
20584 (visium_split_double_move): Likewise.
20585 (visium_expand_copysign): Likewise.
20586 (visium_expand_int_cstore): Likewise.
20587 (visium_expand_fp_cstore): Likewise.
20588 * config/visium/visium.c (visium_pass_by_reference): Likewise.
20589 (visium_function_arg): Likewise.
20590 (visium_function_arg_advance): Likewise.
20591 (visium_libcall_value): Likewise.
20592 (visium_setup_incoming_varargs): Likewise.
20593 (visium_legitimate_constant_p): Likewise.
20594 (visium_legitimate_address_p): Likewise.
20595 (visium_legitimize_address): Likewise.
20596 (visium_secondary_reload): Likewise.
20597 (visium_register_move_cost): Likewise.
20598 (visium_memory_move_cost): Likewise.
20599 (prepare_move_operands): Likewise.
20600 (ok_for_simple_move_operands): Likewise.
20601 (ok_for_simple_move_strict_operands): Likewise.
20602 (ok_for_simple_arith_logic_operands): Likewise.
20603 (visium_function_value_1): Likewise.
20604 (rtx_ok_for_offset_p): Likewise.
20605 (visium_legitimize_reload_address): Likewise.
20606 (visium_split_double_move): Likewise.
20607 (visium_expand_copysign): Likewise.
20608 (visium_expand_int_cstore): Likewise.
20609 (visium_expand_fp_cstore): Likewise.
20610 (visium_split_cstore): Likewise.
20611 (visium_select_cc_mode): Likewise.
20612 (visium_split_cbranch): Likewise.
20613 (output_cbranch): Likewise.
20614 (visium_print_operand_address): Likewise.
20615 * expmed.c (flip_storage_order): Likewise.
20616 * expmed.h (emit_cstore): Likewise.
20617 (flip_storage_order): Likewise.
20618 * genrecog.c (validate_pattern): Likewise.
20619 * hsa-gen.c (gen_hsa_addr): Likewise.
20620 * internal-fn.c (expand_arith_overflow): Likewise.
20621 * ira-color.c (allocno_copy_cost_saving): Likewise.
20622 * lra-assigns.c (find_hard_regno_for_1): Likewise.
20623 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
20624 (process_invariant_for_inheritance): Likewise.
20625 * lra-eliminations.c (move_plus_up): Likewise.
20626 * omp-low.c (lower_oacc_reductions): Likewise.
20627 * simplify-rtx.c (simplify_subreg): Likewise.
20628 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
20629 (TARGET_CHKP_BOUND_MODE): Likewise..
20630 * targhooks.c (default_chkp_bound_mode): Likewise.
20631 (default_setup_incoming_vararg_bounds): Likewise.
20632 * targhooks.h (default_chkp_bound_mode): Likewise.
20633 (default_setup_incoming_vararg_bounds): Likewise.
20634 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
20635 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
20636 (have_whole_vector_shift): Likewise.
20637 * tree-vect-stmts.c (vectorizable_load): Likewise.
20638 * doc/tm.texi: Regenerate.
20640 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
20642 Graceful degrade if Binutils PR21472 is not available.
20645 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
20646 .rodata in flash test fails.
20647 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
20648 * confgure: Regenerate.
20649 * config.in: Regenerate.
20650 * config/avr/avr.c (avr_asm_named_section)
20651 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
20652 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
20653 (avr_asm_init_sections): Same.
20655 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20657 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
20658 (fma<VH:mode>4_intrinsic): Likewise.
20659 (*fmsub<VCVTF:mode>4): Likewise.
20660 (*fmsub<VH:mode>4_intrinsic): Likewise.
20662 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
20665 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
20666 Don't depend on "optimize > 0".
20667 (out_movhi_r_mr, out_movqi_mr_r): Same.
20668 (out_movhi_mr_r, out_movqi_r_mr): Same.
20669 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
20670 io_address_operand on "optimize > 0".
20672 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20674 * tree-loop-distribution.c: Add general explanantion on the pass.
20675 (generate_loops_for_partition): Mark distributed loop.
20676 (pg_add_dependence_edges): New parameter. Handle alias data
20677 dependence specially and record it in the parameter if asked.
20678 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
20679 (init_partition_graph_vertices, add_partition_graph_edge): New.
20680 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
20681 (free_partition_graph_vdata, build_partition_graph): New.
20682 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
20683 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
20684 (data_ref_segment_size, latch_dominated_by_data_ref): New.
20685 (compute_alias_check_pairs, version_loop_by_alias_check): New.
20686 (version_for_distribution_p, finalize_partitions): New.
20687 (distribute_loop): Handle alias data dependence specially. Factor
20688 out loop fusion code as functions and call these functions.
20690 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20692 * tree-loop-distribution.c (classify_partition): New parameter and
20693 better handle reduction statement.
20694 (rdg_build_partitions): Revise comment.
20695 (distribute_loop): Compute statements in all partitions and pass it
20696 to classify_partition.
20698 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20700 * tree-loop-distribution.c (enum partition_type): New.
20701 (struct partition): New field type.
20702 (partition_merge_into): Add parameter. Update partition type.
20703 (data_dep_in_cycle_p, update_type_for_merge): New functions.
20704 (build_rdg_partition_for_vertex): Compute partition type.
20705 (rdg_build_partitions): Dump partition type.
20706 (distribute_loop): Update calls to partition_merge_into.
20708 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20710 * tree-loop-distribution.c (struct ddr_hasher): New.
20711 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
20713 (classify_partition): Call get_data_dependence.
20714 (pg_add_dependence_edges): Ditto.
20715 (distribute_loop): Release data dependence hash table.
20717 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20719 * tree-loop-distribution.c (ref_base_address): Delete.
20720 (similar_memory_accesses): Rename ...
20721 (share_memory_accesses): ... to this. Check if partitions access
20722 the same memory reference.
20723 (distribute_loop): Call share_memory_accesses.
20725 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20727 * tree-loop-distribution.c (struct partition): New field recording
20728 its data reference.
20729 (partition_alloc, partition_free): Init and release data refs.
20730 (partition_merge_into): Merge data refs.
20731 (build_rdg_partition_for_vertex): Collect data refs for partition.
20732 (pg_add_dependence_edges): Change parameters from vector to bitmap.
20734 (distribute_loop): Remve data refs from vertice data of partition
20737 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20739 * tree-loop-distribution.c (params.h): Include header file.
20740 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
20741 (datarefs_vec): New global var.
20742 (create_rdg_vertices): Use datarefs_vec directly.
20743 (free_rdg): Don't free data references.
20744 (build_rdg): Update use. Don't free data references.
20745 (distribute_loop): Compute global variable for data references.
20746 Bail out if there are too many data references.
20748 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20750 * tree-loop-distribution.c (loop_nest): New global var.
20751 (build_rdg): Use loop directly, rather than loop nest.
20752 (pg_add_dependence_edges): Remove loop nest parameter. Use global
20754 (distribute_loop): Compute global variable loop nest. Update use.
20756 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20758 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
20759 (partition_merge_into): New parameter. Dump reason for fusion.
20760 (distribute_loop): Update use of partition_merge_into.
20762 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20764 * tree-loop-distribution.c (bb_top_order_index): New.
20765 (bb_top_order_index_size, bb_top_order_cmp): New.
20766 (stmts_from_loop): Use topological order.
20767 (pass_loop_distribution::execute): Compute and release topological
20768 order for basic blocks.
20770 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20772 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
20775 2017-07-05 Bin Cheng <bin.cheng@arm.com>
20777 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
20778 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
20779 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
20780 * internal-fn.def (LOOP_DIST_ALIAS): New.
20781 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
20782 (fold_loop_internal_call): ... this.
20783 (vect_loop_dist_alias_call): New function.
20784 (set_uid_loop_bbs): Call fold_loop_internal_call.
20785 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
20788 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
20791 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
20792 Require dead FLAGS_REG at the beginning of a peephole.
20794 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
20797 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
20798 arguments in the call to __builtin_ia32_sbb_u32.
20799 (_subborrow_u64): Swap _X and _Y arguments in the call to
20800 __builtin_ia32_sbb_u64.
20802 2017-07-04 Jakub Jelinek <jakub@redhat.com>
20805 * tree-vrp.c (compare_assert_loc): Turn into a function template
20806 with stable template parameter. Only test if a->e is NULL,
20807 !a->e == !b->e has been verified already. Use e == NULL or
20808 e != NULL instead of e or ! e tests. If stable is true, don't use
20809 iterative_hash_expr, on the other side allow a or b or both NULL
20810 and sort the NULLs last.
20811 (process_assert_insertions): Sort using compare_assert_loc<false>
20812 instead of compare_assert_loc, later sort using
20813 compare_assert_loc<true> before calling process_assert_insertions_for
20814 in a loop. Use break instead of continue once seen NULL pointer.
20816 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20818 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
20819 Cortex-R7 and Cortex-R8 processors.
20821 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
20823 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
20824 uninitialized while src is not.
20826 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
20828 * common/config/arm/arm-common.c: Adjust include path for
20830 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
20831 (arm-cpu.h): Create in build directory. Adjust dependency rules.
20832 (arm-cpu-data.h): Likewise.
20833 (arm-cpu-cdata.h): Likewise.
20834 * config/arm/arm-cpu.h: Delete.
20835 * config/arm/arm-cpu-cdata.h: Delete.
20836 * config/arm/arm-cpu-data.h: Delete.
20838 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
20840 * config/arm/arm-cpus.in (cortex-a55): New.
20841 (cortex-a75): Likewise.
20842 (cortex-a75.cortex-a55): Likewise.
20843 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
20845 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
20846 * config/arm/arm-cpu-cdata.h: Regenerate.
20847 * config/arm/arm-cpu-data.h: Regenerate.
20848 * config/arm/arm-cpu.h: Regenerate.
20849 * config/arm/arm-tables.opt: Regenerate.
20850 * config/arm/arm-tune.md: Regenerate.
20852 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
20854 * haifa-sched.c (sched_create_recovery_edges): Update profile.
20856 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
20858 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
20861 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
20863 PR tree-optimization/81292
20864 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
20865 full_string_p, also call adjust_related_strinfos if the adjustment
20866 is simple, otherwise invalidate related strinfos.
20868 2017-07-04 Martin Liska <mliska@suse.cz>
20871 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
20872 newly created variable as DECL_IGNORED_P.
20874 2017-07-04 Martin Liska <mliska@suse.cz>
20877 * ipa-inline.c (inline_small_functions):
20878 Use xstrdup_for_dump.
20880 2017-07-04 Tom de Vries <tom@codesourcery.com>
20882 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
20884 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
20887 * config/darwin.c (darwin_function_switched_text_sections):
20890 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
20892 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
20894 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
20896 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
20898 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20900 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
20901 min_profitable_iters, and th as inclusive lower bounds.
20902 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
20903 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
20904 for min_profitable_iters and min_profitable_estimate.
20905 (vect_transform_loop): Treat th as an inclusive lower bound.
20906 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
20908 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
20911 * config/darwin.c (darwin_function_switched_text_sections):
20912 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
20913 in two pieces, and suppress the use of buf.
20915 2017-07-03 Nathan Sidwell <nathan@acm.org>
20917 * hash-table.h (hash_table_mod1): Fix indentation.
20919 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
20921 PR middle-end/81290
20922 * predict.c (force_edge_cold): Be more careful about propagation
20924 * profile-count.h (profile_probability::guessed,
20925 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
20927 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
20929 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
20931 * doc/invoke.texi (rcpc architecture extension): Document it.
20933 2017-07-03 Richard Biener <rguenther@suse.de>
20935 PR tree-optimization/60510
20936 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
20937 the scalar reduction PHI and use it.
20938 (vectorizable_reduction): Properly guard the single_defuse_cycle
20939 path for non-SLP reduction chains where we cannot use it.
20940 Rework reduc_def/index and vector type deduction. Rework
20941 vector operand gathering during reduction op code-gen.
20942 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
20943 chains dissolve the chain and leave it to non-SLP reduction
20946 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20948 * tree-data-ref.h (dr_alignment): Declare.
20949 * tree-data-ref.c (dr_alignment): New function.
20950 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
20951 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
20953 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
20955 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20957 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
20958 and base_misalignment fields.
20959 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
20960 * tree-data-ref.c: Include builtins.h.
20961 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
20962 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
20963 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
20964 * tree-vect-data-refs.c: Include tree-cfg.h.
20965 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
20966 fields instead of calculating an alignment here.
20967 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
20968 innermost_loop_behavior fields.
20970 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20972 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
20974 (DR_STEP_ALIGNMENT): New macro.
20975 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
20976 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
20977 (create_data_ref): Print it.
20978 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
20979 to tell whether the step preserves vector (mis)alignment.
20980 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
20981 Move the check for an integer step and generalise to all INTEGER_CST.
20982 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
20983 Print the outer step alignment.
20985 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
20987 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
20988 with offset_alignment.
20989 (DR_ALIGNED_TO): Delete.
20990 (DR_OFFSET_ALIGNMENT): New macro.
20991 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
20992 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
20993 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
20994 (create_data_ref): Likewise.
20995 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
20996 (vect_analyze_data_refs): Likewise.
20997 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
20998 creating dummy innermost behavior.
21000 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21002 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
21003 with a "innermost_loop_behavior *" and refeence tree.
21004 * tree-data-ref.c (dr_analyze_innermost): Likewise.
21005 (create_data_ref): Update call accordingly.
21006 * tree-predcom.c (find_looparound_phi): Likewise.
21008 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21010 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
21011 fields with dr_wrt_vec_loop.
21012 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
21013 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
21014 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
21015 (vect_dr_behavior): New function.
21016 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
21017 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
21018 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
21019 track whether the step preserves the misalignment.
21020 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
21021 Use vect_dr_behavior.
21022 (vect_setup_realignment): Update call accordingly.
21023 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
21024 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
21025 call to vect_create_addr_base_for_vector_ref.
21026 (vect_create_cond_for_align_checks): Likewise.
21027 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
21028 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
21029 (vect_recog_mask_conversion_pattern): Likewise.
21030 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
21031 (new_stmt_vec_info): Remove redundant zeroing.
21033 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
21035 * common/config/arm/arm-common.c (arm_be8_option): New function.
21036 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
21037 (ISA_ARMv6): Add isa_bit_be8.
21038 * config/arm/arm.h (arm_be8_option): Add prototype.
21039 (BE8_SPEC_FUNCTION): New define.
21040 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
21041 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
21042 (mlittle-endian): Similarly.
21043 (mbe8, mbe32): New options.
21044 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
21045 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
21047 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21049 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
21051 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21053 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
21054 (cleanup_tree_cfg_bb): Use it.
21055 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
21057 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
21059 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21062 * loop-doloop.c (add_test): Update profile.
21064 2017-07-03 Martin Liska <mliska@suse.cz>
21067 * sanopt.c (rewrite_usage_of_param): New function.
21068 (sanitize_rewrite_addressable_params): Likewise.
21069 (pass_sanopt::execute): Call rewrite_usage_of_param.
21071 2017-07-03 Richard Biener <rguenther@suse.de>
21073 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
21074 back to using VIEW_CONVERT_EXPR.
21076 2017-07-03 Martin Liska <mliska@suse.cz>
21079 * doc/extend.texi: Document when a resolver function is
21080 generated for target_clones.
21082 2017-07-03 Martin Liska <mliska@suse.cz>
21084 * asan.c (asan_emit_stack_protection): Unpoison just red zones
21085 and shadow memory of auto variables which are subject of
21086 use-after-scope sanitization.
21087 (asan_expand_mark_ifn): Add do set only when is_poison.
21089 2016-07-03 Richard Biener <rguenther@suse.de>
21091 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
21093 (vect_force_simple_reduction): Record reduction def -> phi mapping.
21094 (vectorizable_reduction): Perform reduction PHI creation when
21095 visiting a reduction PHI and adjust and simplify code generation
21096 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
21097 (vect_transform_loop): Visit reduction PHIs.
21098 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
21099 defs into the SLP tree.
21100 (vect_build_slp_tree): Reduction defs terminate the recursion.
21101 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
21103 (vect_get_vec_defs_for_stmt_copy): Export.
21104 (vect_get_vec_defs): Likewise.
21105 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
21107 (vect_get_vec_defs_for_stmt_copy): Declare.
21108 (vect_get_vec_defs): Likewise.
21110 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21112 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
21113 parameter with a "loop" parameter and use it instead of the
21114 loop containing DR_STMT. Don't check simple_iv when doing
21115 BB analysis. Describe the two analysis modes in the comment.
21117 2017-07-03 Tom de Vries <tom@codesourcery.com>
21119 PR tree-optimization/69468
21120 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
21121 (find_same_succ_bb): Handle ignore_edge_flags.
21123 2017-07-03 Tom de Vries <tom@codesourcery.com>
21125 PR tree-optimization/81192
21126 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
21128 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
21130 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
21132 2017-07-03 Tom de Vries <tom@codesourcery.com>
21134 PR tree-optimization/81192
21135 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
21136 BB_SAME_SUCC (bb) == NULL.
21138 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21140 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
21143 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21145 * dumpfile.c: Include profile-count.h
21146 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
21148 (insert_cond_bb): Update profile.
21149 * tree-cfg.h (insert_cond_bb): Update prototype.
21150 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
21151 * tree-dump.c: Do not include tree-cfg.
21153 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21155 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
21157 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21159 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
21162 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21164 * tree-complex.c (expand_complex_div_wide): update profile.
21166 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21167 Alan Hayward <alan.hayward@arm.com>
21168 David Sherwood <david.sherwood@arm.com>
21170 * Makefile.in (MACHMODE_H): Remove insn-modes.h
21171 (CORETYPES_H): New define.
21172 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
21173 (insn-modes-inline.h, s-modes-inline-h): New rules.
21174 (generated_files): Add insn-modes-inline.h.
21175 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
21176 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
21177 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
21178 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
21179 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
21180 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
21181 (build/gencodes.o, build/genconditions.o): Likewise.
21182 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
21183 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
21184 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
21185 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
21186 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
21187 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
21188 * coretypes.h: Include everything up to real.h for generators.
21189 Include insn-modes.h first. Include wide-int-print.h after
21190 wide-int.h. Include insn-modes-inline.h and then machmode.h.
21191 * machmode.h: Don't include insn-modes.h here.
21192 * function-tests.c: Remove includes of signop.h, machmode.h,
21193 double-int.h and wide-int.h.
21195 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
21197 * optc-save-gen.awk: Likewise.
21198 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
21199 * godump.c: Remove include of wide-int-print.h.
21200 * pretty-print.h: Likewise.
21201 * wide-int-print.cc: Likewise.
21202 * wide-int.cc: Likewise.
21203 * hash-map-tests.c: Remove include of signop.h.
21204 * hash-set-tests.c: Likewise.
21205 * rtl-tests.c: Likewise.
21206 * mkconfig.sh: Remove include of machmode.h.
21207 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
21209 (emit_insn_modes_inline_h): ...this new function. Emit the code
21210 into an insn-modes-inline.h header file, adding appropriate
21211 include guards and end comments.
21212 (emit_insn_modes_c_header): Remove include of machmode.h.
21213 (emit_min_insn_modes_c_header): Include coretypes.h rather than
21215 (main): Handle -i flag and call emit_insn_modes_inline_h when
21218 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21220 * tree-ssa-strlen.c (strinfo): Rename the length field to
21221 nonzero_chars. Add a full_string_p field.
21222 (compare_nonzero_chars, zero_length_string_p): New functions.
21223 (get_addr_stridx): Add an offset_out parameter.
21224 Use compare_nonzero_chars.
21225 (get_stridx): Update accordingly. Use compare_nonzero_chars.
21226 (new_strinfo): Update after above changes to strinfo.
21227 (set_endptr_and_length): Set full_string_p.
21228 (get_string_length): Update after above changes to strinfo.
21229 (unshare_strinfo): Update call to new_strinfo.
21230 (maybe_invalidate): Likewise.
21231 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
21232 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
21233 as a uhwi instead of an shwi. Update after above changes to
21234 strinfo and new_strinfo.
21235 (zero_length_string): Assert that chainsi contains full strings.
21236 Use zero_length_string_p. Update call to new_strinfo.
21237 (adjust_related_strinfos): Update after above changes to strinfo.
21238 Copy full_string_p from origsi.
21239 (adjust_last_stmt): Use zero_length_string_p.
21240 (handle_builtin_strlen): Update after above changes to strinfo and
21241 new_strinfo. Install the lhs as the string length if the previous
21242 entry didn't describe a full string.
21243 (handle_builtin_strchr): Update after above changes to strinfo
21245 (handle_builtin_strcpy): Likewise.
21246 (handle_builtin_strcat): Likewise.
21247 (handle_builtin_malloc): Likewise.
21248 (handle_pointer_plus): Likewise.
21249 (handle_builtin_memcpy): Likewise. Track nonzero characters
21250 that aren't necessarily followed by a nul terminator.
21251 (handle_char_store): Likewise.
21253 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21255 PR tree-optimization/80769
21256 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
21257 for malloc and calloc. Document the new invariant that all related
21258 strinfos have delayed lengths or none do.
21259 (verify_related_strinfos): Move earlier in file.
21260 (set_endptr_and_length): New function, split out from...
21261 (get_string_length): ...here. Also set the lengths of related
21263 (zero_length_string): Assert that chainsi has known (rather than
21265 (adjust_related_strinfos): Likewise.
21267 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21269 PR tree-optimization/81136
21270 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
21271 assert that two references with the same misalignment have the same
21272 compile-time misalignment if those compile-time misalignments
21275 2017-07-01 Andi Kleen <ak@linux.intel.com>
21277 * print-tree.c (print_node): Print all attributes.
21279 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21281 * cfg.c (scale_bbs_frequencies): New function.
21282 * cfg.h (scale_bbs_frequencies): Declare it.
21283 * cfgloopanal.c (single_likely_exit): Cleanup.
21284 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
21286 (scale_loop_profile): Likewise.
21287 (loop_version): Likewise.
21288 (create_empty_loop_on_edge): Update.
21289 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
21290 scale_loop_frequencies, scale_loop_profile, loopify,
21291 loop_version): Update prototypes.
21292 * modulo-sched.c (sms_schedule): Update.
21293 * predict.c (unlikely_executed_edge_p): Also check probability.
21294 (probably_never_executed_edge_p): Fix typo.
21295 * tree-if-conv.c (version_loop_for_if_conversion): Update.
21296 * tree-parloops.c (gen_parallel_loop): Update.
21297 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
21298 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21299 * tree-ssa-loop-split.c (split_loop): Update.
21300 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
21301 * tree-vect-loop-manip.c (vect_do_peeling): Update.
21302 (vect_loop_versioning): Update.
21303 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21305 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21307 * trans-mem.c (split_bb_make_tm_edge): Update profile.
21309 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21311 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
21312 to keep profile consistent.
21314 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21316 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
21317 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
21318 * profile-count.h (max_safe_multiplier): Make unsigned.
21319 (profile_count::guessed_zero): New.
21321 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21323 * bb-reorder.c (fix_up_crossing_landing_pad,
21324 fix_crossing_conditional_branches): Use make_single_succ_edge
21325 to keep profile consistent.
21327 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21329 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
21332 2017-07-01 Jakub Jelinek <jakub@redhat.com>
21335 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
21336 the right scopes, make sure cond_jump isn't preserved between multiple
21337 iterations. Search for fallthru edge whenever there are 3+ edges and
21338 use find_fallthru_edge for it.
21340 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21342 Patch by Alexander Monakov <amonakov@ispras.ru>
21343 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
21344 probabilities consistently.
21346 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21348 * pa.c (pa_expand_compare_and_swap_loop): Update call of
21349 emit_cmp_and_jump_insns.
21351 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21354 * tree-inline.c (expand_call_inline): Combine profile statuses.
21356 2017-06-30 Andrew Pinski <apinski@cavium.com>
21358 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
21359 fold_stmt returned true.
21361 2017-06-30 Nathan Sidwell <nathan@acm.org>
21363 * ggc.h (empty_string): Delete.
21364 * cfgexpand.c (expand_asm_stmt): Use plain "".
21365 * optabs.c (expand_asm_memory_barrier): Likewise.
21366 * stringpool.c (empty_string): Delete.
21367 (digit_vector, digit_string): Delete.
21368 (ggc_alloc_string): Use plain "", don't optimize single digit
21369 strings. Use ggc_alloc_atomic.
21371 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
21373 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
21374 comparison set and one other set, use the cost of the non-comparison
21377 2017-06-30 Nathan Sidwell <nathan@acm.org>
21379 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
21382 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
21384 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
21385 loops. Remove now unneeded calls to gimple_switch_set_label() that
21386 just set removed labels to NULL_TREE.
21388 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
21390 * tree-ssanames.c (set_range_info_raw): Abstract from ...
21391 (set_range_info): ...here. Only call set_range_info_raw if domain
21393 (set_nonzero_bits): Call set_range_info_raw.
21394 * tree-ssanames.h (set_range_info_raw): New.
21396 2017-06-30 Jakub Jelinek <jakub@redhat.com>
21399 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
21400 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
21401 of nonimmediate_operand and <store_mask_constraint> instead of m
21402 for the input operand. For V8FI iterator, always split if input
21403 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
21404 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
21405 <store_mask_predicate> instead of register_operand and
21406 <store_mask_constraint> instead of v for the input operand. Make
21407 sure both operands aren't MEMs for if not <mask_applied>.
21409 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
21411 * lto-wrapper.c (copy_file) Close both file descriptors before
21414 2017-06-30 Martin Liska <mliska@suse.cz>
21417 * multiple_target.c (create_dispatcher_calls): Make ifunc
21418 also for function that don't have calls or are not referenced.
21420 2017-06-30 Richard Biener <rguenther@suse.de>
21422 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
21423 analyze the first scalar stmt. Move vector type computation
21424 for the BB case here from ...
21425 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
21426 live operation processing in the SLP case properly.
21428 2017-06-30 Richard Biener <rguenther@suse.de>
21430 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
21432 2017-06-30 Martin Liska <mliska@suse.cz>
21435 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
21436 before BUILT_IN_UNWIND_RESUME when ASAN is used.
21438 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
21440 * doc/invoke.texi (AArch64): Add missing options and remove redundant
21443 2017-06-30 Richard Biener <rguenther@suse.de>
21445 PR tree-optimization/81249
21446 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
21447 condition reduction result to original scalar type.
21449 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21451 * profile-count.h (enum profile_quality): Fix typos and whitespace
21454 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21456 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
21457 type for branch probabilities.
21459 2017-06-29 Julian Brown <julian@codesourcery.com>
21460 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21462 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
21463 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
21464 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
21465 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
21467 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21469 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
21470 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
21471 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
21472 CC usage from generic code to here.
21473 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
21474 CC usage into the target macros.
21476 2017-06-29 Maya Rashish <coypu@sdf.org>
21478 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
21481 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21483 * arm/arm-builtins.c: Include profile-count.h
21484 * except.c (sjlj_emit_function_enter): Use
21485 profile_probability::unlikely.
21487 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21489 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
21490 and tocrel_offset be pointer args rather than implicitly using
21492 (legitimate_constant_pool_address_p, rs6000_emit_move,
21493 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
21494 tocrel_offset and use in toc_relative_expr_p call.
21495 (print_operand, print_operand_address): Use static tocrel_base_oac
21496 and tocrel_offset_oac.
21497 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
21500 2017-06-29 Maya Rashish <coypu@sdf.org>
21502 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
21504 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
21506 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
21507 objects, take into account only the alignment of 'op0' and 'mode1' if
21510 2017-06-29 Steve Ellcey <sellcey@cavium.com>
21512 * ccmp.c (ccmp_tree_comparison_p): New function.
21513 (ccmp_candidate_p): Update to use above function.
21514 (get_compare_parts): New function.
21515 (expand_ccmp_next): Update to use new functions.
21516 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
21518 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
21519 take mode as argument.
21520 * ccmp.h (expand_ccmp_expr): Add mode as argument.
21521 * expr.c (expand_expr_real_1): Pass mode as argument.
21523 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
21525 * combine.c (combine_instructions): Print insns to dump_file, together
21528 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21530 * asan.c (asan_emit_stack_protection): Update.
21531 (create_cond_insert_point): Update.
21532 * auto-profile.c (afdo_propagate_circuit): Update.
21533 * basic-block.h (struct edge_def): Turn probability to
21534 profile_probability.
21535 (EDGE_FREQUENCY): Update.
21536 * bb-reorder.c (find_traces_1_round): Update.
21537 (better_edge_p): Update.
21538 (sanitize_hot_paths): Update.
21539 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
21540 (make_single_succ_edge): Update.
21541 (check_bb_profile): Update.
21542 (dump_edge_info): Update.
21543 (update_bb_profile_for_threading): Update.
21544 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
21545 probabilitycount to 0.
21546 * cfgbuild.c (compute_outgoing_frequencies): Update.
21547 * cfgcleanup.c (try_forward_edges): Update.
21548 (outgoing_edges_match): Update.
21549 (try_crossjump_to_edge): Update.
21550 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
21551 (expand_gimple_tailcall): Update.
21552 (construct_init_block): Use make_single_succ_edge.
21553 (construct_exit_block): Use make_single_succ_edge.
21554 * cfghooks.c (verify_flow_info): Update.
21555 (redirect_edge_succ_nodup): Update.
21556 (split_edge): Update.
21557 (account_profile_record): Update.
21558 * cfgloopanal.c (single_likely_exit): Update.
21559 * cfgloopmanip.c (scale_loop_profile): Update.
21560 (set_zero_probability): Remove.
21561 (duplicate_loop_to_header_edge): Update.
21562 * cfgloopmanip.h (loop_version): Update prototype.
21563 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
21564 (force_nonfallthru_and_redirect): Update.
21565 (update_br_prob_note): Update.
21566 (rtl_verify_edges): Update.
21567 (purge_dead_edges): Update.
21568 (rtl_lv_add_condition_to_bb): Update.
21569 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
21570 * cgraphunit.c (init_lowered_empty_function): Update.
21571 (cgraph_node::expand_thunk): Update.
21572 * cilk-common.c: Include profile-count.h
21573 * dojump.c (inv): Remove.
21574 (jumpifnot): Update.
21575 (jumpifnot_1): Update.
21576 (do_jump_1): Update.
21578 (do_jump_by_parts_greater_rtx): Update.
21579 (do_compare_rtx_and_jump): Update.
21580 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
21581 do_jump_1. do_compare_rtx_and_jump): Update prototype.
21582 * dwarf2cfi.c: Include profile-count.h
21583 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
21584 (sjlj_emit_dispatch_table): Likewise.
21585 * explow.c: Include profile-count.h
21586 * expmed.c (emit_store_flag_force): Update.
21587 (do_cmp_and_jump): Update.
21588 * expr.c (compare_by_pieces_d::generate): Update.
21589 (compare_by_pieces_d::finish_mode): Update.
21590 (emit_block_move_via_loop): Update.
21591 (store_expr_with_bounds): Update.
21592 (store_constructor): Update.
21593 (expand_expr_real_2): Update.
21594 (expand_expr_real_1): Update.
21595 * expr.h (try_casesi, try_tablejump): Update prototypes.
21596 * gimple-pretty-print.c (dump_probability): Update.
21597 (dump_profile): New.
21598 (dump_gimple_label): Update.
21599 (dump_gimple_bb_header): Update.
21600 * graph.c (draw_cfg_node_succ_edges): Update.
21601 * hsa-gen.c (convert_switch_statements): Update.
21602 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
21603 (find_if_case_1): Update.
21604 (find_if_case_2): Update.
21605 * internal-fn.c (expand_arith_overflow_result_store): Update.
21606 (expand_addsub_overflow): Update.
21607 (expand_neg_overflow): Update.
21608 (expand_mul_overflow): Update.
21609 (expand_vector_ubsan_overflow): Update.
21610 * ipa-cp.c (good_cloning_opportunity_p): Update.
21611 * ipa-split.c (split_function): Use make_single_succ_edge.
21612 * ipa-utils.c (ipa_merge_profiles): Update.
21613 * loop-doloop.c (add_test): Update.
21614 (doloop_modify): Update.
21615 * loop-unroll.c (compare_and_jump_seq): Update.
21616 (unroll_loop_runtime_iterations): Update.
21617 * lra-constraints.c (lra_inheritance): Update.
21618 * lto-streamer-in.c (input_cfg): Update.
21619 * lto-streamer-out.c (output_cfg): Update.
21620 * mcf.c (adjust_cfg_counts): Update.
21621 * modulo-sched.c (sms_schedule): Update.
21622 * omp-expand.c (expand_omp_for_init_counts): Update.
21623 (extract_omp_for_update_vars): Update.
21624 (expand_omp_ordered_sink): Update.
21625 (expand_omp_for_ordered_loops): Update.
21626 (expand_omp_for_generic): Update.
21627 (expand_omp_for_static_nochunk): Update.
21628 (expand_omp_for_static_chunk): Update.
21629 (expand_cilk_for): Update.
21630 (expand_omp_simd): Update.
21631 (expand_omp_taskloop_for_outer): Update.
21632 (expand_omp_taskloop_for_inner): Update.
21633 * omp-simd-clone.c (simd_clone_adjust): Update.
21634 * optabs.c (expand_doubleword_shift): Update.
21635 (expand_abs): Update.
21636 (emit_cmp_and_jump_insn_1): Update.
21637 (expand_compare_and_swap_loop): Update.
21638 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
21639 * predict.c (predictable_edge_p): Update.
21640 (edge_probability_reliable_p): Update.
21641 (set_even_probabilities): Update.
21642 (combine_predictions_for_insn): Update.
21643 (combine_predictions_for_bb): Update.
21644 (propagate_freq): Update.
21645 (estimate_bb_frequencies): Update.
21646 (force_edge_cold): Update.
21647 * profile-count.c (profile_count::dump): Add missing space into dump.
21648 (profile_count::debug): Add newline.
21649 (profile_count::differs_from_p): Explicitly convert to unsigned.
21650 (profile_count::stream_in): Update.
21651 (profile_probability::dump): New member function.
21652 (profile_probability::debug): New member function.
21653 (profile_probability::differs_from_p): New member function.
21654 (profile_probability::differs_lot_from_p): New member function.
21655 (profile_probability::stream_in): New member function.
21656 (profile_probability::stream_out): New member function.
21657 * profile-count.h (profile_count_quality): Rename to ...
21658 (profile_quality): ... this one.
21659 (profile_probability): New.
21660 (profile_count): Update.
21661 * profile.c (compute_branch_probabilities): Update.
21662 * recog.c (peep2_attempt): Update.
21663 * sched-ebb.c (schedule_ebbs): Update.
21664 * sched-rgn.c (find_single_block_region): Update.
21665 (compute_dom_prob_ps): Update.
21666 (schedule_region): Update.
21667 * sel-sched-ir.c (compute_succs_info): Update.
21668 * stmt.c (struct case_node): Update.
21669 (do_jump_if_equal): Update.
21670 (get_outgoing_edge_probs): Update.
21671 (conditional_probability): Update.
21672 (emit_case_dispatch_table): Update.
21673 (expand_case): Update.
21674 (expand_sjlj_dispatch_table): Update.
21675 (emit_case_nodes): Update.
21676 * targhooks.c: Update.
21677 * tracer.c (better_p): Update.
21678 (find_best_successor): Update.
21679 * trans-mem.c (expand_transaction): Update.
21680 * tree-call-cdce.c: Update.
21681 * tree-cfg.c (gimple_split_edge): Upate.
21682 (move_sese_region_to_fn): Upate.
21683 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
21684 * tree-eh.c (lower_resx): Upate.
21685 (cleanup_empty_eh_move_lp): Upate.
21686 * tree-if-conv.c (version_loop_for_if_conversion): Update.
21687 * tree-inline.c (copy_edges_for_bb): Update.
21688 (copy_cfg_body): Update.
21689 * tree-parloops.c (gen_parallel_loop): Update.
21690 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
21691 (gimple_gen_time_profiler): Update.
21692 * tree-ssa-dce.c (remove_dead_stmt): Update.
21693 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21694 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
21695 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
21696 (unloop_loops): Update.
21697 (try_peel_loop): Update.
21698 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21699 * tree-ssa-loop-split.c (connect_loops): Update.
21700 (split_loop): Update.
21701 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
21702 (hoist_guard): Update.
21703 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
21704 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
21705 (value_replacement): Update.
21706 * tree-ssa-reassoc.c (branch_fixup): Update.
21707 * tree-ssa-tail-merge.c (replace_block_by): Update.
21708 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
21709 (create_edge_and_update_destination_phis): Update.
21710 (compute_path_counts): Update.
21711 (recompute_probabilities): Update.
21712 (update_joiner_offpath_counts): Update.
21713 (freqs_to_counts_path): Update.
21714 (duplicate_thread_path): Update.
21715 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
21716 (struct switch_conv_info): Update.
21717 (gen_inbound_check): Update.
21718 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21719 (vect_do_peeling): Update.
21720 (vect_loop_versioning): Update.
21721 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21722 (optimize_mask_stores): Update.
21723 * ubsan.c (ubsan_expand_null_ifn): Update.
21724 * value-prof.c (gimple_divmod_fixed_value): Update.
21725 (gimple_divmod_fixed_value_transform): Update.
21726 (gimple_mod_pow2): Update.
21727 (gimple_mod_pow2_value_transform): Update.
21728 (gimple_mod_subtract): Update.
21729 (gimple_mod_subtract_transform): Update.
21730 (gimple_ic): Update.
21731 (gimple_stringop_fixed_value): Update.
21732 (gimple_stringops_transform): Update.
21733 * value-prof.h: Update.
21735 2017-06-29 Carl Love <cel@us.ibm.com>
21737 * config/rs6000/rs6000-c.c: Add support for built-in functions
21738 vector signed int vec_signed (vector float);
21739 vector signed long long vec_signed (vector double);
21740 vector signed int vec_signed2 (vector double, vector double);
21741 vector signed int vec_signede (vector double);
21742 vector signed int vec_signedo (vector double);
21743 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
21744 instruction generator.
21745 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
21746 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
21747 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
21749 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
21750 vunsignede_v2df): Add define_expands.
21751 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
21752 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
21753 VEC_UNSIGNEDO): Add definitions.
21754 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
21755 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
21756 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
21757 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
21758 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
21759 * config/rs6000/altivec.h (vec_signed, vec_signed2,
21760 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
21761 vec_unsignede, vec_unsignedo): Add builtin defines.
21762 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
21764 * doc/extend.texi: Update the built-in documentation file for the
21765 new built-in functions.
21767 2017-06-29 Richard Biener <rguenther@suse.de>
21769 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
21770 reduction chains to LOOP_VINFO_REDUCTIONS.
21771 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
21772 SLP reductions after processing reduction chains.
21774 2017-06-29 Nathan Sidwell <nathan@acm.org>
21776 * builtins.c (fold_builtin_FUNCTION): Use
21777 lang_hooks.decl_printable_name.
21779 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
21781 PR middle-end/81194
21782 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
21783 with only one label.
21784 * stmt.c (expand_case): Assert NCASES is greater than one.
21786 2017-06-29 Richard Biener <rguenther@suse.de>
21788 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
21790 (group_case_labels): Likewise.
21791 (find_taken_edge): Push sanity checking on val to workers...
21792 (find_taken_edge_cond_expr): ... here
21793 (find_taken_edge_switch_expr): ... and here, handle cases
21794 with just a default label.
21795 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
21796 (group_case_labels): Likewise.
21797 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
21798 group_case_labels does anything cleanup the CFG again.
21800 2017-06-29 Bin Cheng <bin.cheng@arm.com>
21802 PR tree-optimization/81196
21803 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
21804 exit condition comparing two IVs.
21806 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
21808 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
21809 profile to the dummy entry at the end of the list of architectures.
21810 * config/arm/arm-cpu-cdata.h: Regenerated.
21812 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21813 Michael Collison <michael.collison@arm.com>
21816 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
21818 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
21819 (*aarch64_reg_<mode>3_minus_mask): New pattern.
21820 (*aarch64_<optab>_reg_di3_mask2): New pattern.
21821 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
21822 of shift when the shift amount is masked with constant equal to
21823 the size of the mode.
21824 * config/aarch64/predicates.md (subreg_lowpart_operator): New
21827 2017-06-29 Martin Liska <mliska@suse.cz>
21829 * config/i386/i386.opt: Change range from [1,5] to [0,5].
21831 2017-06-29 Yury Gribov <tetra2005@gmail.com>
21834 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
21837 (edge_growth_cache_entry::edge_growth_cache_entry): New
21839 (reset_edge_growth_cache): Update to use constructor.
21841 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21843 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
21844 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
21845 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
21847 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
21849 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
21850 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
21852 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
21854 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
21855 (*-linux-uclibc*): Add t-uclibc tmake_file.
21856 * config/t-musl: New.
21857 * config/t-uclibc: New.
21859 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
21861 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
21863 (gen_comm_data): Emit architectural setting of arch_prof.
21864 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
21866 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
21867 (armv8-m.base, armv8-m.main): Likewise.
21868 * arm-protos.h (arm_build_target): Add profile field.
21869 (arch_option): Likewise.
21870 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
21872 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
21873 arm_active_target.profile.
21875 2017-06-28 Richard Biener <rguenther@suse.de>
21877 PR middle-end/81227
21878 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
21879 TYPE_OVERFLOW_WRAPS.
21880 * match.pd (negate_expr_p): Likewise.
21881 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
21882 fold_build2, not fold_binary.
21884 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
21886 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
21887 Convert memory address to Pmode.
21888 (aarch64_print_operand): Assert MEM operands are always Pmode.
21890 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
21893 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
21894 Remove redundant if.
21895 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
21896 * config/arm/aarch-common-protos.h
21897 (aarch_forward_to_shift_is_not_shifted_re): Remove.
21898 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
21900 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
21903 * multiple_target.c (create_dispatcher_calls): Set the default
21904 clone to be static, not public.
21906 2017-06-28 Richard Biener <rguenther@suse.de>
21908 * tree-vect-loop.c (vectorizable_reduction): Move special
21909 cond reduction IV var creation ...
21910 (vect_create_epilog_for_reduction): ... here. Remove induction_index
21911 parameter. Use STMT_VINFO_VECTYPE.
21912 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
21915 2017-06-28 Martin Liska <mliska@suse.cz>
21918 * ipa-visibility.c (non_local_p): Handle visibility.
21920 2017-06-28 Martin Liska <mliska@suse.cz>
21923 * common.opt: Add IntegerRange to various options.
21924 * opt-functions.awk (integer_range_info): New function.
21925 * optc-gen.awk: Add integer_range_info to cl_options struct.
21926 * opts-common.c (decode_cmdline_option): Handle
21927 CL_ERR_INT_RANGE_ARG.
21928 (cmdline_handle_error): Likewise.
21929 * opts.c (print_filtered_help): Show valid interval in
21930 when --help is provided.
21931 * opts.h (struct cl_option): Add range_min and range_max fields.
21932 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
21934 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
21936 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
21937 (x * C EQ/NE y * C): New transformation.
21939 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
21941 * genmultilib (combination_space): Accept '+' in option names.
21943 2017-06-28 Martin Liska <mliska@suse.cz>
21946 * asan.c (instrument_derefs): Bail out inner references
21947 that are hard register variables.
21949 2017-06-28 Jakub Jelinek <jakub@redhat.com>
21952 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
21953 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
21955 2017-06-28 Richard Biener <rguenther@suse.de>
21957 * tree-vectorizer.h (vect_get_vec_defs): Remove.
21958 (vect_get_slp_defs): Adjust.
21959 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
21961 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
21963 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
21964 get_initial_defs_for_reduction instead of vect_get_vec_defs.
21965 (vectorizable_reduction): Adjust.
21966 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
21968 (vect_get_slp_defs): Likewise.
21969 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
21970 (vectorizable_bswap): Adjust.
21971 (vectorizable_call): Likewise.
21972 (vectorizable_conversion): Likewise.
21973 (vectorizable_assignment): Likewise.
21974 (vectorizable_shift): Likewise.
21975 (vectorizable_operation): Likewise.
21976 (vectorizable_store): Likewise.
21977 (vectorizable_condition): Likewise.
21978 (vectorizable_comparison): Likewise.
21980 2017-06-28 Michael Collison <michael.collison@arm.com>
21983 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
21985 (arm_gen_movmemqi): Removed unused variable 'i'.
21986 Convert 'for' loop into 'while' loop.
21987 (arm_expand_prologue): Remove last unnecessary set of insn.
21988 (thumb_pop): Remove unused variable 'pushed_words'.
21989 (thumb_exit): Remove last unnecessary set of regs_to_pop.
21991 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21993 * config/s390/predicates.md: Use s390_rel_address_ok_p.
21994 * config/s390/s390-protos.h: Add prototype of
21995 s390_rel_address_ok_p.
21996 * config/s390/s390.c (s390_got_symbol): New function.
21997 (s390_rel_address_ok_p): New function.
21998 (legitimize_pic_address): Use s390_rel_address_ok_p.
21999 (s390_load_got): Use s390_got_symbol.
22000 (s390_option_override): Issue error if
22001 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
22002 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
22004 * config/s390/s390.opt: New option mpic-data-is-text-relative.
22006 2017-06-27 Andrew Pinski <apinski@cavium.com>
22008 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
22009 (X * copysign (1.0, X)): New pattern.
22010 (X * copysign (1.0, -X)): New pattern.
22011 (copysign (-1.0, CST)): New pattern.
22013 2017-06-27 Joseph Myers <joseph@codesourcery.com>
22015 * genmultilib (combination_space): Remove variable.
22016 Validate reuse rules against regular expression for any sequence
22017 of multilib options in any order.
22019 2017-06-27 Michael Collison <michael.collison@arm.com>
22021 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
22022 call aarch64_split_simd_combine.
22023 * (aarch64_combine_internal<mode>): Delete pattern.
22024 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
22025 Allow register and subreg operands.
22027 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22029 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
22030 specific need, just fallback on defaults.
22031 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
22033 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22034 Olivier Hainque <hainque@adacore.com>
22036 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
22038 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
22039 targets. Pick a default if no particular attempt applied.
22040 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
22043 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22045 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
22046 (x86_64-wrs-vxworks7): Likewise.
22048 2017-06-27 Marek Polacek <polacek@redhat.com>
22051 * ubsan.c (instrument_null): Check get_base_address's result for null.
22053 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
22055 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
22057 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
22059 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
22060 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
22061 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
22062 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
22063 New function types.
22064 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
22065 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
22066 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
22067 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
22068 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
22069 BUILT_IN_FEUPDATEENV): New builtins.
22070 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
22071 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
22072 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
22073 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
22075 (builtin_structptr_types): Adjust size.
22076 * tree.c (builtin_structptr_types): Add four entries.
22078 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22079 Olivier Hainque <hainque@adacore.com>
22081 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
22082 (TLS_SYM): New local macro, forcing reference to __tls__ on
22083 link command lines for VxWorks 7 RTPs, triggering initialization
22085 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
22086 OS features TLS support, true for RTPs on VxWorks 7.
22087 * config/vxworks.c (vxworks_override_options): Setup emutls
22090 2017-06-27 Jakub Jelinek <jakub@redhat.com>
22092 * predict.c (test_prediction_value_range): Use -1U instead of -1
22093 to avoid narrowing conversion warning.
22094 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
22095 to avoid narrowing conversion warning.
22096 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
22098 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
22100 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22102 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
22103 64bit configurations.
22104 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
22105 (SIZE_TYPE): Likewise.
22106 * config/vxworks.c (vxworks_emutls_var_fields): Use
22107 long_unsigned_type_node instead of unsigned_type_node as the offset
22108 field type, which is "pointer" mode in emutls.c.
22110 2017-06-27 Jakub Jelinek <jakub@redhat.com>
22113 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
22115 PR middle-end/81207
22116 * gimple-fold.c (replace_call_with_call_and_fold): Handle
22117 gimple_vuse copying separately from gimple_vdef copying.
22119 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22121 * value-prof.c (free_hist): Remove call to memset and the enclosing if
22124 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
22125 Olivier Hainque <hainque@adacore.com>
22127 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
22128 for all vxworks7 targets.
22129 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
22130 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
22131 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
22132 variations for VX6/VX7 and 32/64bits later on in ...
22133 (VXWORKS_LIB_SPEC): Leverage new macros.
22134 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
22135 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
22137 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
22139 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
22140 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
22142 2017-06-26 Carl Love <cel@us.ibm.com>
22144 * config/rs6000/rs6000-c.c: Add support for built-in functions
22145 vector bool char vec_reve (vector bool char);
22146 vector signed char vec_reve (vector signed char);
22147 vector unsigned char vec_reve (vector unsigned char);
22148 vector bool int vec_reve (vector bool int);
22149 vector signed int vec_reve (vector signed int);
22150 vector unsigned int vec_reve (vector unsigned int);
22151 vector bool long long vec_reve (vector bool long long);
22152 vector signed long long vec_reve (vector signed long long);
22153 vector unsigned long long vec_reve (vector unsigned long long);
22154 vector bool short vec_reve (vector bool short);
22155 vector signed short vec_reve (vector signed short);
22156 vector double vec_reve (vector double);
22157 vector float vec_reve (vector float);
22158 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
22159 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
22160 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
22161 (altivec_vreve): New pattern.
22162 * config/rs6000/altivec.h (vec_reve): New define.
22163 * doc/extend.texi (vec_rev): Update the built-in documentation file
22164 for the new built-in functions.
22166 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22168 PR tree-optimization/71815
22169 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
22171 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
22173 (slsr_process_phi): Likewise.
22174 (replace_uncond_cands_and_profitable_phis): Don't replace a
22175 multiply candidate with a stride of 1 (copy or cast).
22176 (phi_incr_cost): Call uses_consumed_by_stmt rather than
22178 (lowest_cost_path): Likewise.
22179 (total_savings): Likewise.
22181 2017-06-26 Richard Biener <rguenther@suse.de>
22184 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
22185 Use def_builtin_pure for all gather builtins.
22187 2017-06-26 Richard Biener <rguenther@suse.de>
22189 PR tree-optimization/81203
22190 * tree-tailcall.c (find_tail_calls): Do not move stmts into
22191 non-dominating BBs.
22193 2017-06-26 Marek Polacek <polacek@redhat.com>
22196 * doc/invoke.texi: Document -Wmultistatement-macros.
22198 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
22200 * doc/sourcebuild.texi (ARM-specific attributes): Document new
22201 arm_neon_ok_no_float_abi effective target.
22203 2017-06-26 Richard Biener <rguenther@suse.de>
22205 PR tree-optimization/80928
22206 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
22207 (copy_bbs): Set BB_DUPLICATED flag early.
22208 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
22210 (execute_on_shrinking_pred): Likewise.
22211 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
22212 BB_DUPLICATED blocks.
22213 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
22214 iterate over all PHIs considering removal of *gsi.
22216 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
22218 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
22221 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22223 * config/rs6000/rs6000-string.c: (expand_block_clear,
22224 do_load_for_compare, select_block_compare_mode,
22225 compute_current_alignment, expand_block_compare,
22226 expand_strncmp_align_check, expand_strn_compare,
22227 expand_block_move, rs6000_output_load_multiple)
22228 Move functions related to string/block move/compare
22229 to a separate file.
22230 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
22231 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
22232 for this function which is now used in two files.
22233 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
22234 * config.gcc: Add rs6000-string.o to extra_objs for
22235 targets powerpc*-*-* and rs6000*-*-*.
22237 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
22240 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
22241 32-bit, since indexed is not valid for DImode.
22242 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
22243 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
22244 (define_peephole2 for Altivec d-form load): Add 32-bit support.
22245 (define_peephole2 for Altivec d-form store): Likewise.
22248 * multiple_target.c (create_dispatcher_calls): Only create the
22249 dispatcher call if the function is the default clone of a
22250 versioned function.
22252 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
22254 PR middle-end/80902
22255 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
22256 a call, force the call to not be a tail call.
22258 2017-06-23 Jeff Law <law@redhat.com>
22260 * doc/contrib.texi: Add entry for Steven Pemberton's work on
22263 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
22265 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
22266 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
22267 handling for early expansion of vector shifts (sl,sr,sra,rl).
22268 (builtin_function_type): Add vector shift right instructions
22269 to the unsigned argument list.
22271 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
22273 rtl-optimizatoin/79286
22274 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
22275 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
22276 trap. PIC register plus a const unspec without offset can never trap.
22278 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
22280 * tree.h (builtin_structptr_type): New type.
22281 (builtin_structptr_types): Declare new array.
22282 * tree.c (builtin_structptr_types): New array.
22283 (free_lang_data, build_common_tree_nodes): Use it.
22285 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
22288 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
22291 2017-06-22 Matt Turner <mattst88@gmail.com>
22293 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
22294 Lake models to skylake case. Assume skylake for unknown
22295 models with clflushopt.
22297 2017-06-22 Jeff Law <law@redhat.com>
22299 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
22300 frame sizes that do not satisfy aarch64_uimm12_shift.
22302 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
22304 * profile-count.h (apply_probability,
22305 apply_scale, probability_in): Fix checks for zero.
22307 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22309 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
22310 * doc/cppdiropts.texi (-I @var{dir}): Document it.
22312 2016-06-22 Richard Biener <rguenther@suse.de>
22314 * tree-vect-loop.c (vect_model_reduction_cost): Handle
22315 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
22316 REDUC_MAX_EXPR support.
22317 (vectorizable_reduction): Likewise.
22318 (vect_create_epilog_for_reduction): Likewise.
22320 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
22322 * match.pd (A / (1 << B) -> A >> B): New.
22323 * generic-match-head.c: Include optabs-tree.h.
22324 * gimple-match-head.c: Likewise.
22325 * optabs-tree.h (target_supports_op_p): New.
22326 * optabs-tree.c (target_supports_op_p): New.
22328 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22330 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
22331 $gcc_cv_ld --help output.
22332 (gcc_cv_ld_demangle): Likewise.
22333 (gcc_cv_ld_eh_frame_hdr): Likewise.
22334 (gcc_cv_ld_pie): Likewise.
22335 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
22336 (gcc_cv_ld_buildid): Likewise.
22337 (gcc_cv_ld_sysroot): Likewise.
22338 (ld_bndplt_support): Likewise.
22339 (ld_pushpopstate_support): Likewise.
22340 * configure: Regenerate.
22341 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
22343 2017-06-21 Jakub Jelinek <jakub@redhat.com>
22346 * config/i386/sse.md (round<mode>2): Renumber match_dup and
22347 operands indexes to avoid gap between operands and match_dups.
22349 2017-06-21 Andrew Pinski <apinski@cavium.com>
22351 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
22352 Increment Arith_shift and Arith_shift_reg by 1.
22353 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
22355 * config/aarch64/aarch64.c (thunderx_tunings): Enable
22356 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
22357 (aarch64_strip_extend): Add new argument and test for it.
22358 (aarch64_cheap_mult_shift_p): New function.
22359 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
22360 add a cost if it is true.
22361 Update calls to aarch64_strip_extend.
22362 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
22364 2017-06-21 Andrew Pinski <apinski@cavium.com>
22366 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
22368 (thunderxt88): Likewise.
22369 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
22370 (thunderx_prefetch_tune): New variable.
22371 (thunderx2t99_prefetch_tune): Update for the correct values.
22372 (thunderxt88_tunings): New variable.
22373 (thunderx_tunings): Use thunderx_prefetch_tune instead of
22374 generic_prefetch_tune.
22375 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
22377 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22379 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
22380 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
22381 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
22382 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
22383 (aarch64_atomic_cas<mode>, GPI): Likewise.
22385 2017-06-21 Martin Liska <mliska@suse.cz>
22387 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
22388 statements on cold and hot labels.
22389 * predict.c (tree_estimate_probability_bb): Remove the
22390 prediction from this place.
22392 2017-06-21 Martin Liska <mliska@suse.cz>
22394 PR tree-optimization/79489
22395 * gimplify.c (maybe_add_early_return_predict_stmt): New
22397 (gimplify_return_expr): Call the function.
22398 * predict.c (tree_estimate_probability_bb): Remove handling
22400 * predict.def: Update comment about early return predictor.
22401 * gimple-predict.h (is_gimple_predict): New function.
22402 * predict.def: Change default value of early return to 66.
22403 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
22405 * passes.def: Put pass_strip_predict_hints to the beginning of
22408 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
22410 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
22411 FUNCTION_DECL declarations.
22412 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
22414 (dwaf2out_decl): Likewise.
22415 * godump.c (go_early_global_decl): Skip call to the real debug hook
22416 for FUNCTION_DECL declarations.
22417 * passes.c (rest_of_decl_compilation): Skip call to the
22418 early_global_decl debug hook for FUNCTION_DECL declarations, unless
22419 -fdump-go-spec is passed.
22421 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
22423 * config/i386/i386.c (struct builtin_isa): New field pure_p.
22424 Reorder for compactness.
22425 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
22426 (def_builtin_pure, def_builtin_pure2): New functions.
22427 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
22429 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
22431 * match.pd (nop_convert): New predicate.
22432 ((A +- CST1) +- CST2): Allow some NOP conversions.
22434 2017-06-21 Jakub Jelinek <jakub@redhat.com>
22437 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
22438 with ctors/dtors if GOVD_SHARED is set.
22440 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22442 * config/aarch64/aarch64.md (movti_aarch64):
22443 Emit mov rather than orr.
22444 (movtf_aarch64): Likewise.
22445 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
22446 Emit mov rather than orr.
22448 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22450 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
22451 Swap alternatives, make integer dup more expensive.
22453 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
22455 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
22456 Return true for non-tls symbols.
22458 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
22460 * config/aarch64/aarch64-cores.def (cortex-a55): New.
22461 (cortex-a75): Likewise.
22462 (cortex-a75.cortex-a55): Likewise.
22463 * config/aarch64/aarch64-tune.md: Regenerate.
22464 * doc/invoke.texi (-mtune): Document new values for -mtune.
22466 2017-06-21 Tom de Vries <tom@codesourcery.com>
22468 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
22469 stack_size feature.
22470 (Effective-Target Keywords, Other attributes): Suggest using
22471 dg-add-options stack_size feature to get stack limit in stack_size
22472 effective target documentation.
22474 2017-06-21 Julian Brown <julian@codesourcery.com>
22475 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22477 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
22478 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
22479 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
22481 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
22482 attribute type list for neon_multiply.
22483 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
22484 attribute type list for neon_multiply.
22485 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
22486 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
22487 attribute type list for neon_multiply.
22488 * config/arm/types.md (crypto_pmull): Add.
22489 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
22490 attribute type list.
22492 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
22494 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
22497 2017-06-20 Jakub Jelinek <jakub@redhat.com>
22499 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
22500 to make sure not to dereference a NULL cost_classes_ptr pointer.
22502 2017-06-20 Carl Love <cel@us.ibm.com>
22504 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22505 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22506 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22507 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22508 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22509 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22510 VMULOSW): New enum "unspec" values.
22511 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
22512 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
22513 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22514 altivec_vmulosw): New patterns.
22515 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22516 VMULOSW): Add definitions.
22518 2017-06-20 Julia Koval <julia.koval@intel.com>
22520 * config/i386/i386.c: Fix rounding expand for new pattern.
22521 * config/i386/subst.md: Fix pattern (parallel -> unspec).
22523 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
22525 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
22526 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
22528 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
22530 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
22533 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
22535 * config/aarch64/aarch64-cores.def: Rearrange to sort by
22536 architecture, then by implementer ID.
22537 * config/aarch64/aarch64-tune.md: Regenerate.
22539 2017-06-20 Richard Biener <rguenther@suse.de>
22541 PR middle-end/81097
22542 * fold-const.c (split_tree): Fold to type before negating.
22544 2017-06-20 David Malcolm <dmalcolm@redhat.com>
22546 * diagnostic-show-locus.c
22547 (selftest::test_fixit_deletion_affecting_newline): New function.
22548 (selftest::diagnostic_show_locus_c_tests): Call it.
22550 2017-06-20 Andreas Schwab <schwab@suse.de>
22553 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
22556 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22558 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
22559 __ARM_FEATURE_COPROC according to support.
22561 2017-06-20 Jakub Jelinek <jakub@redhat.com>
22563 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
22564 Rewritten to avoid overflow for > 32-bit pointers.
22567 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
22568 by removing enum keyword.
22569 (ubsan_type_descriptor): Likewise. Formatting fix.
22572 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
22573 splitter): Require TARGET_SSE2 in the condition.
22575 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
22578 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
22579 for doing vector set of SFmode on ISA 3.0.
22580 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
22581 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
22583 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
22584 SFmode value into a V4SF variable that was extracted from another
22585 V4SF variable without converting the element to double precision
22586 and back to single precision vector format.
22587 (vsx_insert_extract_v4sf_p9_2): Likewise.
22589 2017-06-19 Jakub Jelinek <jakub@redhat.com>
22591 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
22592 in UWHI to avoid undefined overflow.
22595 * ubsan.h (enum ubsan_encode_value_phase): New.
22596 (ubsan_encode_value): Change second argument to
22597 enum ubsan_encode_value_phase with default value of
22598 UBSAN_ENCODE_VALUE_GENERIC.
22599 * ubsan.c (ubsan_encode_value): Change second argument to
22600 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
22601 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
22602 create_tmp_var_raw instead of create_tmp_var and use a
22604 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
22605 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
22606 ubsan_encode_value callers.
22609 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
22610 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
22611 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
22613 2017-06-19 Richard Biener <rguenther@suse.de>
22615 PR middle-end/81118
22616 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
22617 estimates if we changed anything.
22619 2017-06-19 Richard Biener <rguenther@suse.de>
22621 PR tree-optimization/80887
22622 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
22623 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
22624 simplified lookups, then reset mprts_hook.
22625 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
22627 (try_to_simplify): Likewise.
22629 2017-06-19 Martin Liska <mliska@suse.cz>
22632 * gimplify.c (gimplify_switch_expr):
22633 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
22635 2017-06-19 Martin Liska <mliska@suse.cz>
22637 * doc/install.texi: Document that PGO runs in 4 stages.
22639 2017-06-19 Martin Liska <mliska@suse.cz>
22642 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
22643 to dispatcher function name.
22644 * multiple_target.c (replace_function_decl): New function.
22645 (create_dispatcher_calls): Redirect both edges and references.
22647 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
22649 * profile-count.c (profile_count::dump): Dump quality.
22650 (profile_count::differs_from_p): Update for unsigned val.
22651 * profile-count.h (profile_count_quality): New enum.
22652 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
22654 2017-06-19 Richard Biener <rguenther@suse.de>
22656 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
22657 struct function as arg.
22658 (estimate_numbers_of_iterations): Export overload with loop arg.
22659 (free_numbers_of_iterations_estimates_loop): Use an overload of
22660 free_numbers_of_iterations_estimates instead.
22661 * tree-cfg.c (remove_bb): Adjust.
22662 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
22663 * tree-parloops.c (gen_parallel_loop): Likewise.
22664 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
22666 (tree_unroll_loops_completely): Likewise.
22667 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
22668 Use an overload instead and export.
22669 (estimated_loop_iterations): Adjust.
22670 (max_loop_iterations): Likewise.
22671 (likely_max_loop_iterations): Likewise.
22672 (estimate_numbers_of_iterations): Take struct function as arg
22674 (loop_exits_before_overflow): Adjust.
22675 (free_numbers_of_iterations_estimates_loop): Use an overload.
22676 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
22677 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
22679 2017-06-19 Richard Biener <rguenther@suse.de>
22682 * ipa-prop.c (find_constructor_constant_at_offset): Handle
22683 RANGE_EXPR conservatively.
22685 2017-06-16 Carl Love <cel@us.ibm.com>
22687 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22688 definitions for vec_float, vec_float2, vec_floato,
22689 vec_floate built-ins.
22690 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
22691 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
22693 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
22694 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
22695 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
22696 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
22697 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
22698 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
22699 vec_floato): Add builtin defines.
22700 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
22701 Update the built-in documentation file for the new built-in
22704 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22706 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
22707 (mthumb): Mark as the negative of -marm.
22709 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22711 * doc/invoke.texi (ARM Options, -mcpu): Document supported
22713 (ARM Options, -mtune): Document that this accepts the same
22714 extension options as -mcpu.
22715 (ARM Options, -mfpu): Document addition of -mfpu=auto.
22717 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22719 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
22720 permitted extensions.
22722 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22724 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
22725 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
22726 (armv8-m.main): Add option +nodsp.
22727 * config/arm/arm-cpu-cdata.h: Regenerated.
22729 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22731 * config/arm/t-fuchsia: New file.
22732 * config.gcc (arm*-*-fuchsia*): Use it.
22734 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22736 * config/arm/t-symbian: Rewrite for new option infrastructure.
22738 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22740 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
22741 (MULTILIB_REQUIRED): Likewise.
22743 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22745 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
22746 (MULTILIB_RESUE): Likewise.
22747 (MULTILIB_MATCHES): Likewise.
22748 (MULTLIB_REQUIRED): Likewise.
22750 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22752 * config/arm/t-rtems: Rewrite for new option framework.
22754 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22756 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
22757 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
22758 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
22759 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
22760 * config/arm/t-multilib: ... here.
22761 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
22762 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
22763 armv7-a and armv8*-a when A-profile libraries have not been built.
22764 * config/arm/t-rmprofile: Rewrite.
22766 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22768 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
22769 with a backslash. Remove the backslash after substituting unescaped
22771 * doc/fragments.texi (MULTILIB_REUSE): Document it.
22773 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22775 * config.gcc: (arm*-*-*): When building a-profile libraries, force
22776 the driver to pass through the default setting of -mfloat-abi.
22777 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
22779 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
22780 (all_feat_combs): New rule.
22781 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
22783 * config/arm/t-aprofile: Rewrite.
22785 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22787 * config/arm/arm.h (FPUTYPE_AUTO): Define.
22788 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
22789 fpu is not specified by the user/command-line.
22790 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
22791 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
22792 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
22793 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
22794 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
22795 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
22797 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22799 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
22800 * config/arm/t-arm-elf: Rewritten.
22802 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22804 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
22805 have some floating-point instructions.
22806 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
22807 (TARGET_MAYBE_HARD_FLOAT): New macro.
22808 * config/arm/arm-builtins.c (arm_init_builtins): Use
22809 TARGET_MAYBE_HARD_FLOAT.
22810 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
22812 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22814 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
22815 (configargs.h): Include it.
22816 (arm_print_hint_for_fpu_option): New function.
22817 (arm_parse_fpu_option): New function.
22818 (candidate_extension): New class.
22819 (arm_canon_for_multilib): New function.
22820 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
22821 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
22822 (ARCH_CANONICAL_SPECS): New macro.
22823 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
22825 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22827 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
22828 are set after handling multilib fragments. Set target_cpu_default2
22831 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22833 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
22835 (arm*-*-*): Set target_cpu_default2 to a quoted string.
22836 * config/arm/parsecpu.awk (check_cpu): Validate any extension
22838 (check_arch): Likewise.
22839 * config/arm/arm.c (arm_configure_build_target): Handle
22840 TARGET_CPU_DEFAULT being a string constant. Scan any feature
22841 options in the default.
22843 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22845 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
22846 when an option is an alias of another.
22847 * config/arm/parsecpu.awk (optalias): New parser token.
22848 (gen_comm_data): Mark non-alias options as such. Emit entries
22849 for extension aliases.
22850 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
22851 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
22852 (armv6kz, armv6zk, armv6t2): Likewise.
22853 (armv7): Make vfpv3-d16 an alias.
22854 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
22856 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
22857 Sort in canonical order.
22858 (armv8-a): Sort in canonical order.
22859 (armv8.1-a, armv8.2-a): Likewise.
22860 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
22862 (cortex-a9): Sort in canonical order.
22863 * config/arm/arm.c (selftests.h): Include it.
22864 (arm_test_cpu_arch_data): New function.
22865 (arm_run_self_tests): New function.
22866 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
22867 (targetm): Move declaration to the end of the file.
22868 * arm-cpu-cdata.h: Regenerated.
22870 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22872 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
22873 call to target_mode_check describing the type of option passed.
22874 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
22875 (arm_target_thumb_only): Use arm_parse_arch_option_name or
22876 arm_parse_cpu_option_name to match parameters against list of
22878 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
22879 arm_arch_core_flags data structure.
22880 * config/arm/arm-cpu_cdata.h: Regenerated.
22882 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22884 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
22886 (arm_print_hint_for_cpu_option): Likewise.
22887 (arm_print_hint_for_arch_option): Likewise.
22888 (arm_parse_cpu_option_name): Likewise.
22889 (arm_parse_arch_option_name): Likewise.
22890 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
22891 of entries in the all_fpus list.
22892 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
22893 (arm_parse_cpu_option_name): Declare.
22894 (arm_parse_arch_option_name): Declare.
22895 (arm_parse_option_features): Declare.
22896 (arm_intialize_isa): Declare.
22897 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
22899 (gen_comm_data): ... here. Make definitions non-static.
22900 * config/arm/arm-cpu-data.h: Regenerated.
22901 * config/arm/arm-cpu-cdata.h: Regenerated.
22903 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22905 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
22906 (cpu_arch_extension): New structure.
22907 (cpu_arch_option, arch_option, cpu_option): New structures.
22908 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
22909 architecture types.
22910 (gen_data): Generate new format data tables.
22911 * config/arm/arm.c (cpu_tune): New structure.
22912 (cpu_option, processors): Delete.
22913 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
22914 (arm_print_hint_for_cpu_option): ... this and ...
22915 (arm_print_hint_for_arch_option): ... this.
22916 (arm_parse_arch_cpu_name): Delete. Replace with ...
22917 (arm_parse_cpu_option_name): ... this and ...
22918 (arm_parse_arch_option_name): ... this.
22919 (arm_unrecognized_feature): Change type of target parameter to
22921 (arm_parse_arch_cpu_features): Delete. Replace with ...
22922 (arm_parse_option_features): ... this.
22923 (arm_configure_build_target): Rework to use new configuration data
22925 (arm_print_tune_info): Rework for new configuration data tables.
22926 * config/arm/arm-cpu-data.h: Regenerated.
22927 * config/arm/arm-cpu.h: Regenerated.
22929 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22931 * Makefile.in (OBJS): Move sbitmap.o from here ...
22932 (OBJS-libcommon): ... to here.
22934 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22936 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
22937 (ISA_ALL_CRYPTO): New macro.
22938 (ISA_ALL_SIMD): New macro
22939 (ISA_ALL_FP): New macro.
22940 * config/arm/arm.c (fpu_bitlist): Update initializer.
22941 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
22943 (arm9e): Add fpu. Add option for nofp
22944 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
22945 (arm926ej-s, arm1026ej-s): Likewise.
22946 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
22947 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
22948 neon-fp16, neon-vfpv4, nofp and nosimd.
22949 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
22950 (cortex-a8): Add fpu. Add option for nofp.
22951 (cortex-a9): Add fpu. Add options for nosimd and nofp.
22952 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
22953 (cortex-r4f): Add fpu.
22954 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
22955 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
22957 (cortex-r8): Likewise.
22958 (cortex-m4): Add fpu. Add option for nofp.
22959 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
22960 (cortex-a17.cortex-a7): Likewise.
22961 (cortex-a32): Add fpu. Add options for crypto and nofp.
22962 (cortex-a35, cortex-a53): Likewise.
22963 (cortex-a57): Add fpu. Add option for crypto.
22964 (cortex-a72, cortex-a73): Likewise.
22965 (exynos-m1): Likewise.
22966 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
22967 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
22968 (cortex-m33): Add fpu. Add option for nofp.
22969 * config/arm/arm-cpu-cdata.h: Regenerated
22970 * config/arm/arm-cpu-data.h: Regenerated.
22972 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22974 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
22975 (armv5te, armv5tej): Likewise.
22976 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
22977 (armv7): Add options fp and vfpv3-d16.
22978 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
22979 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
22981 (armv7ve): Likewise.
22982 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
22983 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
22984 (armv8-a): Add nocrypto option.
22985 (armv8.1-a, armv8.2-a): Likewise.
22986 (armv8-m.main): add options fp, fp.dp and nofp.
22988 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
22990 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
22992 (armv8-a+crc): Delete.
22993 (armv8.1-a): Add options simd, crypto and nofp.
22994 (armv8.2-a): Add options fp16, simd, crypto and nofp.
22995 (armv8.2-a+fp16): Delete.
22996 (armv8-m.main): Add option dsp.
22997 (armv8-m.main+dsp): Delete.
22998 (cortex-a8): Add fpu. Add nofp option.
22999 (cortex-a9): Add fpu. Add nofp and nosimd options.
23000 * config/arm/parsecpu.awk (gen_data): Generate option tables and
23001 link to main cpu and architecture data structures.
23002 (gen_comm_data): Only put isa attributes from the main architecture
23004 (option): New statement for architecture and CPU entries.
23005 * arm.c (struct cpu_option): New structure.
23006 (struct processors): Add entry for options.
23007 (arm_unrecognized_feature): New function.
23008 (arm_parse_arch_cpu_name): Ignore any characters after the first
23010 (arm_parse_arch_cpu_feature): New function.
23011 (arm_configure_build_target): Separate out any CPU and architecture
23012 features and parse separately. Don't error out if -mfpu=auto is
23013 used with only an architecture string.
23014 (arm_print_asm_arch_directives): New function.
23015 (arm_file_start): Call it.
23016 * config/arm/arm-cpu-cdata.h: Regenerated.
23017 * config/arm/arm-cpu-data.h: Likewise.
23018 * config/arm/arm-tables.opt: Likewise.
23020 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23022 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
23023 assembler when it is not -mfpu=auto.
23025 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23027 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
23028 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
23029 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
23030 (ASM_CPU_SPEC): Rewrite.
23031 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
23032 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
23033 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
23034 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
23035 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
23036 copied string is NUL-terminated. Also strip any characters prefixed
23038 (arm_rewrite_selected_arch): New function.
23039 (arm_rewrite_march): New function.
23041 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23043 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
23044 (x_arm_cpu_string, x_arm_tune_string): Likewise.
23045 (march, mcpu, mtune): Convert to string-based options.
23046 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
23047 (arm_parse_arch_cpu_name): New function.
23048 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
23049 identify selected architecture or CPU.
23050 (arm_option_save): New function.
23051 (TARGET_OPTION_SAVE): Redefine.
23052 (arm_option_restore): Restore string options.
23053 (arm_option_print): Print string options.
23055 2017-06-16 Martin Sebor <msebor@redhat.com>
23057 PR tree-optimization/80933
23058 PR tree-optimization/80934
23059 * builtins.c (fold_builtin_3): Do not handle bcmp here.
23060 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
23061 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
23062 (gimple_fold_builtin): Call them.
23064 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23066 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
23067 as unlikely; update profile.
23069 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23071 * predict.c (force_edge_cold): Handle declaring edges impossible
23074 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23076 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
23078 (try_unroll_loop_completely): Fix reporting.
23080 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23082 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
23084 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
23087 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
23088 if given a non-constant argument for an intrinsic which requires a
23091 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23093 * profile.c (compare_freqs): New function.
23094 (branch_prob): Sort edge list.
23095 (find_spanning_tree): Assume that the list is priority sorted.
23097 2017-06-16 Richard Biener <rguenther@suse.de>
23099 PR tree-optimization/81090
23100 * passes.def (pass_record_bounds): Remove.
23101 * tree-pass.h (make_pass_record_bounds): Likewise.
23102 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
23103 make_pass_record_bounds): Likewise.
23104 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
23105 not free niter estimates at the beginning but at the end.
23106 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
23108 2017-06-16 Richard Biener <rguenther@suse.de>
23110 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
23111 initializer to workaround ICE in host GCC 4.8.
23113 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23115 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
23117 (clone_inlined_nodes): Update.
23119 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23121 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
23122 prefetch settings, and enable prefetching by default at -O3.
23124 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23126 * config/aarch64/aarch64.c (aarch64_override_options_internal):
23127 Set flag_prefetch_loop_arrays according to tuning data.
23129 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23131 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
23132 New tune structure.
23133 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
23134 [Unrelated to main purpose of the patch] Place the pointer field last
23135 to enable type checking errors when tune structure are wrongly merged.
23136 * config/aarch64/aarch64.c (generic_prefetch_tune,)
23137 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
23138 (thunderx2t99_prefetch_tune): New tune constants.
23139 (tune_params *_tunings): Update all tunings (no functional change).
23140 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
23141 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
23142 from tunings structures.
23144 2017-06-16 Jakub Jelinek <jakub@redhat.com>
23147 * ubsan.c (instrument_null): Add T argument, use it instead
23148 of computing it based on IS_LHS.
23149 (instrument_object_size): Likewise.
23150 (pass_ubsan::execute): Adjust instrument_null and
23151 instrument_object_size callers to pass gimple_get_lhs or
23152 gimple_assign_rhs1 result to it. Use instrument_null instead of
23153 calling get_base_address and instrument_mem_ref. Handle
23154 aggregate call arguments for object-size sanitization.
23156 2017-06-16 Yury Gribov <tetra2005@gmail.com>
23158 PR tree-optimization/81089
23159 * tree-vrp.c (is_masked_range_test): Validate operands of
23162 2017-06-15 Martin Sebor <msebor@redhat.com>
23165 * dumpfile.c (dump_register): Avoid calling memset to initialize
23166 a class with a default ctor.
23167 * gcc.c (struct compiler): Remove const qualification.
23168 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
23169 * hash-table.h: Ditto.
23170 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
23172 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
23173 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
23175 * params.h (struct param_info): Make struct members non-const.
23176 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
23177 with default initialization.
23178 * vec.h (vec_copy_construct, vec_default_construct): New helper
23180 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
23181 with vec_copy_construct.
23182 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
23183 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
23184 * doc/invoke.texi (-Wclass-memaccess): Document.
23186 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23188 * emit-rtl.h (is_leaf): Update comment about local
23189 register allocator.
23191 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23194 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
23195 for a variable to have a section before checking if the section has a
23197 Set section to.persistent if persistent attribute is set.
23198 Warn if .persistent attribute is used on an automatic variable.
23200 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
23202 PR rtl-optimization/80474
23203 * reorg.c (update_block): Do not ignore instructions in a delay slot.
23205 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
23207 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
23210 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
23212 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
23213 (casesi): Emit bounds checking as RTL.
23214 (casesi_internal_mips16_<mode>): Remove bounds checking.
23216 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
23218 * config/xtensa/xtensa.c (xtensa_option_override): Append
23219 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
23220 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
23221 xtensa_doloop_hooks): Define unconditionally.
23222 (xtensa_reorg_loops): Only call reorg_loops in the presence of
23224 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
23225 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
23226 for it in xtensa_option_override.
23227 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
23228 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
23230 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
23232 * doc/cppopts.texi: Document '-' special value to -MF.
23234 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
23236 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
23237 (cortex_a53_fconst): Likewise.
23238 (cortex_a53_fpmul): Likewise.
23239 (cortex_a53_f_load_64): Likewise.
23240 (cortex_a53_f_load_many): Likewise.
23241 (cortex_a53_advsimd_alu): Likewise.
23242 (cortex_a53_advsimd_alu_q): Likewise.
23243 (cortex_a53_advsimd_mul): Likewise.
23244 (cortex_a53_advsimd_mul_q): Likewise.
23245 (fpmac bypass): Add new bypass for fpmac-fpmac case.
23246 Add missing fmul, r2f_cvt and fconst cases.
23248 2017-06-14 Richard Biener <rguenther@suse.de>
23250 PR middle-end/81088
23251 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
23253 (fold_binary_loc): When associating do not treat pre-existing
23254 TREE_OVERFLOW on literal constants as a reason to allow
23255 TREE_OVERFLOW on associated literal constants.
23257 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23259 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
23260 (MASK_FEATURES): New macro.
23261 * config/sparc/sparc.c (sparc_option_override): Remove the special
23262 handling of -mfpu and generalize it to all MASK_FEATURES switches.
23264 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23266 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
23267 a division of 0 if non-call exceptions are enabled.
23269 2017-06-14 Andrew Pinski <apinski@cavium.com>
23270 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23273 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
23274 Improve vector initialization code gen for only variable case.
23276 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23278 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
23280 2017-06-14 Richard Biener <rguenther@suse.de>
23282 PR tree-optimization/81083
23283 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
23286 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23288 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
23289 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
23290 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
23291 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
23292 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
23293 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
23295 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23297 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
23298 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
23300 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23302 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
23304 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23306 * config/rs6000/t-rtems: Don't handle SPE.
23308 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23310 * config/rs6000/t-linux: Don't handle SPE.
23312 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23314 * config/rs6000/eabispe.h: Delete file.
23316 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23318 * config/rs6000/t-spe: Delete file.
23320 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23322 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
23323 (rs6000_legitimate_offset_address_p): Return false for anything in
23324 V2SImode or V2SFmode.
23326 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23328 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
23329 except V2SF and V2SI. Rearrange the vector modes, and add comments.
23330 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
23332 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
23333 (rs6000_legitimate_offset_address_p): Ditto.
23334 (rs6000_emit_move): Ditto.
23335 (rs6000_init_builtins): Remove V4HI_type_node.
23337 2017-06-13 Martin Liska <mliska@suse.cz>
23340 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
23341 (gate_asan): Likewise.
23342 * asan.h (asan_no_sanitize_address_p): Remove the function.
23343 (sanitize_flags_p): New function.
23344 * builtins.def: Fix coding style.
23345 * common.opt: Use renamed enum value.
23346 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
23347 * doc/extend.texi: Document no_sanitize attribute.
23348 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
23349 to SANITIZE_UNDEFINED_NONDEFAULT.
23350 * gcc.c (sanitize_spec_function): Use the renamed enum value.
23351 * gimple-fold.c (optimize_atomic_compare_exchange_p):
23352 Use sanitize_flags_p.
23353 * gimplify.c (gimplify_function_tree): Likewise.
23354 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
23355 * opts.c (parse_no_sanitize_attribute): New function.
23356 (common_handle_option): Use renamed enum value.
23357 * opts.h (parse_no_sanitize_attribute): Declare.
23358 * tree.c (sanitize_flags_p): New function.
23359 * tree.h: Declared here.
23360 * tsan.c: Use sanitize_flags_p.
23361 * ubsan.c (ubsan_expand_null_ifn): Likewise.
23362 (instrument_mem_ref): Likewise.
23363 (instrument_bool_enum_load): Likewise.
23364 (do_ubsan_in_current_function): Remove the function.
23365 (pass_ubsan::execute): Use sanitize_flags_p.
23366 * ubsan.h: Remove do_ubsan_in_current_function
23367 * tree-cfg.c (print_no_sanitize_attr_value): New function.
23368 (dump_function_to_file): Use it here.
23370 2017-06-13 Martin Jambor <mjambor@suse.cz>
23372 PR tree-optimization/80803
23373 PR tree-optimization/81063
23374 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
23375 (propagate_subaccesses_across_link): Enqueue subtree whenever
23376 necessary instead of relying on the caller.
23378 2017-06-13 Martin Jambor <mjambor@suse.cz>
23380 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
23381 that have a first_link.
23382 (sort_and_splice_var_accesses): Do not check first_link before
23384 (subtree_mark_written_and_enqueue): Likewise.
23385 (propagate_all_subaccesses): Likewise and do not stop at first
23386 parent with a first_link.
23388 2017-06-13 Martin Jambor <mjambor@suse.cz>
23390 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
23393 2017-06-13 Yury Gribov <tetra2005@gmail.com>
23395 * match.pd: New pattern.
23397 2017-06-13 Yury Gribov <tetra2005@gmail.com>
23399 * tree-vrp.c (is_masked_range_test): New function.
23400 (register_edge_assert_for): Determine ranges for
23403 2017-06-13 Yury Gribov <tetra2005@gmail.com>
23405 PR tree-optimization/67328
23406 * fold-const.c (maskable_range_p): New function.
23407 (build_range_check): Generate bittests if possible.
23409 2017-06-13 Martin Liska <mliska@suse.cz>
23411 * gimple-pretty-print.c (dump_probability): Add new argument.
23412 (dump_edge_probability): Dump both probability and count.
23413 (dump_gimple_label): Likewise.
23414 (dump_gimple_bb_header): Likewise.
23416 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
23419 * config/avr/avr-devices.c: Fix indentation.
23420 * config/avr/gen-avr-mmcu-specs.c: Dito.
23422 2017-06-13 Richard Biener <rguenther@suse.de>
23424 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
23425 instead get vector type from stmt_info.
23426 (vectorizable_reduction): Adjust. Remove dead code.
23428 2017-06-13 Richard Biener <rguenther@suse.de>
23430 PR middle-end/81065
23431 * fold-const.c (extract_muldiv_1): Remove bogus distribution
23432 case of C * (x * C2 + C3).
23433 (fold_addr_of_array_ref_difference): Properly fold index difference.
23435 2017-06-12 David S. Miller <davem@davemloft.net>
23438 * config/sparc/sparc.md (return expander): Emit frame blockage if
23439 function uses alloca.
23441 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
23443 * combine.c (make_field_assignment): Check len rather than the mode
23444 precision when calling force_to_mode.
23446 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
23448 Support multilibs and devices that see flash in RAM address range.
23451 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
23452 (avr_mcu_t) <flash_pm_offset>: New field.
23453 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
23454 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
23455 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
23456 (AVR_TINY_PM_OFFSET): Remove macro.
23457 * config/avr/avr.opt (-mshort-calls): New option.
23458 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23459 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
23460 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
23461 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
23462 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
23463 instead of avr_arch->have_jmp_call.
23464 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
23465 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
23466 avr_arch->flash_pm_offset to define.
23467 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
23468 new field flash_pm_offset. Add entry for avrxmega3.
23469 (avr_texinfo): Add entry for avrxmega3.
23470 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
23471 attiny212, attiny214,
23472 attiny412, attiny414, attiny416, attiny417,
23473 attiny814, attiny816, attiny817,
23474 attiny1614, attiny1616, attiny1617,
23475 attiny3214, attiny3216, attiny3217.
23476 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
23477 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
23478 (avr_print_operand_address) [AVR_TINY]: Same.
23479 (avr_asm_init_sections) <readonly_data_section>: Only patch
23480 callback if avr_arch->flash_pm_offset = 0.
23481 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
23482 for rodata if avr_arch->flash_pm_offset != 0.
23483 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
23484 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
23485 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
23486 (m_options): Append opt_rcall.
23487 (m_dirnames): Append dir_rcall.
23488 * config/avr/t-multilib: Regenerate.
23490 * configure.ac [target=avr]: Check whether avrxmega3 default
23491 linker description file works as needed.
23492 * configure: Regenerate.
23493 * doc/avr-mmcu.texi: Regenerate.
23494 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
23495 <__AVR_ARCH__>: Document avrxmega3 and 103.
23496 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
23497 <__AVR_SHORT_CALLS__>: Document it.
23498 <__AVR_PM_BASE_ADDRESS__>: Document it.
23499 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
23500 (AVR Variable Attributes) <progmem>: Document this is
23501 not needed for avrxmega3.
23502 (AVR Named Address Spaces) <__flash>: Dito.
23504 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
23506 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
23508 2017-06-12 Doug Rupp <rupp@adacore.com>
23510 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
23511 Append vxworks-stdint.h to the tm_file list.
23512 * config/vxworks-stdint.h: New file.
23514 2017-06-12 Martin Liska <mliska@suse.cz>
23516 PR tree-optimization/81041
23517 * tree-profile.c (gimple_gen_ic_func_profiler):
23518 Create an extra BB in profile-generate
23519 (gimple_gen_time_profiler): Likewise.
23521 2017-06-12 Jakub Jelinek <jakub@redhat.com>
23523 PR tree-optimization/81003
23524 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
23525 (update_range_test): Use it instead of force_gimple_operand_gsi.
23527 2017-06-12 Richard Biener <rguenther@suse.de>
23529 PR tree-optimization/81053
23530 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
23531 with backedge value not defined in loop. Simplify def stmt
23534 2017-06-11 Tom de Vries <tom@codesourcery.com>
23537 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
23539 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
23540 nvptx_cannot_force_const_mem.
23542 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23544 * opts.c (finish_options): Move test for flag_split_stack after
23545 it has been initialized.
23547 2017-06-11 Jason Merrill <jason@redhat.com>
23549 * tree.h (id_equal): New.
23550 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
23551 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
23552 instead of strcmp of IDENTIFIER_POINTER.
23554 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23556 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
23557 (mark_all_inlined_calls_cdtor): Fix formating.
23558 (inline_transform): Rescale profile before inlining.
23560 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23562 * cgraph.h (cgraph_edge::clone): Update prototype.
23563 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
23564 (cgraph_node::create_clone): Update.
23565 (cgraph_node::create_version_clone): Update.
23566 * tree-inline.c (copy_bb): Update.
23567 (expand_call_inline): Update.
23569 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
23571 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
23572 factored out from ...
23573 (rs6000_emit_prologue): ... here.
23575 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
23577 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
23578 factored out from ...
23579 (rs6000_emit_prologue): ... here.
23581 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23583 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
23585 (handle_missing_profiles): Fix computation of tp_first_run.
23586 (counts_to_freqs): Do not touch freqs when count is 0.
23588 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
23590 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
23593 2017-06-10 Tom de Vries <tom@codesourcery.com>
23595 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
23596 attributes): Document signal effective target.
23598 2017-06-10 Tom de Vries <tom@codesourcery.com>
23600 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
23601 Document effective target stack_size.
23603 2017-06-09 David Malcolm <dmalcolm@redhat.com>
23605 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
23606 to the edit_context if they can be auto-applied.
23608 2017-06-9 Ian Lance Taylor <iant@golang.org>
23610 * opts.c (finish_options): If -fsplit-stack, disable implicit
23611 -forder-blocks-and-partition.
23612 * doc/invoke.texi (Optimize Options): Document that when using
23613 -fsplit-stack -forder-blocks-and-partition is not implicitly
23616 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23618 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
23619 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
23620 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
23621 * builtins.def (abort, trap, unreachable): Declare cold.
23622 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
23623 * tree-core.h (ECF_COLD): New.
23624 * tree.c (set_call_expr_flags): Handle ECF_COLD.
23625 (build_common_builtin_nodes): Mark unreachable and abort as cold.
23627 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23629 * predict.c (unlikely_executed_stmt_p): Cleanup.
23631 2017-06-09 Richard Biener <rguenther@suse.de>
23633 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
23634 model if the ref is always written to.
23636 2017-06-09 Tamar Christina <tamar.christina@arm.com>
23638 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
23640 2017-06-09 Tamar Christina <tamar.christina@arm.com>
23642 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
23645 2017-06-09 Tom de Vries <tom@codesourcery.com>
23648 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
23649 "target cannot support label values" when encountering LABEL_REF.
23651 2017-06-09 Martin Liska <mliska@suse.cz>
23653 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
23654 (gimple_gen_ic_func_profiler): Emit direct comparison
23655 of __gcov_indirect_call_callee with NULL.
23656 (gimple_gen_time_profiler): Change probability from
23657 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
23659 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23661 * profile.c (edge_gcov_counts): Turn to pointer.
23662 (compute_branch_probabilities, compute_branch_probabilities): Update.
23663 (branch_prob): Do not clear edge_gcov_count.
23664 * profile.h (edge_gcov_counts): Turn to pointer.
23665 (edge_gcov_count): Update.
23667 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
23669 * gimple.h (gimple_check_failed): Mark cold.
23671 2017-06-09 Richard Biener <rguenther@suse.de>
23673 PR tree-optimization/66623
23674 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
23675 refactor check_reduction into two parts, properly computing
23676 whether we have to check reduction validity for outer loop
23679 2017-06-09 Richard Biener <rguenther@suse.de>
23681 PR tree-optimization/79483
23682 * graphite-scop-detection.c (order): New global.
23683 (get_order): Compute bb to order mapping that satisfies code
23684 generation constraints.
23685 (cmp_pbbs): New helper.
23686 (build_scops): Start domwalk at entry block, sort generated
23689 2017-06-09 Richard Biener <rguenther@suse.de>
23691 PR middle-end/81007
23692 * ipa-polymorphic-call.c
23693 (ipa_polymorphic_call_context::restrict_to_inner_class):
23694 Skip FIELD_DECLs with error_mark_node type.
23695 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
23698 2017-06-09 Martin Liska <mliska@suse.cz>
23700 * predict.c (struct branch_predictor): New struct.
23701 (test_prediction_value_range): New test.
23702 (predict_c_tests): New function.
23703 * selftest-run-tests.c (selftest::run_tests): Run the function.
23704 * selftest.h: Declare new tests.
23706 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
23709 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
23710 gen_add3_insn did not fail.
23711 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
23712 r0, construct that number in a temporary reg and add that reg to r0.
23713 If asked to put the result in r0 as well, fail.
23715 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
23717 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23718 for early expansion of vec_eqv.
23720 2017-06-08 Jakub Jelinek <jakub@redhat.com>
23722 PR middle-end/81005
23723 * ubsan.c (instrument_null): Avoid pointless code temporary.
23724 (pass_ubsan::execute): Instrument aggregate arguments of calls.
23726 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
23730 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
23733 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
23736 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23738 * predict.c (unlikely_executed_edge_p): Move ahead.
23739 (probably_never_executed_edge_p): Use it.
23741 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
23743 PR middle-end/79988
23744 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
23745 gimple_call_builtin_p call.
23747 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23749 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
23750 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
23751 rtl_check_failed_type2, rtl_check_failed_code1,
23752 rtl_check_failed_code2, rtl_check_failed_code_mode,
23753 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
23754 rtvec_check_failed_bounds, rtl_check_failed_flag,
23755 _fatal_insn_not_found, _fatal_insn): Likewise.
23756 * tree.h (tree_contains_struct_check_failed,
23757 tree_check_failed, tree_not_check_failed,
23758 tree_class_check_failed, tree_range_check_failed,
23759 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
23760 tree_vec_elt_check_failed, phi_node_elt_check_failed,
23761 tree_operand_check_failed, omp_clause_check_failed,
23762 omp_clause_operand_check_failed, omp_clause_range_check_failed):
23765 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23767 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
23768 flag_branch_probabilities.
23769 * ipa-inline.c (edge_badness): Likewise.
23770 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
23771 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
23772 * predict.c (maybe_hot_frequency_p): Likewise.
23773 (probably_never_executed): Likewise.
23774 * sched-ebb.c (schedule_ebbs): Likewise.
23775 * sched-rgn.c (find_single_block_region): Likewise.
23776 * tracer.c (tail_duplicate): Likewise.
23778 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23780 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
23781 longer requires x_flag_profile_use.
23783 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23785 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
23786 instead of flag_reorder_blocks_and_partition.
23787 * dbxout.c (dbxout_function_end): Likewise.
23788 * dwarf2out.c (gen_subprogram_die): Likewise.
23789 * haifa-sched.c (sched_create_recovery_edges): Likewise.
23790 * hw-doloop.c (reorg_loops): Likewise.
23791 * varasm.c (assemble_start_function,
23792 assemble_end_function): Likewise.
23793 (decide_function_section): Do not check for
23794 flag_reorder_blocks_and_partition.
23796 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
23798 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
23800 (chkp_get_hard_register_fake_addr_expr): Ditto.
23801 (chkp_build_addr_expr): Add check for hard reg case.
23802 (chkp_parse_array_and_component_ref): Ditto.
23803 (chkp_find_bounds_1): Ditto.
23804 (chkp_process_stmt): Don't generate bounds store for
23807 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23809 * predict.c (maybe_hot_bb_p): Do not check profile status.
23810 (maybe_hot_edge_p): Likewise.
23811 (probably_never_executed): Check for zero counts even if profile
23813 (unlikely_executed_edge_p): New function.
23814 (unlikely_executed_stmt_p): New function.
23815 (unlikely_executed_bb_p): New function.
23816 (set_even_probabilities): Use unlikely predicates.
23817 (combine_predictions_for_bb): Likewise.
23818 (predict_paths_for_bb): Likewise.
23819 (predict_paths_leading_to_edge): Likewise.
23820 (determine_unlikely_bbs): New function.
23821 (estimate_bb_frequencies): Use it.
23822 (compute_function_frequency): Use zero counts even if profile is
23824 * profile-count.h: Fix typo.
23826 2017-08-08 Julia Koval <julia.koval@intel.com>
23828 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
23829 _mm512_mask_cvtsepi16_storeu_epi8,
23830 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
23831 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
23832 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
23833 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
23834 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
23835 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
23836 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
23837 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
23838 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
23839 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
23840 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
23841 __builtin_ia32_pmovuswb256mem_mask,
23842 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
23843 __builtin_ia32_pmovwb512mem_mask): New builtins.
23845 2017-08-08 Julia Koval <julia.koval@intel.com>
23847 PR target/73350,80862
23848 * config/i386/subst.md (round): Fix round pattern.
23849 * config/i386/i386.c (ix86_erase_embedded_rounding):
23850 Fix erasing rounding for the fixed pattern.
23852 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
23854 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
23856 2017-06-08 Martin Liska <mliska@suse.cz>
23858 PR gcov-profile/80911
23859 * gcov.c (block_info::block_info): New constructor.
23861 2017-06-07 Carl Love <cel@us.ibm.com>
23863 * config/rs6000/rs6000-c: The return type of the following
23864 built-in functions was implemented as int not long long. Fix sign
23865 of return value for the unsigned version of vec_mulo and vec_mule.
23866 vector unsigned long long vec_bperm (vector unsigned long long,
23867 vector unsigned char)
23868 vector signed long long vec_mule (vector signed int,
23870 vector unsigned long long vec_mule (vector unsigned int,
23871 vector unsigned int)
23872 vector signed long long vec_mulo (vector signed int,
23874 vector unsigned long long vec_mulo (vector unsigned int,
23875 vector unsigned int)
23876 * doc/extend.texi: Fix the documentation for the built-in
23879 2017-06-07 Carl Love <cel@us.ibm.com>
23882 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
23885 2017-06-07 Carl Love <cel@us.ibm.com>
23887 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
23888 support, Generate doublehv for signed int/float for BE case only.
23890 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
23892 * doc/invoke.texi (mcx16): Rewrite.
23894 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23896 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
23897 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
23898 *mov<mode>_softfloat, and an anonymous splitter): Use
23899 nonimmediate_operand instead of rs6000_nonimmediate_operand.
23901 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23903 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
23905 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
23906 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
23907 (rs6000_debug_reg_global): Adjust.
23908 (rs6000_init_hard_regno_mode_ok): Adjust.
23909 (rs6000_dbx_register_number): Adjust.
23910 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
23911 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
23912 Remove SPE_ACC and SPEFSCR.
23913 (REG_ALLOC_ORDER): Ditto.
23914 (FRAME_POINTER_REGNUM): Change to 111.
23915 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
23916 (REG_CLASS_NAMES): Ditto.
23917 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
23918 (REGISTER_NAMES): Ditto.
23919 (ADDITIONAL_REG_NAMES): Ditto.
23920 (rs6000_reg_names): Ditto.
23921 * config/rs6000/rs6000.md: Renumber some register number
23924 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23926 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
23928 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
23929 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
23931 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
23932 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
23933 Delete the SPE high registers.
23934 (REG_ALLOC_ORDER): Ditto.
23935 (enum reg_class): Remove SPE_HIGH_REGS.
23936 (REG_CLASS_NAMES): Ditto.
23937 (REG_CLASS_CONTENTS): Delete the SPE high registers.
23938 (REGISTER_NAMES): Ditto.
23939 (rs6000_reg_names): Ditto.
23940 * doc/tm.texi.in: Remove SPE as example.
23941 * doc/tm.texi: Regenerate.
23943 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23945 * config/rs6000/8540.md (ppc8540_brinc): Delete.
23946 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
23947 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
23948 * config/rs6000/rs6000.md (type): Remove "brinc".
23950 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23952 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
23953 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
23954 * config/rs6000/linuxspe.h: Delete file.
23955 * config/rs6000/rs6000.md: Don't include spe.md.
23956 * config/rs6000/spe.h: Delete file.
23957 * config/rs6000/spe.md: Delete file.
23958 * config/rs6000/t-rs6000: Remove spe.md.
23960 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23962 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
23963 (reg_or_none500mem_operand): Delete.
23964 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
23965 instead of reg_or_none500mem_operand.
23967 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23969 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23970 handling of SPE flags.
23971 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
23973 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
23975 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
23977 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
23978 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
23979 paired_divv2sf3): Similar.
23980 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
23981 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
23982 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
23984 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
23985 Rename the paired_* instruction patterns.
23986 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
23988 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
23989 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
23990 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
23991 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
23992 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
23993 PAIRED_VECTOR_MODE.
23994 (struct machine_function): Delete field spe_insn_chain_scanned_p.
23995 (spe_func_has_64bit_regs_p): Delete.
23996 (spe_expand_predicate_builtin): Delete.
23997 (spe_expand_evsel_builtin): Delete.
23998 (TARGET_DWARF_REGISTER_SPAN): Do not define.
23999 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
24000 (invalid_e500_subreg): Delete.
24001 (rs6000_legitimize_address): Always force_reg op2 as well, for
24002 paired single memory accesses.
24003 (rs6000_member_type_forces_blk): Delete.
24004 (rs6000_spe_function_arg): Delete.
24005 (rs6000_expand_unop_builtin): Delete SPE handling.
24006 (rs6000_expand_binop_builtin): Ditto.
24007 (spe_expand_stv_builtin): Delete.
24008 (bdesc_2arg_spe): Delete.
24009 (spe_expand_builtin): Delete.
24010 (spe_expand_predicate_builtin): Delete.
24011 (spe_expand_evsel_builtin): Delete.
24012 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
24013 (spe_init_builtins): Delete.
24014 (spe_func_has_64bit_regs_p): Delete.
24015 (savres_routine_name): Delete "info" parameter. Adjust callers.
24016 (rs6000_emit_stack_reset): Ditto.
24017 (rs6000_dwarf_register_span): Delete.
24018 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
24019 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
24020 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
24022 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
24024 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
24025 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
24026 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
24027 mulv2sf3, divv2sf3): Delete expanders.
24029 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24031 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
24033 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24035 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
24036 * config/rs6000/rs6000.c: Ditto.
24038 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24040 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
24041 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
24042 comparison_operator.
24044 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24046 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
24047 * config/rs6000/rs6000.opt: Ditto.
24048 * config/rs6000/t-rtems: Ditto.
24050 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24052 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
24053 TARGET_E500_SINGLE by 0, simplify.
24054 * config/rs6000/rs6000.c: Ditto.
24055 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
24056 (spe_build_register_parallel): Delete.
24057 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
24058 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
24059 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
24060 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
24061 (E500_CONVERT): Delete.
24062 * config/rs6000/spe.md: Remove many patterns and all define_constants.
24064 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24066 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
24067 * config/rs6000/dfp.md: Ditto.
24068 (negdd2, *negdd2_fpr): Merge.
24069 (absdd2, *absdd2_fpr): Merge.
24070 (negtd2, *negtd2_fpr): Merge.
24071 (abstd2, *abstd2_fpr): Merge.
24072 * config/rs6000/e500.h: Delete file.
24073 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
24074 TARGET_FPRS by 1 and simplify.
24075 * config/rs6000/rs6000-c.c: Ditto.
24076 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
24077 TARGET_DF_SPE by 0.
24078 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
24080 * config/rs6000/rs6000.md: Ditto.
24081 (floatdidf2, *floatdidf2_fpr): Merge.
24082 (move_from_CR_gt_bit): Delete.
24083 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
24084 (E500_CR_IOR_COMPARE): Delete.
24085 (All patterns that require !TARGET_FPRS): Delete.
24086 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
24088 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24090 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
24092 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24094 * graphds.c (add_edge): Intitialize edge's attached data.
24095 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
24096 pointer parameter. Call pointed function on each edge during
24097 graph traversing. Skip traversing the edge when the function
24099 (graphds_dfs, graphds_scc): Ditto.
24100 (for_each_edge): New parameter. Pass the new parameter to callback
24102 * graphds.h (skip_edge_callback): New function pointer type.
24103 (graphds_dfs, graphds_scc): New function pointer parameter.
24104 (graphds_edge_callback, for_each_edge): New parameter.
24106 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24108 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
24109 out code checking if runtime alias check is possible to below ...
24110 Call the new function.
24111 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
24112 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
24114 2017-06-07 Marek Polacek <polacek@redhat.com>
24117 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
24118 TYPE_OVERFLOW_WRAPS check.
24120 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24122 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
24123 if versioning is required.
24124 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
24125 peeling with the check for versioning.
24127 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24129 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
24130 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
24131 Set true to new parameter if new ssa variable is defined.
24132 (vect_gen_vector_loop_niters): Refactor. Set range information
24133 for the new vector loop bound variable.
24134 (vect_do_peeling): Ditto.
24136 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24138 * tree-affine.c (ssa.h): Include header file.
24139 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
24140 has wrapping overflow behavior.
24142 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24144 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
24146 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24148 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
24149 (tree_to_aff_combination): ... here.
24151 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24153 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
24154 reg_pressure model function.
24155 (ivopts_global_cost_for_size): Delete.
24156 (determine_set_costs, iv_ca_recount_cost): Call new model function
24157 ivopts_estimate_reg_pressure.
24159 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24161 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
24162 expensive than udiv. Remove floating point cases from mod.
24164 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24166 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
24167 Increase idiv cost.
24169 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24171 * config/aarch64/aarch64.md
24172 (copysignsf3): Fix mask generation.
24174 2017-06-07 Jakub Jelinek <jakub@redhat.com>
24176 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
24178 (class dump_manager): Add register_dumps method.
24179 * dumpfile.c: Include langhooks.h.
24180 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
24181 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
24182 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
24183 (dump_manager::dump_register): Start with 512 entries instead of 32.
24184 (dump_manager::register_dumps): New method.
24185 * toplev.c (general_init): Instead of invoking register_dumps
24186 langhook, invoke register_dumps method on the dump manager.
24187 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
24190 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
24192 * doc/md.texi: Clarify the restrictions on a define_insn condition.
24193 Say that # requires an associated define_split to exist, and that
24194 the define_split must be suitable for use after register allocation.
24196 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24198 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
24199 (compute_outgoing_frequencies): Also initialize zero counts.
24200 (find_many_sub_basic_blocks): Do not produce uninitialized profile
24201 around loops; preserve more of profile when nothing changes.
24203 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
24205 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
24207 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
24208 * config/arm/arm-cpu-cdata.h: Regenerate.
24209 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
24210 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
24211 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
24212 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
24213 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
24215 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
24217 * config/arm/t-rmprofile: Likewise.
24218 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
24220 2017-06-06 David S. Miller <davem@davemloft.net>
24223 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
24224 blockage if function uses alloca.
24226 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24228 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
24229 New "uid" fields to hold pretty-print IDs of group and ref.
24230 Memory references are now identified as <group_id>:<ref_id>
24231 instead of using [random] addresses.
24232 (dump_mem_details): Simplify, no functional change.
24233 (dump_mem_ref): Simplify and make output more concise.
24234 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
24235 (find_or_create_group): Initialize group uid.
24236 (record_ref): Initialize ref uid. Improve debug output.
24237 (prune_group_by_reuse, should_issue_prefetch_p,)
24238 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
24239 (mark_nontemporal_store, determine_loop_nest_reuse):
24240 Improve debug output.
24242 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24244 * dbgcnt.def (prefetch): New debug counter.
24245 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
24246 (schedule_prefetches): Stop issueing prefetches if debug counter
24249 2017-06-06 Tom de Vries <tom@codesourcery.com>
24251 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
24252 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
24254 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24256 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
24257 Use aarch64_reg_or_zero predicate for operand 4.
24258 (aarch64_compare_and_swap<mode> define_insn_and_split):
24259 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
24260 (aarch64_store_exclusive<mode>): Likewise for operand 2.
24262 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24264 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
24265 (arm_compute_save_core_reg_mask): This.
24266 (thumb1_compute_save_reg_mask): Rename into ...
24267 (thumb1_compute_save_core_reg_mask): This.
24268 (arm_compute_save_reg0_reg12_mask): Adapt comment.
24269 (arm_compute_frame_layout): Likewise.
24271 2017-06-06 Richard Biener <rguenther@suse.de>
24273 PR tree-optimization/80974
24274 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
24275 keep or clear leaders SSA info.
24277 2017-06-06 Tom de Vries <tom@codesourcery.com>
24279 * config/nvptx/nvptx.c (split_mode_p): New function.
24280 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
24282 2017-06-06 Tom de Vries <tom@codesourcery.com>
24284 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
24286 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24289 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
24292 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24294 * shrink-wrap.c (handle_simple_exit): Update profile.
24295 (try_shrink_wrapping): Upate profile.
24297 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24299 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
24300 (tree_guess_outgoing_edge_probabilities): New.
24301 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
24302 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
24304 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24306 * ipa-split.c (split_function): Initialize return bb profile.
24308 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24310 * profile.c (compute_branch_probabilities): Also initialize
24311 EXIT_BLOCK profile.
24313 2017-06-06 Richard Biener <rguenther@suse.de>
24315 PR tree-optimization/80928
24316 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
24317 (vect_analyze_loop_operations): Properly guard analysis for
24319 (vect_transform_loop): Likewise.
24320 (vect_analyze_loop_2): Also reset SLP type on PHIs.
24321 (vect_model_induction_cost): Do not cost for pure SLP.
24322 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
24323 of induction in inner loop vectorization.
24324 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
24325 (vect_get_and_check_slp_defs): Handle vect_induction_def.
24326 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
24328 (vect_analyze_slp_cost_1): Cost induction.
24329 (vect_detect_hybrid_slp_stmts): Handle PHIs.
24330 (vect_get_slp_vect_defs): Likewise.
24331 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
24332 (vect_transform_stmt): Handle SLP reductions.
24333 * tree-vectorizer.h (vectorizable_induction): Adjust.
24335 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24337 * config/rs6000/rs6000.c (make_resolver_func): Update
24338 init_lowered_empty_function call.
24340 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
24342 * doc/invoke.texi: Document the -fprofile-abs-path option.
24343 * common.opt (fprofile-abs-path): New option.
24344 * gcov-io.h (gcov_write_filename): Declare.
24345 * gcov-io.c (gcov_write_filename): New function.
24346 * coverage.c (coverage_begin_function): Use gcov_write_filename.
24347 * profile.c (output_location): Likewise.
24349 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24351 * shring-wrap.c: Revert accidental commit.
24353 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
24355 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
24357 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24359 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
24361 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
24362 profile in callgraph edge.
24363 * profile-count.h (apply_probability): If THIS is 0, then result is 0
24364 (apply_scale): Likewise.
24365 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
24366 Also scale profile when inlining function with zero profile.
24367 (initialize_cfun): Update exit block profile even when it is zero.
24368 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
24369 when profile is read.
24371 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24373 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
24374 (CLONE_*): New constants to define the processors we can generate
24375 code for with the target_clone attribute.
24376 (rs6000_clone_map): New array to identify which clone processors
24377 the current program is running on.
24378 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
24379 target_clone attribute.
24380 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
24381 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
24382 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
24383 (cpu_expand_builtin): Add support for target_clone attribute.
24384 (rs6000_valid_attribute_p): Allow "default" attribute.
24385 (get_decl_name): New debug function to simplify printing the
24386 current function name in debugging statements.
24387 (rs6000_clone_priority): New functions to support the target_clone
24388 attribute, and be able to generate code to switch between ISA 2.05
24389 through ISA 3.0 (power6 through power9).
24390 (rs6000_compare_version_priority): Likewise.
24391 (rs6000_get_function_versions_dispatcher): Likewise.
24392 (make_resolver_func): Likewise.
24393 (add_condition_to_bb): Likewise.
24394 (dispatch_function_versions): Likewise.
24395 (rs6000_generate_version_dispatcher_body): Likewise.
24396 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
24397 (fusion_gpr_load_p): Fix a spacing issue.
24398 * doc/extend.texi (Common Function Attributes): Document that the
24399 PowerPC supports the target_clone attribute.
24401 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24403 * config/arm/arm.h: explain F symbol found in description of ARM
24404 register allocation in its legend.
24406 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24408 * config/mips/frame-header-opt.c: Include profile-count.h.
24409 * config/riscv/riscv.c: Include profile-count.h
24411 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24413 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
24415 (sm_set_flag_if_changed): Add bbs field.
24416 (execute_sm_if_changed_flag_set): Pass BBS.
24417 (execute_sm): Update.
24419 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24421 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
24424 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24426 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
24427 (peephole2): New peephole2 to emit the above.
24428 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
24430 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24432 * config/aarch64/aarch64.c (define_peephole2 above
24433 *sub_<shift>_<mode>): New peephole.
24435 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24437 * config/i386/i386.c (make_resolver_func): Update.
24438 * Makefile.in: Add profile-count.h and profile-count.o
24439 * auto-profile.c (afdo_indirect_call): Update to new API.
24440 (afdo_set_bb_count): Update.
24441 (afdo_propagate_edge): Update.
24442 (afdo_propagate_circuit): Update.
24443 (afdo_calculate_branch_prob): Update.
24444 (afdo_annotate_cfg): Update.
24445 * basic-block.h: Include profile-count.h
24446 (struct edge_def): Turn count to profile_count.
24447 (struct basic_block_def): Likewie.
24448 (REG_BR_PROB_BASE): Move to profile-count.h
24449 (RDIV): Move to profile-count.h
24450 * bb-reorder.c (max_entry_count): Turn to profile_count.
24451 (find_traces): Update.
24452 (rotate_loop):Update.
24453 (connect_traces):Update.
24454 (sanitize_hot_paths):Update.
24455 * bt-load.c (migrate_btr_defs): Update.
24456 * cfg.c (RDIV): Remove.
24457 (init_flow): Use alloc_block.
24458 (alloc_block): Uninitialize count.
24459 (unchecked_make_edge): Uninitialize count.
24460 (check_bb_profile): Update.
24461 (dump_edge_info): Update.
24462 (dump_bb_info): Update.
24463 (update_bb_profile_for_threading): Update.
24464 (scale_bbs_frequencies_int): Update.
24465 (scale_bbs_frequencies_gcov_type): Update.
24466 (scale_bbs_frequencies_profile_count): New.
24467 * cfg.h (update_bb_profile_for_threading): Update.
24468 (scale_bbs_frequencies_profile_count): Declare.
24469 * cfgbuild.c (compute_outgoing_frequencies): Update.
24470 (find_many_sub_basic_blocks): Update.
24471 * cfgcleanup.c (try_forward_edges): Update.
24472 (try_crossjump_to_edge): Update.
24473 * cfgexpand.c (expand_gimple_tailcall): Update.
24474 (construct_exit_block): Update.
24475 * cfghooks.c (verify_flow_info): Update.
24476 (dump_bb_for_graph): Update.
24477 (split_edge): Update.
24478 (make_forwarder_block): Update.
24479 (duplicate_block): Update.
24480 (account_profile_record): Update.
24481 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
24482 (get_estimated_loop_iterations): Update.
24483 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
24484 (single_likely_exit): Update.
24485 * cfgloopmanip.c (scale_loop_profile): Update.
24487 (set_zero_probability): Update.
24488 (lv_adjust_loop_entry_edge): Update.
24489 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
24490 (purge_dead_edges): Update.
24491 (rtl_account_profile_record): Update.
24492 * cgraph.c (cgraph_node::create): Uninitialize count.
24493 (symbol_table::create_edge): Uninitialize count.
24494 (cgraph_update_edges_for_call_stmt_node): Update.
24495 (cgraph_edge::dump_edge_flags): Update.
24496 (cgraph_node::dump): Update.
24497 (cgraph_edge::maybe_hot_p): Update.
24498 * cgraph.h: Include profile-count.h
24499 (create_clone), create_edge, create_indirect_edge): Update.
24500 (cgraph_node): Turn count to profile_count.
24501 (cgraph_edge0: Likewise.
24502 (make_speculative, clone): Update.
24503 (create_edge): Update.
24504 (init_lowered_empty_function): Update.
24505 * cgraphclones.c (cgraph_edge::clone): Update.
24506 (duplicate_thunk_for_node): Update.
24507 (cgraph_node::create_clone): Update.
24508 * cgraphunit.c (cgraph_node::analyze): Update.
24509 (cgraph_node::expand_thunk): Update.
24510 * final.c (dump_basic_block_info): Update.
24511 * gimple-streamer-in.c (input_bb): Update.
24512 * gimple-streamer-out.c (output_bb): Update.
24513 * graphite.c (print_global_statistics): Update.
24514 (print_graphite_scop_statistics): Update.
24515 * hsa-brig.c: Include basic-block.h.
24516 * hsa-dump.c: Include basic-block.h.
24517 * hsa-gen.c (T sum_slice): Update.
24518 (convert_switch_statements):Update.
24519 * hsa-regalloc.c: Include basic-block.h.
24520 * ipa-chkp.c (chkp_produce_thunks): Update.
24521 * ipa-cp.c (struct caller_statistics): Update.
24522 (init_caller_stats): Update.
24523 (gather_caller_stats): Update.
24524 (ipcp_cloning_candidate_p): Update.
24525 (good_cloning_opportunity_p): Update.
24526 (get_info_about_necessary_edges): Update.
24527 (dump_profile_updates): Update.
24528 (update_profiling_info): Update.
24529 (update_specialized_profile): Update.
24530 (perhaps_add_new_callers): Update.
24531 (decide_about_value): Update.
24532 (ipa_cp_c_finalize): Update.
24533 * ipa-devirt.c (struct odr_type_warn_count): Update.
24534 (struct decl_warn_count): Update.
24535 (struct final_warning_record): Update.
24536 (possible_polymorphic_call_targets): Update.
24537 (ipa_devirt): Update.
24538 * ipa-fnsummary.c (redirect_to_unreachable): Update.
24539 * ipa-icf.c (sem_function::merge): Update.
24540 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
24541 * ipa-inline.c (compute_uninlined_call_time): Update.
24542 (compute_inlined_call_time): Update.
24543 (want_inline_small_function_p): Update.
24544 (want_inline_self_recursive_call_p): Update.
24545 (edge_badness): Update.
24546 (lookup_recursive_calls): Update.
24547 (recursive_inlining): Update.
24548 (inline_small_functions): Update.
24549 (dump_overall_stats): Update.
24550 (dump_inline_stats): Update.
24551 * ipa-profile.c (ipa_profile_generate_summary): Update.
24552 (ipa_propagate_frequency): Update.
24553 (ipa_profile): Update.
24554 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24555 * ipa-utils.c (ipa_merge_profiles): Update.
24556 * loop-doloop.c (doloop_modify): Update.
24557 * loop-unroll.c (report_unroll): Update.
24558 (unroll_loop_runtime_iterations): Update.
24559 * lto-cgraph.c (lto_output_edge): Update.
24560 (lto_output_node): Update.
24561 (input_node): Update.
24562 (input_edge): Update.
24563 (merge_profile_summaries): Update.
24564 * lto-streamer-in.c (input_cfg): Update.
24565 * lto-streamer-out.c (output_cfg): Update.
24566 * mcf.c (create_fixup_graph): Update.
24567 (adjust_cfg_counts): Update.
24568 (sum_edge_counts): Update.
24569 * modulo-sched.c (sms_schedule): Update.
24570 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
24571 * predict.c (maybe_hot_count_p): Update.
24572 (probably_never_executed): Update.
24573 (dump_prediction): Update.
24574 (combine_predictions_for_bb): Update.
24575 (propagate_freq): Update.
24576 (handle_missing_profiles): Update.
24577 (counts_to_freqs): Update.
24578 (rebuild_frequencies): Update.
24579 (force_edge_cold): Update.
24580 * predict.h: Include profile-count.h
24581 (maybe_hot_count_p, counts_to_freqs): UPdate.
24582 * print-rtl-function.c: Do not include cfg.h
24583 * print-rtl.c: Include basic-block.h
24584 * profile-count.c: New file.
24585 * profile-count.h: New file.
24586 * profile.c (is_edge_inconsistent): Update.
24587 (correct_negative_edge_counts): Update.
24588 (is_inconsistent): Update.
24589 (set_bb_counts): Update.
24590 (read_profile_edge_counts): Update.
24591 (compute_frequency_overlap): Update.
24592 (compute_branch_probabilities): Update; Initialize and deinitialize
24594 (branch_prob): Update.
24595 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
24596 (edge_gcov_count): New.
24597 (bb_gcov_count): New.
24598 * shrink-wrap.c (try_shrink_wrapping): Update.
24599 * tracer.c (better_p): Update.
24600 * trans-mem.c (expand_transaction): Update.
24601 (ipa_tm_insert_irr_call): Update.
24602 (ipa_tm_insert_gettmclone_call): Update.
24603 * tree-call-cdce.c: Update.
24604 * tree-cfg.c (gimple_duplicate_sese_region): Update.
24605 (gimple_duplicate_sese_tail): Update.
24606 (gimple_account_profile_record): Update.
24607 (execute_fixup_cfg): Update.
24608 * tree-inline.c (copy_bb): Update.
24609 (copy_edges_for_bb): Update.
24610 (initialize_cfun): Update.
24611 (freqs_to_counts): Update.
24612 (copy_cfg_body): Update.
24613 (expand_call_inline): Update.
24614 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
24615 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
24616 (try_unroll_loop_completely): Update.
24617 (try_peel_loop): Update.
24618 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24619 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
24620 * tree-ssa-loop-split.c (connect_loops): Update.
24621 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
24622 * tree-ssa-reassoc.c (branch_fixup): Update.
24623 * tree-ssa-tail-merge.c (replace_block_by): Update.
24624 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
24625 (compute_path_counts): Update.
24626 (update_profile): Update.
24627 (recompute_probabilities): Update.
24628 (update_joiner_offpath_counts): Update.
24629 (estimated_freqs_path): Update.
24630 (freqs_to_counts_path): Update.
24631 (clear_counts_path): Update.
24632 (ssa_fix_duplicate_block_edges): Update.
24633 (duplicate_thread_path): Update.
24634 * tree-switch-conversion.c (case_bit_test_cmp): Update.
24635 (struct switch_conv_info): Update.
24636 * tree-tailcall.c (decrease_profile): Update.
24637 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
24638 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24639 * value-prof.c (check_counter): Update.
24640 (gimple_divmod_fixed_value): Update.
24641 (gimple_mod_pow2): Update.
24642 (gimple_mod_subtract): Update.
24643 (gimple_ic_transform): Update.
24644 (gimple_stringop_fixed_value): Update.
24645 * value-prof.h (gimple_ic): Update.
24647 2017-06-02 Carl Love <cel@us.ibm.com>
24649 * config/rs6000/rs6000-c: Add support for built-in functions
24650 vector double vec_doublee (vector signed int);
24651 vector double vec_doublee (vector unsigned int);
24652 vector double vec_doublee (vector float);
24653 vector double vec_doubleh (vector signed int);
24654 vector double vec_doubleh (vector unsigned int);
24655 vector double vec_doubleh (vector float);
24656 vector double vec_doublel (vector signed int);
24657 vector double vec_doublel (vector unsigned int);
24658 vector double vec_doublel (vector float);
24659 vector double vec_doubleo (vector signed int);
24660 vector double vec_doubleo (vector unsigned int);
24661 vector double vec_doubleo (vector float);.
24662 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
24663 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
24665 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
24666 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
24667 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
24669 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
24670 vec_doublel, vec_doubleh.
24671 * doc/extend.texi: Update the built-in documentation file for the
24672 new built-in functions.
24674 2017-06-02 David Malcolm <dmalcolm@redhat.com>
24677 * ipa-inline-analysis.c (free_growth_caches): Set
24678 edge_removal_hook_holder to NULL after removing it.
24680 2017-06-02 Sudakshina Das <sudi.das@arm.com>
24682 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
24683 comparision with zero.
24685 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
24686 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
24687 for early expansion of vec_min and vec_max builtins.
24688 (builtin_function_type): Add min/max unsigned variants to those
24689 identified as having unsigned arguments.
24691 2017-06-02 Olivier Hainque <hainque@adacore.com>
24693 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
24695 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24697 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
24698 Use VALL_F16 iterator rather than VALL.
24700 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24702 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
24703 Emit CBNZ inside loop when doing a strong exchange and comparing
24704 against zero. Generate the CC flags after the loop.
24706 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
24708 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
24709 (dl_section_ref): New.
24710 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
24711 On AIX, append an expression to subtract the size of the
24712 section length to dl_section_ref.
24714 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
24716 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
24717 for early expansion of vector absolute builtins.
24719 2017-06-02 Richard Biener <rguenther@suse.de>
24721 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
24722 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
24724 2017-06-02 Richard Biener <rguenther@suse.de>
24726 PR tree-optimization/80948
24727 * tree-tailcall.c (find_tail_calls): Track stmts to move in
24728 stmt order as well.
24730 2017-06-02 Richard Biener <rguenther@suse.de>
24732 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
24734 * tree-vect-stmts.c (process_use): Do not mark backedge defs
24735 for inductions as relevant.
24737 2017-06-02 Richard Biener <rguenther@suse.de>
24739 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
24740 (vectorizable_induction): ... this. Remove dead code.
24742 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
24744 * builtins. (expand_builtin_alloca): Remove second parameter and
24745 infer its value from the first parameter instead.
24746 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
24748 2017-06-02 Jakub Jelinek <jakub@redhat.com>
24750 PR rtl-optimization/80903
24751 * loop-doloop.c (add_test): Unshare sequence.
24753 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
24755 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
24757 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
24759 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
24761 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
24762 xlogue_layout::get_instance, logue_layout::xlogue_layout,
24763 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
24764 (xlogue_layout::get_stub_rtx): Make static.
24765 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
24766 (xlogue_layout::compute_stub_managed_regs): Rename to...
24767 (xlogue_layout::count_stub_managed_regs): ...this.
24768 (xlogue_layout::is_stub_managed_reg): New function.
24769 (xlogue_layout::m_stub_names): Rename to...
24770 (xlogue_layout::s_stub_names): ...this, make static.
24771 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
24772 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
24773 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
24774 xlogue_layout::s_stub_names): Instantiate statics.
24775 (stub_managed_regs): Remove.
24776 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
24777 (disable_call_ms2sysv_xlogues): Rename to...
24778 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
24779 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
24781 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
24782 change after reload_completed.
24783 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
24785 (ix86_expand_prologue): Likewise.
24786 (ix86_expand_epilogue): Likewise.
24787 (ix86_expand_split_stack_prologue): Likewise.
24788 (ix86_compute_frame_layout): Remove frame parameter ...
24789 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
24790 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
24792 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
24793 (ix86_frame): Move from here ...
24794 * config/i386/i386.h (ix86_frame): ... to here.
24795 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
24796 complete ix86_frame data structure instead. Remove some_ld_name.
24798 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
24800 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
24801 symbols that hold a DECL_VALUE_EXPR.
24803 2017-06-01 Martin Jambor <mjambor@suse.cz>
24805 PR tree-optimization/80898
24806 * tree-sra.c (process_subtree_disqualification): Removed.
24807 (disqualify_candidate): Do not acll
24808 process_subtree_disqualification.
24809 (subtree_mark_written_and_enqueue): New function.
24810 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
24811 RHS has been disqualified and re-queue LHS if necessary. Apart
24812 from that, ignore disqualified RHS.
24814 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24816 * config/s390/s390.c (s390_emit_epilogue): Disable early return
24817 address fetch for z10 or later.
24819 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24821 * config/arc/arc.md (tst_movb): Add guard when splitting.
24823 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24825 * config/arc/arc.c (arc_can_eliminate): Test against
24826 arc_frame_pointer_needed.
24828 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24830 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
24831 to prevent store reordering.
24832 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
24833 (type): Add block type.
24834 (stack_tie): Define special instruction to be used in
24837 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24839 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
24840 constraint. It is not valid for the pattern.
24841 (noncommutative_binary_comparison): Likewise.
24843 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24845 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
24848 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24850 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
24851 be used by the reg-alloc.
24853 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24855 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
24856 reg-alloc when having mul64 or mul32x16 instructions.
24857 (mulsidi3): Likewise.
24858 (umulsidi3): Likewise.
24859 (mulsi32x16): New pattern.
24860 (mulsi64): Likewise.
24861 (mulsidi64): Likewise.
24862 (umulsidi64): Likewise.
24863 (MUL32x16_REG): Define.
24864 (mul64_600): Use MUL32x16_REG.
24865 (mac64_600): Likewise.
24866 (umul64_600): Likewise.
24867 (umac64_600): Likewise.
24869 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
24871 * config/arc/arc.md (mulsi3_700): Make it commutative.
24873 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
24875 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
24877 (*sign_extendsidi2_insn): Likewise for movstosw.
24879 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
24881 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
24882 the type of the input discriminant value. Convert the
24883 discriminant value of signedness vary.
24885 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
24887 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
24888 Add to -Wall section.
24890 2017-06-01 Richard Biener <rguenther@suse.de>
24892 PR middle-end/66313
24893 * fold-const.c (fold_plusminus_mult_expr): If the factored
24894 factor may be zero use a wrapping type for the inner operation.
24895 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
24896 and handle moved defs.
24897 (process_assignment): Properly guard the unary op case. Return a
24898 tri-state indicating that moving the stmt before the call may allow
24899 to continue. Pass through to_move.
24900 (find_tail_calls): Handle moving unrelated defs before
24903 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
24906 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
24907 splitter result in the canonical way.
24909 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
24911 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
24912 also for 32bit target. Update insn attributes.
24913 (zero-extendsidi2 splitter): Allow all registers for operand 1.
24915 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
24917 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
24918 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
24919 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
24920 (_mm_maskz_min_ss): New intrinsics.
24922 2017-05-31 Martin Liska <mliska@suse.cz>
24924 * tree-vect-loop.c (vect_create_epilog_for_reduction):
24925 Change comment style to one we normally use.
24926 (vectorizable_reduction): Likewise.
24927 (vectorizable_induction): Likewise.
24928 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
24929 (vectorizable_call): Likewise.
24930 (vectorizable_simd_clone_call): Likewise.
24931 (vectorizable_conversion): Likewise.
24932 (vectorizable_assignment): Likewise.
24933 (vectorizable_shift): Likewise.
24934 (vectorizable_operation): Likewise.
24935 (vectorizable_store): Likewise.
24936 (vectorizable_load): Likewise.
24937 * tree-vectorizer.h: Likewise.
24939 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
24941 * passes.c (emergency_dump_function): New.
24942 * tree-pass.h (emergency_dump_function): Declare.
24943 * plugin.c (plugins_internal_error_function): Remove.
24944 * plugin.h (plugins_internal_error_function): Remove declaration.
24945 * toplev.c (internal_error_function): New static function. Use it...
24946 (general_init): ...here.
24948 2017-05-31 Graham Markall <graham.markall@embecosm.com>
24950 * config/arc/arc.c (arc_print_operand): Handle constant operands.
24951 (arc_rtx_costs): Add costs for new patterns.
24952 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
24953 * config/arc/predicates.md: Add _1_2_3_operand predicate.
24955 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
24957 * tree-ssa-strlen.c (get_next_strinfo): New function.
24958 (get_stridx_plus_constant): Use it.
24959 (zero_length_string): Likewise.
24960 (adjust_related_strinfos): Likewise.
24961 (adjust_last_stmt): Likewise.
24963 2017-05-31 Richard Biener <rguenther@suse.de>
24966 * config/i386/i386.c (ix86_expand_builtin): Remove assert
24967 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
24969 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
24971 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
24972 loop_vinfo argument and use of dependence distance vectors.
24973 Check instead whether the two references differ only in their
24974 initial value and assume that they have the same alignment if the
24975 difference is a multiple of the vector alignment.
24976 (vect_analyze_data_refs_alignment): Update call accordingly.
24978 2017-05-31 Martin Liska <mliska@suse.cz>
24981 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
24983 2017-05-31 Bin Cheng <bin.cheng@arm.com>
24985 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
24986 (create_intersect_range_checks): Move from ...
24987 * tree-data-ref.c (create_intersect_range_checks_index)
24988 (create_intersect_range_checks): ... to here.
24989 (create_runtime_alias_checks): New function factored from ...
24990 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
24991 here. Call above function.
24992 * tree-data-ref.h (create_runtime_alias_checks): New function.
24994 2017-05-31 Bin Cheng <bin.cheng@arm.com>
24996 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
24997 segment length for dr_b and compute it in wide_int.
24999 2017-05-31 Richard Biener <rguenther@suse.de>
25001 PR tree-optimization/80906
25002 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
25003 and pass through iv_map.
25004 (copy_bb_and_scalar_dependences): Adjust.
25005 (translate_pending_phi_nodes): Likewise.
25006 (copy_loop_close_phi_args): Handle code-generating IVs instead
25009 2017-05-30 David Malcolm <dmalcolm@redhat.com>
25011 * diagnostic-color.c (color_dict): Add "type-diff".
25012 (parse_gcc_colors): Update comment.
25013 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
25014 -fdiagnostics-show-template-tree and -fno-elide-type.
25015 (GCC_COLORS): Add type-diff to example.
25017 (-fdiagnostics-show-template-tree): New.
25018 (-fno-elide-type): New.
25019 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
25020 the pp_format_decoder callback. Call any m_format_postprocessor's
25022 (pretty_printer::pretty_printer): Initialize
25023 m_format_postprocessor.
25024 (pretty_printer::~pretty_printer): Delete any
25025 m_format_postprocessor.
25026 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
25027 (class format_postprocessor): New class.
25028 (struct pretty_printer::format_decoder): Document the new parameters.
25029 (struct pretty_printer::m_format_postprocessor): New field.
25030 * tree-diagnostic.c (default_tree_printer): Update for new
25031 bool and const char ** params.
25032 * tree-diagnostic.h (default_tree_printer): Likewise.
25034 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
25036 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
25037 (lwa_operand): Delete rs6000_gen_cell_microcode test.
25038 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
25039 rs6000_gen_cell_microcode code.
25040 (rs6000_final_prescan_insn): Delete.
25041 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
25042 "warn-cell-microcode" entries.
25043 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
25044 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
25045 throughout. Change cc_reg_not_micro_cr0_operand to
25046 cc_reg_not_cr0_operand throughout.
25047 (*extendhi<mode>2_noload): Delete.
25048 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
25049 (mwarn-cell-microcode): Delete.
25050 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
25051 -mgen-cell-microcode and -mwarn-cell-microcode.
25053 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
25056 * config/i386/constraints.md (Yd): New constraint.
25058 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
25059 and (?Yd, r) alternatives. Update insn attributes.
25060 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
25061 and (?*Yd, r) alternatives. Update insn attributes.
25062 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
25064 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
25066 * gimplify.c (gimplify_modify_expr): Don't create a
25067 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
25070 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
25072 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
25074 2017-05-30 Richard Biener <rguenther@suse.de>
25076 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
25077 and reduc_def fields.
25078 (STMT_VINFO_REDUC_TYPE): New define.
25079 (STMT_VINFO_REDUC_DEF): Likewise.
25080 (vect_force_simple_reduction): Adjust prototype.
25081 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
25082 (vect_is_simple_reduction): Remove check_reduction argument.
25083 (vect_force_simple_reduction): Adjust and set
25084 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
25085 (vectorizable_reduction): Do not re-do reduction analysis
25086 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
25087 * tree-parloops.c (gather_scalar_reductions): Adjust.
25089 2017-05-30 Richard Biener <rguenther@suse.de>
25091 PR middle-end/80901
25092 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
25095 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
25097 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
25098 Introduce unknown_misalignment parameter and remove vf.
25099 (vect_peeling_hash_get_lowest_cost):
25100 Pass unknown_misalignment parameter.
25101 (vect_enhance_data_refs_alignment):
25102 Fix unsupportable data ref treatment.
25104 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25106 * tree-vect-data-refs.c (vect_get_data_access_cost):
25107 Workaround for SLP handling.
25108 (vect_enhance_data_refs_alignment):
25109 Compute costs for doing no peeling at all, compare to the best
25110 peeling costs so far and avoid peeling if cheaper.
25112 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25114 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
25115 Return peeling info and set costs to zero for unlimited cost
25117 (vect_enhance_data_refs_alignment): Also inspect all datarefs
25118 with unknown misalignment. Compute and costs for unknown
25119 misalignment, compare them to the costs for known misalignment
25120 and choose the cheapest for peeling.
25122 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25124 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
25125 (vect_get_peeling_costs_all_drs): Create function.
25126 (vect_peeling_hash_get_lowest_cost):
25127 Use vect_get_peeling_costs_all_drs.
25128 (vect_peeling_supportable): Create function.
25129 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
25131 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25133 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
25134 DR_HAS_NEGATIVE_STEP.
25135 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
25136 (vect_enhance_data_refs_alignment): Use.
25137 (vect_duplicate_ssa_name_ptr_info): Use.
25138 * tree-vectorizer.h (dr_misalignment): Use.
25139 (known_alignment_for_access_p): Use.
25141 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
25144 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
25146 (has_section_name): New function.
25148 2017-05-30 Martin Liska <mliska@suse.cz>
25151 * auto-profile.c (get_function_decl_from_block): Fix
25154 2017-05-30 Richard Biener <rguenther@suse.de>
25156 PR middle-end/80876
25157 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
25159 2017-05-30 Martin Liska <mliska@suse.cz>
25161 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
25162 * dumpfile.h (struct dump_file_info): Remove ctors.
25164 2017-05-30 Martin Liska <mliska@suse.cz>
25166 * predict.def: Fix GNU coding style.
25168 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
25170 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25171 Mark 'to' argument with ATTRIBUTE_UNUSED.
25173 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
25175 * config/xtensa/xtensa.c (xtensa_emit_call): Use
25176 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
25177 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
25180 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
25182 * doc/install.texi (Options specification): Restore entry of
25183 --enable-sjlj-exceptions.
25185 2017-05-27 Michael Eager <eager@eagercon.com>
25188 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
25190 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
25192 * config/microblaze/microblaze.h
25193 (FIXED_REGISTERS): Update in macro.
25194 (CALL_USED_REGISTERS): Update in macro.
25196 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
25198 * doc/install.texi: Add links to macOS binary distributions.
25200 2017-05-27 Jakub Jelinek <jakub@redhat.com>
25204 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25206 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
25208 2017-05-26 Martin Liska <mliska@suse.cz>
25210 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
25212 2017-05-26 Martin Liska <mliska@suse.cz>
25214 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
25215 always leading ';; '.
25216 (dump_bb_info): Likewise.
25217 (brief_dump_cfg): Likewise.
25218 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
25219 * dumpfile.c: Remove usage of TDF_VERBOSE.
25220 * dumpfile.h (enum dump_kind): Likewise.
25221 (dump_gimple_bb_header): Do not use TDF_COMMENT.
25222 * print-tree.c (debug_verbose): Remove.
25223 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
25224 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
25225 * tree-diagnostic.c (default_tree_printer): Replace
25226 TDF_DIAGNOSTIC with TDF_SLIM.
25228 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25230 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
25231 in parameter loop, rather than loop_vinfo.
25232 (create_intersect_range_checks): Ditto.
25233 (vect_create_cond_for_alias_checks): Update call to above functions.
25235 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25237 PR tree-optimization/80815
25238 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
25239 for merging runtime alias checks. Handle negative DR_STEPs.
25241 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25243 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
25245 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
25246 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
25247 out code pruning runtime alias checks.
25248 * tree-data-ref.c (prune_runtime_alias_test_list): New function
25249 factored out from above.
25250 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
25252 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
25254 (prune_runtime_alias_test_list): New decalaration.
25256 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25258 * tree-vect-data-refs.c (compare_tree): Rename and move ...
25259 * tree-data-ref.c (data_ref_compare_tree): ... to here.
25260 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
25261 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
25262 (operator==, comp_dr_with_seg_len_pair): Ditto.
25263 (vect_prune_runtime_alias_test_list): Ditto.
25265 2017-05-26 Martin Liska <mliska@suse.cz>
25268 * params.def: Bound partial-inlining-entry-probability param.
25270 2017-05-26 Marek Polacek <polacek@redhat.com>
25273 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
25276 2017-05-26 Richard Biener <rguenther@suse.de>
25278 PR tree-optimization/80842
25279 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
25282 2017-05-26 Richard Biener <rguenther@suse.de>
25284 PR tree-optimization/80844
25285 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
25287 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
25289 * doc/md.texi (Machine Constraints): Update x86 family
25290 machine constraints section to match 'config/i386/constraints.md'.
25292 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
25294 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
25296 2017-05-25 Nathan Sidwell <nathan@acm.org>
25298 * doc/invoke.texi (--enable-languages): Update documentation.
25300 2017-05-25 Martin Liska <mliska@suse.cz>
25302 * dumpfile.c: Add TDF_FOLDING.
25303 * dumpfile.h (enum dump_kind): Likewise.
25304 * genmatch.c (dt_simplify::gen_1): Use it.
25306 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25308 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
25310 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25312 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
25313 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
25315 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25317 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
25318 * match.pd (X == C): Rewrite it here.
25319 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
25320 with_certain_nonzero_bits2): New predicates.
25321 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
25323 2017-05-24 Nathan Sidwell <nathan@acm.org>
25325 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
25328 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
25331 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
25333 * config/powerpcspe: New port. Files are copied from the rs6000
25334 port, with "rs6000" in filenames replaced by "powerpcspe".
25336 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
25338 PR rtl-optimization/80754
25339 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
25341 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
25343 * config/sparc/sparc.md (length): Return the correct value for -mflat
25344 sibcalls to match output_sibcall.
25346 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
25350 * config/rs6000/rs6000.c (struct machine_function): Add new field
25352 (rs6000_get_separate_components): Init that field, use it.
25353 (rs6000_components_for_bb): Use the field.
25355 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
25357 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
25359 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
25361 PR middle-end/80823
25362 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
25364 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25367 * config/s390/s390.c (s390_check_qrst_address): Check incoming
25368 address against address_operand predicate.
25369 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
25371 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
25373 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
25374 parameters passed indirectly.
25376 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
25378 * config/i386/i386.md (*movdi_internal): Remove SSE4
25379 alternative 18 (?r, *v). Update insn attributes.
25380 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
25381 Update insn attributes.
25382 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
25383 Update insn attributes.
25384 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
25385 alternative 1 (r, v). Remove isa attribute.
25386 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
25387 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
25388 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
25390 2017-05-23 Tom de Vries <tom@codesourcery.com>
25392 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
25395 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25397 * cgraphunit.c (symbol_table::process_new_functions): Update.
25398 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
25399 (inline_generate_summary): Rename to ...
25400 (ipa_fn_summary_generate): ... this one.
25401 (inline_read_summary): Rename to ...
25402 (ipa_fn_summary_read): ... this one.
25403 (inline_write_summary): Rename to ...
25404 (ipa_fn_summary_write): ... this one.
25405 (inline_free_summary): Rename to ...
25406 (ipa_free_fn_summary): ... this one.
25407 (pass_data_local_fn_summary, pass_local_fn_summary,
25408 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
25409 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
25410 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
25411 make_pass_ipa_fn_summary): New.
25412 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
25413 inline_write_summary, inline_free_summary): Remove.
25414 (ipa_free_fn_summary) : New.
25415 * ipa-inline.c (ipa_inline): Update.
25416 (pass_ipa_inline): Do not generate summaries.
25417 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
25419 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
25420 and add pass_ipa_fn_summary.
25421 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
25423 (make_pass_inline_parameters): Remove.
25425 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
25427 * omp-low.c (struct omp_context): Remove "default_kind" member.
25430 * omp-offload.c (execute_oacc_device_lower): Remove the
25431 parallelism dimensions function attributes for unparallelized
25432 OpenACC kernels constructs.
25434 2017-05-23 Martin Liska <mliska@suse.cz>
25436 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
25438 (cgraph_edge::make_speculative): Likewise.
25439 (cgraph_edge::resolve_speculation): Likewise.
25440 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
25441 (cgraph_node::dump): Likewise.
25442 * cgraph.h: Likewise.
25443 * cgraphunit.c (analyze_functions): Likewise.
25444 (symbol_table::compile): Likewise.
25445 * ipa-cp.c (print_all_lattices): Likewise.
25446 (determine_versionability): Likewise.
25447 (initialize_node_lattices): Likewise.
25448 (ipcp_verify_propagated_values): Likewise.
25449 (estimate_local_effects): Likewise.
25450 (update_profiling_info): Likewise.
25451 (create_specialized_node): Likewise.
25452 (perhaps_add_new_callers): Likewise.
25453 (decide_about_value): Likewise.
25454 (decide_whether_version_node): Likewise.
25455 (identify_dead_nodes): Likewise.
25456 (ipcp_store_bits_results): Likewise.
25457 * ipa-devirt.c (dump_targets): Likewise.
25458 (ipa_devirt): Likewise.
25459 * ipa-icf.c (sem_item::dump): Likewise.
25460 (sem_function::equals): Likewise.
25461 (sem_variable::equals): Likewise.
25462 (sem_item_optimizer::read_section): Likewise.
25463 (sem_item_optimizer::execute): Likewise.
25464 (congruence_class::dump): Likewise.
25465 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
25466 (dump_inline_summary): Likewise.
25467 (estimate_node_size_and_time): Likewise.
25468 (inline_analyze_function): Likewise.
25469 * ipa-inline-transform.c (inline_call): Likewise.
25470 * ipa-inline.c (report_inline_failed_reason): Likewise.
25471 (want_early_inline_function_p): Likewise.
25472 (edge_badness): Likewise.
25473 (update_edge_key): Likewise.
25474 (inline_small_functions): Likewise.
25475 * ipa-profile.c (ipa_profile): Likewise.
25476 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
25477 (ipa_make_edge_direct_to_target): Likewise.
25478 (remove_described_reference): Likewise.
25479 (ipa_impossible_devirt_target): Likewise.
25480 (propagate_controlled_uses): Likewise.
25481 (ipa_print_node_params): Likewise.
25482 (ipcp_transform_function): Likewise.
25483 * ipa-pure-const.c (pure_const_read_summary): Likewise.
25484 (propagate_pure_const): Likewise.
25485 * ipa-reference.c (generate_summary): Likewise.
25486 (read_write_all_from_decl): Likewise.
25487 (propagate): Likewise.
25488 (ipa_reference_read_optimization_summary): Likewise.
25489 * ipa-utils.c (ipa_merge_profiles): Likewise.
25490 * ipa.c (walk_polymorphic_call_targets): Likewise.
25491 (symbol_table::remove_unreachable_nodes): Likewise.
25492 (ipa_single_use): Likewise.
25493 * passes.c (execute_todo): Likewise.
25494 * predict.c (drop_profile): Likewise.
25495 * symtab.c (symtab_node::get_dump_name): New function.
25496 (symtab_node::dump_name): Likewise.
25497 (symtab_node::dump_asm_name): Likewise.
25498 (symtab_node::dump_references): Likewise.
25499 (symtab_node::dump_referring): Likewise.
25500 (symtab_node::dump_base): Likewise.
25501 (symtab_node::debug_symtab): Likewise.
25502 * tree-sra.c (convert_callers_for_node): Likewise.
25503 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
25504 * value-prof.c (init_node_map): Likewise.
25506 2017-05-23 Martin Liska <mliska@suse.cz>
25508 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
25509 and symtab_node::debug_symtab to symbol_table::debug.
25510 * cgraphunit.c (analyze_functions): Use the renamed function.
25511 (symbol_table::compile): Likewise.
25512 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
25513 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
25514 * passes.c (execute_todo): Likewise.
25515 * symtab.c (symbol_table::dump): New function.
25516 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
25518 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25520 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
25521 that nonconst implies exec.
25523 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25525 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
25526 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
25527 (inline_edge_summary_vec): Turn into ...
25528 (ipa_call_summaries): ... this one.
25529 (redirect_to_unreachable, edge_set_predicate,
25530 evaluate_properties_for_edge, inline_summary_alloc,
25531 reset_ipa_call_summary, reset_inline_summary,
25532 inline_summary_t::duplicate): Update.
25533 (inline_edge_duplication_hook): Turn to ...
25534 (ipa_call_summary_t::duplicate): ... this one.
25535 (inline_edge_removal_hook): Turn to ...
25536 (ipa_call_summary_t::remove): ... this one.
25537 (dump_inline_edge_summary): Turn to ...
25538 (dump_ipa_call_summary): ... this one.
25539 (estimate_function_body_sizes): Update.
25540 (inline_update_callee_summaries): Update.
25541 (remap_edge_change_prob): Update.
25542 (remap_edge_summaries): Update.
25543 (inline_merge_summary): Update.
25544 (do_estimate_edge_time): Update.
25545 (inline_generate_summary): Update.
25546 (inline_read_section): Update.
25547 (inline_read_summary): Update.
25548 (inline_free_summary): Update.
25549 * ipa-inline.c (can_inline_edge_p): Update.
25550 (compute_inlined_call_time): Update.
25551 (want_inline_small_function_p): Update.
25552 (edge_badness): Update.
25553 (early_inliner): Update.
25554 * ipa-inline.h (inline_edge_summary): Turn to ...
25555 (ipa_call_summary): ... this one.
25556 (ipa_call_summary_t): New class.
25557 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
25558 (ipa_call_summaries): New.
25559 (inline_edge_summary): Remove.
25560 (estimate_edge_growth): Update.
25561 * ipa-profile.c (ipa_propagate_frequency_1): Update.
25562 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
25563 * ipa-split.c (execute_split_functions): Update.
25564 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
25566 2017-05-23 Tom de Vries <tom@codesourcery.com>
25568 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
25569 attributes): Document rdrand effective target.
25571 2017-05-23 Tom de Vries <tom@codesourcery.com>
25573 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
25574 attributes): Sort alphabetically.
25576 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
25578 * config/avr/genmultilib.awk: Use gsub instead of gensub.
25580 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
25583 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
25584 V2DF/V2DI splat into two separate patterns, one that handles
25585 registers, and the other that only handles memory. Drop support
25586 for splatting from a GPR on ISA 2.07 and then splitting the
25587 splat into direct move and splat.
25588 (vsx_splat_<mode>_reg): Likewise.
25589 (vsx_splat_<mode>_mem): Likewise.
25591 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
25593 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
25595 2017-05-22 Jakub Jelinek <jakub@redhat.com>
25597 PR middle-end/80809
25598 * omp-low.c (finish_taskreg_remap): New function.
25599 (finish_taskreg_scan): If unit size of ctx->record_type
25600 is non-constant, unshare the size expression and replace
25601 decls in it with possible outer var refs.
25603 PR middle-end/80809
25604 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
25605 GOVD_SHARED rather than GOVD_PRIVATE with it.
25606 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
25607 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
25609 PR middle-end/80853
25610 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
25611 as last argument to build_outer_var_ref for pointer bases of array
25612 section reductions.
25614 2017-05-19 Martin Sebor <msebor@redhat.com>
25616 * print-tree.c (print_node): Print DECL_READ_P flag.
25618 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25620 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
25621 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
25622 * cgraph.c: Likewise.
25623 * cgraphunit.c: Likewise.
25624 * gengtype.c: Likewise.
25625 * ipa-cp.c: Likewise.
25626 * ipa-devirt.c: Likewise.
25627 * ipa-icf.c: Likewise.
25628 * ipa-predicate.c: Likewise.
25629 * ipa-profile.c: Likewise.
25630 * ipa-prop.c: Likewise.
25631 * ipa-split.c: Likewise.
25633 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
25634 edge_predicate_pool, dump_inline_hints,
25635 inline_summary::account_size_time, redirect_to_unreachable,
25636 edge_set_predicate, set_hint_predicate,
25637 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
25638 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
25639 inline_summary_t::remove, remap_hint_predicate_after_duplication,
25640 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
25641 ipa_call_summary_t::remove, initialize_growth_caches,
25642 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
25643 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
25644 mark_modified, unmodified_parm_1, unmodified_parm,
25645 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
25646 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
25647 compute_bb_predicates, will_be_nonconstant_expr_predicate,
25648 will_be_nonconstant_predicate, record_modified_bb_info,
25649 get_minimal_bb, record_modified, param_change_prob,
25650 phi_result_unknown_predicate, predicate_for_phi_result,
25651 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
25652 estimate_function_body_sizes, compute_inline_parameters,
25653 compute_inline_parameters_for_curren, pass_data_inline_parameters,
25654 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
25655 inline_update_callee_summaries, remap_edge_change_prob,
25656 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
25657 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
25658 inline_analyze_function, inline_summary_t::insert,
25659 inline_generate_summary, read_ipa_call_summary, inline_read_section,
25660 inline_read_summary, write_ipa_call_summary, inline_write_summary,
25661 inline_free_summary): Move to ipa-fnsummary.h
25662 (predicate_t): Remove.
25663 * ipa-fnsummary.c: New file.
25664 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
25665 (enum inline_hints_vals, inline_hints, agg_position_info,
25666 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
25667 inline_summaries, ipa_call_summary, ipa_call_summary_t,
25668 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
25669 dump_inline_summary, dump_inline_hints, inline_generate_summary,
25670 inline_read_summary, inline_write_summary, inline_free_summary,
25671 inline_analyze_function, initialize_inline_failed,
25672 inline_merge_summary, inline_update_overall_summary,
25673 compute_inline_parameters): Move to ipa-fnsummary.h
25674 * ipa-fnsummary.h: New file.
25675 * ipa-inline-transform.h: Include ipa-inline.h.
25676 * ipa-inline.c: LIkewise.
25678 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25680 * ipa-inline.c (edge_badness): Use inlined_time instead of
25681 inline_summaries->get.
25683 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25685 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
25687 2017-05-22 Nathan Sidwell <nathan@acm.org>
25689 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
25690 (fdump-lang): Document 'raw' option.
25691 * dumpfile.h (TDI_tu): Delete.
25692 * dumpfile.c (dump_files): Remove translation-unit.
25693 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
25695 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
25697 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
25698 command option from $(AWK) call.
25699 * config/avr/genmultilib.awk: Simplify and rewrite so that it
25700 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
25701 [FORMAT]: Remove handling of variable.
25702 * config/avr/t-multilib: Regenerate.
25704 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25706 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
25708 (dump_inline_summary): Do not print self_time.
25709 (estimate_function_body_sizes): Do not set self_time.
25710 (compute_inline_parameters): Likewise.
25711 (inline_read_section, inline_write_summary): Do not stream self_time.
25712 * ipa-inline.h (inline_summary): Drop self_time.
25714 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
25716 * ipa-inline-analysis.c (account_size_time): Rename to ...
25717 (inline_summary::account_size_time): ... this one.
25718 (reset_ipa_call_summary): Turn to ...
25719 (ipa_call_summary::reset): ... this one.
25720 (reset_inline_summary): Turn to ...
25721 (inline_summary::reset): ... this one.
25722 (inline_summary_t::remove): Update.
25723 (inline_summary_t::duplicate): Update.
25724 (ipa_call_summary_t::remove): Update.
25725 (dump_inline_summary): Update.
25726 (estimate_function_body_sizes): Update.
25727 (compute_inline_parameters): Update.
25728 (estimate_node_size_and_time): Update.
25729 (inline_merge_summary): Update.
25730 (inline_update_overall_summary): Update.
25731 (inline_read_section): Update.
25732 (inline_write_summary): Update.
25733 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
25734 add account_size_time and reset member functions.
25735 (ipa_call_summary): Add reset function.
25736 * ipa-predicate.h (predicate::operator &): Constify.
25738 2017-05-22 Richard Biener <rguenther@suse.de>
25740 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
25742 2017-05-19 Jason Merrill <jason@redhat.com>
25744 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
25746 2017-05-19 Marek Polacek <polacek@redhat.com>
25749 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
25750 TYPE_OVERFLOW_WRAPS checks.
25752 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
25754 * tree-core.h (enum omp_clause_default_kind): Add
25755 "OMP_CLAUSE_DEFAULT_PRESENT".
25756 * tree-pretty-print.c (dump_omp_clause): Handle it.
25757 * gimplify.c (enum gimplify_omp_var_data): Add
25758 "GOVD_MAP_FORCE_PRESENT".
25759 (gimplify_adjust_omp_clauses_1): Map it to
25760 "GOMP_MAP_FORCE_PRESENT".
25761 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
25763 * gimplify.c (oacc_default_clause): Clarify.
25765 2017-05-19 Nathan Sidwell <nathan@acm.org>
25767 LANG_HOOK_REGISTER_DUMPS
25768 * toplev.c (general_init): Call register dump lang hook.
25769 * doc/invoke.texi: Document -fdump-lang option family.
25770 * dumpfile.c (dump_files): Remove class dump here.
25771 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
25772 * dumpfile.h (tree_dump_index): Remove TDI_class.
25773 * langhooks-def.h (lhd_register_dumps): Declare.
25774 (LANG_HOOKS_REGISTER_DUMPS): Define.
25775 (LANG_HOOKS_INITIALIZER): Add it.
25776 * langhooks.c (lhd_register_dumps): Define.
25777 * langhooks.h (struct lang_hooks): Add register_dumps.
25779 2017-05-19 Nathan Sidwell <nathan@acm.org>
25781 * context.h (context::set_passes): New.
25782 * context.c (context::context): Do not create pass manager.
25783 * toplev.c (general_init): Create pass manager here.
25785 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
25787 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
25788 use this splitter if two add or or instructions would also work for
25789 the constant we want to generate.
25791 2017-05-19 Richard Biener <rguenther@suse.de>
25794 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
25795 predicate evaluation.
25797 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
25799 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
25801 * ipa-inline.c (want_inline_small_function_p): Do not cast to
25804 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
25806 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
25807 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
25808 (inline_edge_summary_vec): Turn into ...
25809 (ipa_call_summaries): ... this one.
25810 (redirect_to_unreachable, edge_set_predicate,
25811 evaluate_properties_for_edge, inline_summary_alloc,
25812 reset_ipa_call_summary, reset_inline_summary,
25813 inline_summary_t::duplicate): Update.
25814 (inline_edge_duplication_hook): Turn to ...
25815 (ipa_call_summary_t::duplicate): ... this one.
25816 (inline_edge_removal_hook): Turn to ...
25817 (ipa_call_summary_t::remove): ... this one.
25818 (dump_inline_edge_summary): Turn to ...
25819 (dump_ipa_call_summary): ... this one.
25820 (estimate_function_body_sizes): Update.
25821 (inline_update_callee_summaries): Update.
25822 (remap_edge_change_prob): Update.
25823 (remap_edge_summaries): Update.
25824 (inline_merge_summary): Update.
25825 (do_estimate_edge_time): Update.
25826 (inline_generate_summary): Update.
25827 (inline_read_section): Update.
25828 (inline_read_summary): Update.
25829 (inline_free_summary): Update.
25830 * ipa-inline.c (can_inline_edge_p): Update.
25831 (compute_inlined_call_time): Update.
25832 (want_inline_small_function_p): Update.
25833 (edge_badness): Update.
25834 (early_inliner): Update.
25835 * ipa-inline.h (inline_edge_summary): Turn to ...
25836 (ipa_call_summary): ... this one.
25837 (ipa_call_summary_t): New class.
25838 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
25839 (ipa_call_summaries): New.
25840 (inline_edge_summary): Remove.
25841 (estimate_edge_growth): Update.
25842 * ipa-profile.c (ipa_propagate_frequency_1): Update.
25843 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
25844 * ipa-split.c (execute_split_functions): Update.
25845 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
25847 2017-05-19 Richard Biener <rguenther@suse.de>
25849 PR middle-end/80764
25850 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
25852 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
25854 * config/rs6000/rs6000.c (struct machine_function): Add field
25855 fpr_is_wrapped_separately.
25856 (rs6000_get_separate_components): Use 64 components. Handle the
25857 new FPR components.
25858 (rs6000_components_for_bb): Handle the FPR components.
25859 (rs6000_emit_prologue_components): Handle the FPR components.
25860 (rs6000_emit_epilogue_components): Handle the FPR components.
25861 (rs6000_set_handled_components): Handle the FPR components.
25862 (rs6000_emit_prologue): Don't output prologue code for those FPRs
25863 that are already separately shrink-wrapped.
25864 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
25865 that are already separately shrink-wrapped.
25867 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25870 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
25873 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
25874 (define_peephole2 for Altivec d-form load): Add peepholes to catch
25875 cases where the register allocator uses a move and an offsettable
25876 memory operation to/from a FPR register on ISA 2.06/2.07.
25877 (define_peephole2 for Altivec d-form store): Likewise.
25879 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
25882 * config/i386/mmx.md (*mov<mode>_internal): Enable
25883 alternatives 11, 12, 13 and 14 also for 32bit targets.
25884 Remove alternatives 15, 16, 17 and 18.
25885 * config/i386/sse.md (vec_concatv2di): Change
25886 alternative (!x, *y) to (x, ?!*Yn).
25888 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
25890 * dumpfile.h (enum dump_kind): Remove stray comma.
25892 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
25894 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
25895 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
25896 predicate::num_conditions
25897 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
25898 (CHANGED): turn into predicate::changed.
25899 (agg_position_info): Move to ipa-predicate.h
25900 (add_condition, predicate::add_clause, predicate::operator &=,
25901 predicate::or_with, predicate::evaluate, predicate::probability,
25902 dump_condition, dump_clause, predicate::dump,
25903 predicate::remap_after_duplication, predicate::remap_after_inlining,
25904 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
25905 (evaluate_conditions_for_known_args): Update.
25906 (set_cond_stmt_execution_predicate): Update.
25907 * ipa-inline.h: Include ipa-predicate.h
25908 (condition, inline_param_summary, conditions, agg_position_info,
25909 predicate): Move to ipa-predicate.h
25910 * ipa-predicate.c: New file.
25911 * ipa-predicate.h: New file.
25913 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
25915 * final.c (leaf_function_p): Check we are not in a sequence.
25917 2017-05-18 Martin Liska <mliska@suse.cz>
25919 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
25920 * dumpfile.c (dump_register): Use new enum dump_kind.
25921 (get_dump_file_name): Likewise.
25922 (dump_enable_all): Likewise.
25923 (dump_switch_p_1): Likewise.
25924 (enable_rtl_dump_file): Remove usage of TDF_RTL.
25925 * dumpfile.h (enum dump_kind): New enum type.
25926 (struct dump_file_info): Create constructor and
25927 format fields and comments.
25928 * passes.c (pass_manager::register_one_dump_file):
25930 * statistics.c (statistics_early_init): Likewise.
25931 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
25932 TDF_TREE with TDF_SLIM.
25933 (gather_memory_references_ref): Likewise.
25935 2017-05-18 Martin Liska <mliska@suse.cz>
25937 * vec.h (struct vnull): Use it.
25939 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
25941 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
25942 (true_predicate, false_predicate, true_predicate_p,
25943 false_predicate_p): Remove.
25944 (single_cond_predicate, not_inlined_predicate): Turn to member function
25946 (add_condition): Update.
25947 (add_clause): Turn to...
25948 (predicate::add_clause): ... this one; update; allow passing NULL
25950 (and_predicates): Turn to ...
25951 (predicate::operator &=): ... this one.
25952 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
25953 (or_predicates): Turn to ...
25954 (predicate::or_with): ... this one.
25955 (evaluate_predicate): Turn to ...
25956 (predicate::evaluate): ... this one.
25957 (predicate_probability): Turn to ...
25958 (predicate::probability): ... this one.
25959 (dump_condition): Update.
25960 (dump_predicate): Turn to ...
25961 (predicate::dump): ... this one.
25962 (account_size_time): Update.
25963 (edge_set_predicate): Update.
25964 (set_hint_predicate): UPdate.
25965 (evaluate_conditions_for_known_args): Update.
25966 (evaluate_properties_for_edge): Update.
25967 (remap_predicate_after_duplication): Turn to...
25968 (predicate::remap_after_duplication): ... this one.
25969 (remap_hint_predicate_after_duplication): Update.
25970 (inline_summary_t::duplicate): UPdate.
25971 (dump_inline_edge_summary): Update.
25972 (dump_inline_summary): Update.
25973 (set_cond_stmt_execution_predicate): Update.
25974 (set_switch_stmt_execution_predicate): Update.
25975 (compute_bb_predicates): Update.
25976 (will_be_nonconstant_expr_predicate): Update.
25977 (will_be_nonconstant_predicate): Update.
25978 (phi_result_unknown_predicate): Update.
25979 (predicate_for_phi_result): Update.
25980 (array_index_predicate): Update.
25981 (estimate_function_body_sizes): Update.
25982 (estimate_node_size_and_time): Update.
25983 (estimate_ipcp_clone_size_and_time): Update.
25984 (remap_predicate): Rename to ...
25985 (predicate::remap_after_inlining): ... this one.
25986 (remap_hint_predicate): Update.
25987 (inline_merge_summary): Update.
25988 (inline_update_overall_summary): Update.
25989 (estimate_size_after_inlining): Update.
25990 (read_predicate): Rename to ...
25991 (predicate::stream_in): ... this one.
25992 (read_inline_edge_summary): Update.
25993 (write_predicate): Rename to ...
25994 (predicate::stream_out): ... this one.
25995 (write_inline_edge_summary): Update.
25996 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
25997 (clause_t): Turn to uint32_t
25998 (predicate): Turn to class; implement constructor and operators
26000 (size_time_entry): Update.
26001 (inline_summary): Update.
26002 (inline_edge_summary): Update.
26004 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
26006 * fold-const.c (fold_binary_loc): Move transformation...
26007 * match.pd (C - X CMP X): ... here.
26009 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
26011 * config/sparc/sparc.c (sparc_option_override): Set function
26012 alignment for -mcpu=niagara7 to 64 to match the I$ line.
26013 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
26015 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
26017 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
26019 2017-05-18 Marek Polacek <polacek@redhat.com>
26022 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
26023 (pass_ubsan::execute): Call gimple_assign_single_p instead of
26024 gimple_assign_load_p.
26026 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
26028 PR middle-end/80692
26029 * real.c (do_compare): Give decimal_do_compare preference over
26030 comparing just the signs.
26032 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
26034 * doc/md.texi (Canonicalization of Instructions): Describe the
26035 canonical form of instructions that inherently set a condition
26038 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
26040 PR middle-end/80775
26041 * tree-cfg.c: Move deletion of unreachable case statements to after
26042 the merging of consecutive case labels.
26044 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
26046 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
26047 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
26048 restoring of callee-saved registers.
26050 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
26052 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
26053 * config/visium/visium.c (single_set_and_flags): Likewise.
26054 * config/visium/visium.md (Substitutions): Likewise.
26056 2017-05-17 Martin Liska <mliska@suse.cz>
26058 * cfg.c: Introduce dump_flags_t type and
26059 use it instead of int type.
26061 * cfghooks.c: Likewise.
26062 * cfghooks.h (struct cfg_hooks): Likewise.
26063 * cfgrtl.c: Likewise.
26064 * cfgrtl.h: Likewise.
26065 * cgraph.c (cgraph_node::get_body): Likewise.
26066 * coretypes.h: Likewise.
26067 * domwalk.c: Likewise.
26068 * domwalk.h: Likewise.
26069 * dumpfile.c (struct dump_option_value_info): Likewise.
26070 (dump_enable_all): Likewise.
26071 (dump_switch_p_1): Likewise.
26072 (opt_info_switch_p): Likewise.
26073 * dumpfile.h (enum tree_dump_index): Likewise.
26074 (struct dump_file_info): Likewise.
26075 * genemit.c: Likewise.
26076 * generic-match-head.c: Likewise.
26077 * gengtype.c (open_base_files): Likewise.
26078 * gimple-pretty-print.c: Likewise.
26079 * gimple-pretty-print.h: Likewise.
26080 * graph.c (print_graph_cfg): Likewise.
26081 * graphite-scop-detection.c (dot_all_sese): Likewise.
26082 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
26083 * loop-unroll.c (report_unroll): Likewise.
26084 * passes.c (pass_manager::register_one_dump_file): Likewise.
26085 * print-tree.c: Likewise.
26086 * statistics.c: Likewise.
26087 * tree-cfg.c: Likewise.
26088 * tree-cfg.h: Likewise.
26089 * tree-dfa.c: Likewise.
26090 * tree-dfa.h: Likewise.
26091 * tree-dump.c (dump_function): Likewise.
26092 * tree-dump.h (struct dump_info): Likewise.
26093 * tree-pretty-print.c: Likewise.
26094 * tree-pretty-print.h: Likewise.
26095 * tree-ssa-live.c: Likewise.
26096 * tree-ssa-live.h: Likewise.
26097 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
26098 * tree-vect-loop.c: Likewise.
26099 * tree-vect-slp.c: Likewise.
26101 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
26102 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26104 PR tree-optimization/80457
26105 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
26106 of all arguments to a statement as scalar_to_vec operations.
26107 (vectorizable_call): Adjust call to vect_model_simple_cost for
26109 (vectorizable_conversion): Likewise.
26110 (vectorizable_assignment): Likewise.
26111 (vectorizable_shift): Likewise.
26112 (vectorizable_operation): Likewise.
26113 (vectorizable_comparison): Likewise.
26114 (vect_is_simple_cond): Record the def types for operands.
26115 (vectorizable_condition): Likewise, call vect_model_simple_cost.
26116 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
26117 for statement argument count.
26119 2017-05-16 Carl Love <cel@us.ibm.com>
26121 * config/rs6000/rs6000-c: Add support for built-in functions
26122 vector unsigned long long vec_bperm (vector unsigned long long,
26123 vector unsigned char)
26124 vector signed long long vec_mule (vector signed int,
26126 vector unsigned long long vec_mule (vector unsigned int,
26127 vector unsigned int)
26128 vector signed long long vec_mulo (vector signed int,
26130 vector unsigned long long vec_mulo (vector unsigned int,
26131 vector unsigned int)
26132 vector signed char vec_sldw (vector signed char,
26133 vector signed char,
26135 vector unsigned char vec_sldw (vector unsigned char,
26136 vector unsigned char,
26138 vector signed short vec_sldw (vector signed short,
26139 vector signed short,
26141 vector unsigned short vec_sldw (vector unsigned short,
26142 vector unsigned short,
26144 vector signed int vec_sldw (vector signed int,
26147 vector unsigned int vec_sldw (vector unsigned int,
26148 vector unsigned int,
26150 vector signed long long vec_sldw (vector signed long long,
26151 vector signed long long,
26153 vector unsigned long long vec_sldw (vector unsigned long long,
26154 vector unsigned long long,
26156 * config/rs6000/rs6000-c: Add support for built-in functions
26157 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
26158 * config/rs6000/altivec.h: Add defintion for vec_sldw.
26159 * doc/extend.texi: Update the built-in documentation for the
26160 new built-in functions.
26162 2017-05-16 Marek Polacek <polacek@redhat.com>
26166 * tree.c (save_expr): Don't fold the expression.
26168 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
26170 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
26171 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
26172 and (?*y,m). Update insn attributes.
26174 2017-05-16 Martin Liska <mliska@suse.cz>
26176 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
26177 flags argument of print_gimple_stmt, print_gimple_expr,
26178 print_generic_stmt and print_generic_expr.
26179 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
26180 * coretypes.h: Likewise.
26181 * except.c (dump_eh_tree): Likewise.
26182 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
26183 * gimple-pretty-print.h: Likewise.
26184 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
26185 (backprop::push_to_worklist): Likewise.
26186 (backprop::pop_from_worklist): Likewise.
26187 (backprop::process_use): Likewise.
26188 (backprop::intersect_uses): Likewise.
26189 (note_replacement): Likewise.
26190 * gimple-ssa-store-merging.c
26191 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
26192 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
26193 (pass_store_merging::execute): Likewise.
26194 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
26195 (ssa_base_cand_dump_callback): Likewise.
26196 (dump_incr_vec): Likewise.
26197 (replace_refs): Likewise.
26198 (replace_mult_candidate): Likewise.
26199 (create_add_on_incoming_edge): Likewise.
26200 (create_phi_basis): Likewise.
26201 (insert_initializers): Likewise.
26202 (all_phi_incrs_profitable): Likewise.
26203 (introduce_cast_before_cand): Likewise.
26204 (replace_one_candidate): Likewise.
26205 * gimplify.c (gimplify_expr): Likewise.
26206 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
26207 (set_rename): Likewise.
26208 (rename_uses): Likewise.
26209 (copy_loop_phi_nodes): Likewise.
26210 (add_close_phis_to_merge_points): Likewise.
26211 (copy_loop_close_phi_args): Likewise.
26212 (copy_cond_phi_args): Likewise.
26213 (graphite_copy_stmts_from_block): Likewise.
26214 (translate_pending_phi_nodes): Likewise.
26215 * graphite-poly.c (print_pdr): Likewise.
26216 (dump_gbb_cases): Likewise.
26217 (dump_gbb_conditions): Likewise.
26218 (print_scop_params): Likewise.
26219 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
26220 (build_cross_bb_scalars_use): Likewise.
26221 (gather_bbs::before_dom_children): Likewise.
26222 * hsa-dump.c (dump_hsa_immed): Likewise.
26223 * ipa-cp.c (print_ipcp_constant_value): Likewise.
26224 (get_replacement_map): Likewise.
26225 * ipa-inline-analysis.c (dump_condition): Likewise.
26226 (estimate_function_body_sizes): Likewise.
26227 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
26228 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
26229 * ipa-prop.c (ipa_dump_param): Likewise.
26230 (ipa_print_node_jump_functions_for_edge): Likewise.
26231 (ipa_modify_call_arguments): Likewise.
26232 (ipa_modify_expr): Likewise.
26233 (ipa_dump_param_adjustments): Likewise.
26234 (ipa_dump_agg_replacement_values): Likewise.
26235 (ipcp_modif_dom_walker::before_dom_children): Likewise.
26236 * ipa-pure-const.c (check_stmt): Likewise.
26237 (pass_nothrow::execute): Likewise.
26238 * ipa-split.c (execute_split_functions): Likewise.
26239 * omp-offload.c (dump_oacc_loop_part): Likewise.
26240 (dump_oacc_loop): Likewise.
26241 * trans-mem.c (tm_log_emit): Likewise.
26242 (tm_memopt_accumulate_memops): Likewise.
26243 (dump_tm_memopt_set): Likewise.
26244 (dump_tm_memopt_transform): Likewise.
26245 * tree-cfg.c (gimple_verify_flow_info): Likewise.
26246 (print_loop): Likewise.
26247 * tree-chkp-opt.c (chkp_print_addr): Likewise.
26248 (chkp_gather_checks_info): Likewise.
26249 (chkp_get_check_result): Likewise.
26250 (chkp_remove_check_if_pass): Likewise.
26251 (chkp_use_outer_bounds_if_possible): Likewise.
26252 (chkp_reduce_bounds_lifetime): Likewise.
26253 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
26254 (chkp_mark_completed_bounds): Likewise.
26255 (chkp_register_incomplete_bounds): Likewise.
26256 (chkp_mark_invalid_bounds): Likewise.
26257 (chkp_maybe_copy_and_register_bounds): Likewise.
26258 (chkp_build_returned_bound): Likewise.
26259 (chkp_get_bound_for_parm): Likewise.
26260 (chkp_build_bndldx): Likewise.
26261 (chkp_get_bounds_by_definition): Likewise.
26262 (chkp_generate_extern_var_bounds): Likewise.
26263 (chkp_get_bounds_for_decl_addr): Likewise.
26264 * tree-chrec.c (chrec_apply): Likewise.
26265 * tree-data-ref.c (dump_data_reference): Likewise.
26266 (dump_subscript): Likewise.
26267 (dump_data_dependence_relation): Likewise.
26268 (analyze_overlapping_iterations): Likewise.
26269 * tree-inline.c (expand_call_inline): Likewise.
26270 (tree_function_versioning): Likewise.
26271 * tree-into-ssa.c (dump_defs_stack): Likewise.
26272 (dump_currdefs): Likewise.
26273 (dump_names_replaced_by): Likewise.
26274 (dump_update_ssa): Likewise.
26275 (update_ssa): Likewise.
26276 * tree-object-size.c (pass_object_sizes::execute): Likewise.
26277 * tree-parloops.c (build_new_reduction): Likewise.
26278 (try_create_reduction_list): Likewise.
26279 (ref_conflicts_with_region): Likewise.
26280 (oacc_entry_exit_ok_1): Likewise.
26281 (oacc_entry_exit_single_gang): Likewise.
26282 * tree-pretty-print.h: Likewise.
26283 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
26284 (get_scalar_evolution): Likewise.
26285 (add_to_evolution): Likewise.
26286 (get_loop_exit_condition): Likewise.
26287 (analyze_evolution_in_loop): Likewise.
26288 (analyze_initial_condition): Likewise.
26289 (analyze_scalar_evolution): Likewise.
26290 (instantiate_scev): Likewise.
26291 (number_of_latch_executions): Likewise.
26292 (gather_chrec_stats): Likewise.
26293 (final_value_replacement_loop): Likewise.
26294 (scev_const_prop): Likewise.
26295 * tree-sra.c (dump_access): Likewise.
26296 (disqualify_candidate): Likewise.
26297 (create_access): Likewise.
26298 (reject): Likewise.
26299 (maybe_add_sra_candidate): Likewise.
26300 (create_access_replacement): Likewise.
26301 (analyze_access_subtree): Likewise.
26302 (analyze_all_variable_accesses): Likewise.
26303 (sra_modify_assign): Likewise.
26304 (initialize_constant_pool_replacements): Likewise.
26305 (find_param_candidates): Likewise.
26306 (decide_one_param_reduction): Likewise.
26307 (replace_removed_params_ssa_names): Likewise.
26308 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
26309 * tree-ssa-copy.c (dump_copy_of): Likewise.
26310 (copy_prop_visit_cond_stmt): Likewise.
26311 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
26312 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
26313 (record_equivalences_from_stmt): Likewise.
26314 * tree-ssa-dse.c (compute_trims): Likewise.
26315 (delete_dead_call): Likewise.
26316 (delete_dead_assignment): Likewise.
26317 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
26318 (forward_propagate_into_cond): Likewise.
26319 (pass_forwprop::execute): Likewise.
26320 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
26321 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
26323 (move_computations_worker): Likewise.
26324 (execute_sm): Likewise.
26325 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
26326 (remove_exits_and_undefined_stmts): Likewise.
26327 (remove_redundant_iv_tests): Likewise.
26328 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
26329 (adjust_iv_update_pos): Likewise.
26330 * tree-ssa-math-opts.c (bswap_replace): Likewise.
26331 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
26332 (value_replacement): Likewise.
26333 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
26334 * tree-ssa-pre.c (print_pre_expr): Likewise.
26335 (get_representative_for): Likewise.
26336 (create_expression_by_pieces): Likewise.
26337 (insert_into_preds_of_block): Likewise.
26338 (eliminate_insert): Likewise.
26339 (eliminate_dom_walker::before_dom_children): Likewise.
26340 (eliminate): Likewise.
26341 (remove_dead_inserted_code): Likewise.
26342 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
26343 * tree-ssa-reassoc.c (get_rank): Likewise.
26344 (eliminate_duplicate_pair): Likewise.
26345 (eliminate_plus_minus_pair): Likewise.
26346 (eliminate_not_pairs): Likewise.
26347 (undistribute_ops_list): Likewise.
26348 (eliminate_redundant_comparison): Likewise.
26349 (update_range_test): Likewise.
26350 (optimize_range_tests_var_bound): Likewise.
26351 (optimize_vec_cond_expr): Likewise.
26352 (rewrite_expr_tree): Likewise.
26353 (rewrite_expr_tree_parallel): Likewise.
26354 (linearize_expr): Likewise.
26355 (break_up_subtract): Likewise.
26356 (linearize_expr_tree): Likewise.
26357 (attempt_builtin_powi): Likewise.
26358 (attempt_builtin_copysign): Likewise.
26359 (transform_stmt_to_copy): Likewise.
26360 (transform_stmt_to_multiply): Likewise.
26361 (dump_ops_vector): Likewise.
26362 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
26363 (print_scc): Likewise.
26364 (set_ssa_val_to): Likewise.
26365 (visit_reference_op_store): Likewise.
26366 (visit_use): Likewise.
26367 (sccvn_dom_walker::before_dom_children): Likewise.
26368 (run_scc_vn): Likewise.
26369 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
26371 (expr_hash_elt::print): Likewise.
26372 (const_and_copies::pop_to_marker): Likewise.
26373 (const_and_copies::record_const_or_copy_raw): Likewise.
26374 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
26375 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
26376 (dump_predicates): Likewise.
26377 (find_uninit_use): Likewise.
26378 (warn_uninitialized_phi): Likewise.
26379 (pass_late_warn_uninitialized::execute): Likewise.
26380 * tree-ssa.c (verify_vssa): Likewise.
26381 (verify_ssa): Likewise.
26382 (maybe_optimize_var): Likewise.
26383 * tree-vrp.c (dump_value_range): Likewise.
26384 (dump_all_value_ranges): Likewise.
26385 (dump_asserts_for): Likewise.
26386 (register_edge_assert_for_2): Likewise.
26387 (vrp_visit_cond_stmt): Likewise.
26388 (vrp_visit_switch_stmt): Likewise.
26389 (vrp_visit_stmt): Likewise.
26390 (vrp_visit_phi_node): Likewise.
26391 (simplify_cond_using_ranges_1): Likewise.
26392 (fold_predicate_in): Likewise.
26393 (evrp_dom_walker::before_dom_children): Likewise.
26394 (evrp_dom_walker::push_value_range): Likewise.
26395 (evrp_dom_walker::pop_value_range): Likewise.
26396 (execute_early_vrp): Likewise.
26398 2017-05-16 Richard Biener <rguenther@suse.de>
26400 * dwarf2out.c (loc_list_from_tree_1): Do not create
26401 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
26403 2017-05-16 Richard Biener <rguenther@suse.de>
26405 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
26407 (note_variable_value_in_expr): If we resolved the decl ref
26408 do not push to the stack.
26410 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
26412 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
26413 operations in fast-math mode.
26414 (vaddq_f16): Likewise.
26415 (vmul_f16): Likewise.
26416 (vmulq_f16): Likewise.
26417 (vsub_f16): Likewise.
26418 (vsubq_f16): Likewise.
26419 * config/arm/neon.md (add<mode>3): New.
26421 (fma:<VH:mode>3): New. Also remove outdated comment.
26424 2017-05-16 Martin Liska <mliska@suse.cz>
26428 * ipa-devirt.c (warn_types_mismatch): Fix typo.
26429 (odr_types_equivalent_p): Likewise.
26431 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
26433 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
26435 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
26438 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
26439 non-interunit SSE move alternatives with '?'.
26440 (zero-extendsidi peephole2): New peephole to skip intermediate
26441 general register in SSE zero-extend sequence.
26443 2017-05-15 Jeff Law <law@redhat.com>
26445 * reorg.c (relax_delay_slots): Create a new variable to hold
26446 the temporary target rather than clobbering TARGET_LABEL.
26448 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
26449 missing argument to extract_bit_field call.
26450 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
26452 2017-05-15 Martin Liska <mliska@suse.cz>
26455 * gcc.c (process_command): Do not allow empty argument of -o option.
26457 2017-05-15 Renlin Li <renlin.li@arm.com>
26459 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
26460 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
26461 * config/aarch64/constraints.md (Usf): Add long call check.
26462 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
26463 (call_value): Likewise.
26464 (sibcall): Likewise.
26465 (sibcall_value): Likewise.
26467 (call_value_insn): New.
26468 (sibcall_insn): Update rtx pattern.
26469 (sibcall_value_insn): Likewise.
26470 (call_internal): Remove.
26471 (call_value_internal): Likewise.
26472 (sibcall_internal): Likewise.
26473 (sibcall_value_internal): Likewise.
26474 (call_reg): Likewise.
26475 (call_symbol): Likewise.
26476 (call_value_reg): Likewise.
26477 (call_value_symbol): Likewise.
26479 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
26482 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
26484 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
26486 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
26487 compatible with CCGOCmode and with CCZmode.
26489 2017-05-14 Martin Sebor <msebor@redhat.com>
26491 PR middle-end/77671
26492 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
26493 (gimple_fold_builtin_snprintf): Same.
26494 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
26495 (gimple_fold_builtin_snprintf): Same.
26496 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
26497 of character types.
26498 (is_call_safe): New function.
26499 (try_substitute_return_value): Call it.
26500 (try_simplify_call): New function.
26501 (pass_sprintf_length::handle_gimple_call): Call it.
26503 2017-05-14 Martin Sebor <msebor@redhat.com>
26505 PR middle-end/80669
26506 * builtins.c (expand_builtin_stpncpy): Simplify.
26508 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
26510 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
26511 * config/i386/i386.h
26512 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
26513 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
26514 (struct machine_function): Add new members call_ms2sysv,
26515 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
26516 (struct machine_frame_state): New fields sp_realigned and
26517 sp_realigned_offset.
26518 * config/i386/i386.c
26519 (enum xlogue_stub): New enum.
26520 (enum xlogue_stub_sets): New enum.
26521 (class xlogue_layout): New class.
26522 (struct ix86_frame): New fields stack_realign_allocate_offset,
26523 stack_realign_offset and outlined_save_offset. Modify comments to
26524 detail stack layout when using out-of-line stubs.
26525 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
26526 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
26527 -mcall-ms2sysv-xlogues.
26528 (stub_managed_regs): New static variable.
26529 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
26530 registers managed by out-of-line stub.
26531 (disable_call_ms2sysv_xlogues): New function.
26532 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
26533 m->call_ms2sysv when appropriate and compute frame layout for
26535 (sp_valid_at, fp_valid_at): New inline functions.
26536 (choose_basereg): New function.
26537 (choose_baseaddr): Add align parameter, use choose_basereg and modify
26539 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
26540 Use align parameter of choose_baseaddr to generated aligned SSE movs
26542 (pro_epilogue_adjust_stack): Modify to track
26543 machine_frame_state::sp_realigned.
26544 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
26545 (ix86_nsaved_sseregs): Likewise.
26546 (ix86_emit_save_regs): Likewise.
26547 (ix86_emit_save_regs_using_mov): Likewise.
26548 (ix86_emit_save_sse_regs_using_mov): Likewise.
26549 (get_scratch_register_on_entry): Likewise.
26550 (gen_frame_set): New function.
26551 (gen_frame_load): Likewise.
26552 (gen_frame_store): Likewise.
26553 (emit_outlined_ms2sysv_save): Likewise.
26554 (emit_outlined_ms2sysv_restore): Likewise.
26555 (ix86_expand_prologue): Modify stack re-alignment code and call
26556 emit_outlined_ms2sysv_save when appropriate.
26557 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
26558 parameter rtx_insn *insn, which allows the function to be used to only
26559 generate the notes.
26560 (ix86_expand_epilogue): Modify validity checks of frame and stack
26561 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
26562 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
26563 * config/i386/predicates.md
26564 (save_multiple): New predicate.
26565 (restore_multiple): Likewise.
26566 * config/i386/sse.md
26567 (save_multiple<mode>): New pattern.
26568 (save_multiple_realign<mode>): Likewise.
26569 (restore_multiple<mode>): Likewise.
26570 (restore_multiple_and_return<mode>): Likewise.
26571 (restore_multiple_leave_return<mode>): Likewise.
26572 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
26574 2017-05-14 Julia Koval <julia.koval@intel.com>
26576 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
26577 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
26578 (__builtin_ia32_xsetbv): New builtins.
26579 * config/i386/i386.c (ix86_expand_special_args_builtin):
26581 (ix86_expand_builtin): Special expand for new intrinsics.
26582 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
26583 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
26584 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
26586 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26588 * cfganal.c (inverted_post_order_compute): Change argument type
26590 * cfganal.h (inverted_post_order_compute): Adjust prototype.
26591 * df-core.c (rest_of_handle_df_initialize): Adjust.
26592 (rest_of_handle_df_finish): Likewise.
26593 (df_analyze_1): Likewise.
26594 (df_analyze): Likewise.
26595 (loop_inverted_post_order_compute): Change argument to be a vec *.
26596 (df_analyze_loop): Adjust.
26597 (df_get_n_blocks): Likewise.
26598 (df_get_postorder): Likewise.
26599 * df.h (struct df_d): Change field to be a vec.
26600 * lcm.c (compute_laterin): Adjust.
26601 (compute_available): Likewise.
26602 * lra-lives.c (lra_create_live_ranges_1): Likewise.
26603 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
26604 * tree-ssa-pre.c (compute_antic): Likewise.
26606 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26608 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
26609 (depth_first_search::depth_first_search): Change structure init
26610 function to this constructor.
26611 (depth_first_search::add_bb): Rename function to this member.
26612 (depth_first_search::execute): Likewise.
26613 (flow_dfs_compute_reverse_finish): Adjust.
26615 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26617 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
26618 (longest_simple_path): Likewise.
26619 * shrink-wrap.c (spread_components): Likewise.
26620 (disqualify_problematic_components): Likewise.
26621 (emit_common_heads_for_components): Likewise.
26622 (emit_common_tails_for_components): Likewise.
26623 (insert_prologue_epilogue_for_components): Likewise.
26625 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26627 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
26630 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26632 * df-core.c (df_set_blocks): Start using auto_bitmap.
26633 (df_compact_blocks): Likewise.
26634 * df-problems.c (df_rd_confluence_n): Likewise.
26635 * df-scan.c (df_insn_rescan_all): Likewise.
26636 (df_process_deferred_rescans): Likewise.
26637 (df_update_entry_block_defs): Likewise.
26638 (df_update_exit_block_uses): Likewise.
26639 (df_entry_block_bitmap_verify): Likewise.
26640 (df_exit_block_bitmap_verify): Likewise.
26641 (df_scan_verify): Likewise.
26642 * lra-constraints.c (lra_constraints): Likewise.
26643 (undo_optional_reloads): Likewise.
26644 (lra_undo_inheritance): Likewise.
26645 * lra-remat.c (calculate_gen_cands): Likewise.
26646 (do_remat): Likewise.
26647 * lra-spills.c (assign_spill_hard_regs): Likewise.
26648 (spill_pseudos): Likewise.
26649 * tree-ssa-pre.c (bitmap_set_and): Likewise.
26650 (bitmap_set_subtract_values): Likewise.
26652 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26654 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
26655 management with auto_bitmap.
26656 (fix_inter_tick): Likewise.
26657 (fix_recovery_deps): Likewise.
26658 * ira.c (add_store_equivs): Likewise.
26659 (find_moveable_pseudos): Likewise.
26660 (split_live_ranges_for_shrink_wrap): Likewise.
26661 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
26662 (rtx_reuse_manager::seen_def_p): Likewise.
26663 (rtx_reuse_manager::set_seen_def): Likewise.
26664 * print-rtl.h (class rtx_reuse_manager): Likewise.
26666 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26668 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
26670 (migrate_btr_def): Likewise.
26671 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
26672 * df-core.c (loop_post_order_compute): Likewise.
26673 (loop_inverted_post_order_compute): Likewise.
26674 * hsa-common.h: Likewise.
26675 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
26676 * init-regs.c (initialize_uninitialized_regs): Likewise.
26677 * ipa-inline.c (resolve_noninline_speculation): Likewise.
26678 (inline_small_functions): Likewise.
26679 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
26680 * ira.c (combine_and_move_insns): Likewise.
26681 (build_insn_chain): Likewise.
26682 * loop-invariant.c (find_invariants): Likewise.
26683 * lower-subreg.c (propagate_pseudo_copies): Likewise.
26684 * predict.c (tree_predict_by_opcode): Likewise.
26685 (predict_paths_leading_to): Likewise.
26686 (predict_paths_leading_to_edge): Likewise.
26687 (estimate_loops_at_level): Likewise.
26688 (estimate_loops): Likewise.
26689 * shrink-wrap.c (try_shrink_wrapping): Likewise.
26690 (spread_components): Likewise.
26691 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
26692 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
26693 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
26694 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
26695 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
26696 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
26697 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
26698 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
26699 (mark_threaded_blocks): Likewise.
26700 (thread_through_all_blocks): Likewise.
26701 * tree-ssa.c (verify_ssa): Likewise.
26702 (execute_update_addresses_taken): Likewise.
26703 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
26705 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26707 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
26709 (post_order_compute): Likewise.
26710 (inverted_post_order_compute): Likewise.
26711 (pre_and_rev_post_order_compute_fn): Likewise.
26713 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26715 * genrecog.c (int_set::int_set): Explicitly construct our
26716 auto_vec base class.
26717 * vec.h (auto_vec::auto_vec): New constructor.
26719 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26721 * bitmap.h (class auto_bitmap): New constructor taking
26722 bitmap_obstack * argument.
26724 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26726 * bitmap.h (class auto_bitmap): Change type of m_bits to
26727 bitmap_head, and adjust ctor / dtor and member operators.
26729 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
26731 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
26732 when returned register mode doesn't match original mode.
26734 2017-05-12 Jeff Law <law@redhat.com>
26735 Jakub Jelinek <jakub@redhat.com>
26737 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
26738 we look for cc setter after the compare-elim changes.
26739 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
26740 within the vector to match what compare-elim now expects.
26741 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
26742 (xorsi3_flags, one_cmplsi2_flags): Likewise.
26744 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
26745 after the compare-elim changes.
26746 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
26747 the vector to match what compare-elim now expects.
26748 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
26749 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
26750 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
26751 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
26752 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
26754 * config/visium/visium.c (single_set_and_flags): Fix where
26755 we look for cc setter after the compare-elim changes.
26756 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
26757 with the vector to match what compare-elim now expects.
26758 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
26759 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
26760 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
26761 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
26762 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
26763 (neg<mode>2_insn_set_overflow): Likewise.
26765 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
26767 PR middle-end/79794
26768 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
26769 maybe_expand_insn call, set ops[0].target. If still set after call,
26770 set alt_rtl. Add extra arg to recursive calls.
26771 (extract_bit_field): Add alt_rtl argument. Pass to
26773 * expmed.h (extract_bit_field): Fix prototype.
26774 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
26775 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
26776 to extract_bit_field_calls.
26777 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
26778 Pass alt_rtl to extract_bit_field calls.
26779 * calls.c (store_unaligned_arguments_into_psuedos)
26780 load_register_parameters): Pass extra NULL to extract_bit_field calls.
26781 * optabs.c (maybe_legitimize_operand): Clear op->target when call
26783 * optabs.h (struct expand_operand): Add target bitfield.
26785 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
26787 * compare-elim.c (try_eliminate_compare): Canonicalize
26788 operation with embedded compare to
26789 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
26790 (set (reg) (operation)].
26792 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
26794 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
26797 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
26798 cost of adding a carry flag for ADC instruction.
26799 [case MINUS]: Ignore the cost of subtracting a carry flag
26800 for SBB instruction.
26802 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
26804 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
26806 * config/rs6000/bmiintrin.h: New file.
26807 * config/rs6000/bmi2intrin.h: New file.
26808 * config/rs6000/x86intrin.h: New file.
26810 2017-05-12 Jeff Law <law@redhat.com>
26812 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
26815 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
26817 PR middle-end/80707
26818 * tree-cfg.c: Remove cfg edges of unreachable case statements.
26820 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26822 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
26823 early expansion of vector divide builtins.
26824 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
26825 builtins identified as having unsigned arguments.
26827 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26829 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
26830 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
26831 expansion of vector logical operations (and, andc, or, xor,
26834 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26836 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
26837 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
26839 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
26841 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
26842 early GIMPLE expansion of vector multiplies.
26844 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26846 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
26847 TARGET_HAVE_MOVT conditional.
26848 (movt splitter): Likewise.
26850 2017-05-12 Richard Biener <rguenther@suse.de>
26852 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
26853 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
26854 Fold all stmts not inplace.
26856 2017-05-12 Richard Biener <rguenther@suse.de>
26858 PR tree-optimization/80713
26859 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
26860 inserted_exprs bit for not removed stmts.
26862 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
26864 PR middle-end/69921
26865 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
26866 parallelized" attribute for parallelized OpenACC kernels.
26867 * omp-offload.c (execute_oacc_device_lower): Use it.
26869 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
26870 Set "oacc kernels" attribute.
26871 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
26872 parameter. Adjust all users.
26873 (oacc_fn_attrib_kernels_p): Remove function.
26874 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
26875 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
26876 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
26877 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
26878 assert "oacc kernels" attribute is set.
26880 2017-05-11 Carl Love <cel@us.ibm.com>
26882 * config/rs6000/rs6000-c: Add support for built-in functions
26883 vector unsigned char vec_popcnt (vector signed char)
26884 vector unsigned char vec_popcnt (vector unsigned char)
26885 vector unsigned short vec_popcnt (vector signed short)
26886 vector unsigned short vec_popcnt (vector unsigned short)
26887 vector unsigned int vec_popcnt (vector signed int)
26888 vector unsigned int vec_popcnt (vector unsigned int)
26889 vector unsigned long long vec_popcnt (vector signed long long)
26890 vector unsigned long long vec_popcnt (vector unsigned long long)
26891 vector signed long long vec_slo (vector signed long long,
26892 vector signed char)
26893 vector signed long long vec_slo (vector signed long long,
26894 vector unsigned char)
26895 vector unsigned long long vec_slo (vector unsigned long long,
26896 vector signed char)
26897 vector unsigned long long vec_slo (vector unsigned long long,
26898 vector unsigned char)
26899 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
26900 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
26901 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
26902 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
26903 * doc/extend.texi: Update the built-in documentation file for the
26904 new built-in functions.
26906 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
26908 * attribs.h (sorted_attr_string): Move machine independent
26909 functions for target clone support from the i386 port to common
26910 code. Rename ix86_function_versions to common_function_versions.
26911 Rename make_name to make_unique_name.
26912 (common_function_versions): Likewise.
26913 (make_unique_name): Likewise.
26914 (make_dispatcher_decl): Likewise.
26915 (is_function_default_version): Likewise.
26916 * attribs.c (attr_strcmp): Likewise.
26917 (sorted_attr_string): Likewise.
26918 (common_function_versions): Likewise.
26919 (make_unique_name): Likewise.
26920 (make_dispatcher_decl): Likewise.
26921 (is_function_default_version): Likewise.
26922 * config/i386/i386.c (attr_strcmp): Likewise.
26923 (sorted_attr_string): Likewise.
26924 (ix86_function_versions): Likewise.
26925 (make_name): Likewise.
26926 (make_dispatcher_decl): Likewise.
26927 (is_function_default_version): Likewise.
26928 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
26930 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26933 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
26934 Account for direct move costs for vec_construct of integer
26937 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
26940 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
26941 (UNSPEC_STX_ATOMIC): Ditto.
26942 (loaddi_via_sse): New insn.
26943 (storedi_via_sse): Ditto.
26944 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
26945 Update corresponding peephole2 patterns.
26946 (atomic_storedi_fpu): Ditto.
26948 2017-05-11 Julia Koval <julia.koval@intel.com>
26950 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
26951 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
26953 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
26954 (__builtin_ia32_rsqrt14ss_mask): New builtins.
26955 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
26957 2017-05-11 Nathan Sidwell <nathan@acm.org>
26959 * graphite-poly.c: Include dumpfile.h.
26961 * dumpfle.h (dump_function): Declare here ...
26962 * tree-dump.h (dump_function): ... not here.
26963 * dumpfile.c: #include tree-cfg.h.
26964 (dump_function): Move here from ...
26965 * tree-dump.c (dump_function): ... here.
26966 * gimplify.c: #include splay-tree.h, not tree-dump.h.
26967 * graphite-poly.c: Don't include tree-dump.h.
26968 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
26969 * print-tree.c: Likewise.
26970 * stor-layout.c: Likewise.
26971 * tree-nested.c: Likewise.
26973 * dumpfile.c (dump_start): Use TDF_FLAGS.
26974 (dump_enable_all): Fix TDF_KIND check thinko.
26976 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
26978 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26979 array entries to represent two legal parameterizations of the
26980 overloaded __builtin_cmpb function, as represented by the
26981 P6_OV_BUILTIN_CMPB constant.
26982 (altivec_resolve_overloaded_builtin): Add special case handling
26983 for the __builtin_cmpb function, as represented by the
26984 P6_OV_BUILTIN_CMPB constant.
26985 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
26986 (BU_P6_64BIT_2): New macro.
26987 (BU_P6_OVERLOAD_2): New macro
26988 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
26989 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
26990 (CMPB): Add overload support to represent both 32-bit and 64-bit
26991 compare-bytes function.
26992 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
26993 support for TARGET_CMPB.
26994 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
26995 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26996 documentation of the __builtin_cmpb overloaded built-in function.
26998 2017-05-11 Richard Biener <rguenther@suse.de>
27000 PR tree-optimization/80705
27001 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
27002 bases are not vectorizable.
27004 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27006 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
27007 when counting register pressure.
27009 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27011 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
27012 (iv_ca_more_deps): Renamed to ...
27013 (iv_ca_compare_deps): ... this.
27014 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
27016 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27018 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
27020 (determine_group_iv_costs): ... here.
27021 (find_inv_vars_cb): Record inv var if it's not recorded before.
27023 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27025 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
27026 (get_shiftadd_cost): Ditto.
27028 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27030 * tree-ssa-address.c: Include header file.
27031 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
27033 (add_to_parts): Refactor.
27034 (addr_to_parts): New parameter. Update use of move_hint_to_base.
27035 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
27038 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27040 PR tree-optimization/53090
27041 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
27043 (extract_cond_operands): Detect condition with IV on both sides
27044 and return COMP_IV_EXPR_2.
27045 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
27046 (rewrite_use_compare): Simplify by removing call to function
27047 extract_cond_operands.
27049 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27051 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
27052 (extract_cond_operands): Detect condition comparing against non-
27053 invariant bound and return appropriate enum value.
27054 (find_interesting_uses_cond): Update use of extract_cond_operands.
27055 Handle its return value accordingly.
27056 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
27058 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27060 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
27061 nonlinear iv_use computation in loop invariant sensitive way.
27063 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27065 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
27066 (find_iv_candidates): Call relate_compare_use_with_all_cands.
27068 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27070 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
27071 (dump_cand): Support iv_cand.inv_exprs.
27072 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
27074 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
27077 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27079 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
27081 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
27082 as local function. Include necessary header files.
27083 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
27085 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27087 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
27089 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27091 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
27092 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
27093 RSHIFT_EXPR and BIT_NOT_EXPR.
27095 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27097 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
27098 (adjust_setup_cost): New parameter supporting round up adjustment.
27099 (struct address_cost_data): Delete.
27100 (force_expr_to_var_cost): Don't bound cost with spill_cost.
27101 (split_address_cost, ptr_difference_cost): Delete.
27102 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
27103 (struct ainc_cost_data): New struct.
27104 (get_address_cost_ainc): New function.
27105 (get_address_cost, get_computation_cost): Reimplement.
27106 (determine_group_iv_cost_address): Record inv_expr for all uses of
27108 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
27109 (iv_ca_has_deps): Reimplemented to ...
27110 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
27112 (iv_ca_extend): Call iv_ca_more_deps.
27114 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27116 * tree-ssa-address.c (struct mem_address): Move to header file.
27117 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
27118 * tree-ssa-address.h (struct mem_address): Move from C file.
27119 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
27121 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27123 * tree-affine.h (aff_combination_type): New interface.
27124 (aff_combination_zero_p): Remove static.
27125 (aff_combination_const_p): New interface.
27126 (aff_combination_singleton_var_p): New interfaces.
27128 2017-05-11 Richard Biener <rguenther@suse.de>
27130 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
27131 Skip unreachable blocks and destinations.
27132 (eliminate): Move stmt removal and fixup ...
27133 (fini_eliminate): ... here. Skip inserted exprs.
27134 (pass_pre::execute): Move fini_pre after fini_eliminate.
27135 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
27136 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
27137 PRE to get rid of dead code that has invalid SSA form and
27138 split critical edges again.
27140 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27142 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
27144 2017-05-11 Richard Biener <rguenther@suse.de>
27146 * passes.c (execute_function_todo): Verify loops if they are
27147 said to be up-to-date.
27148 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
27149 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
27151 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
27154 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
27155 handle calling assemble_external ourself.
27158 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
27159 modes with zero size. Enhance comment.
27161 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27163 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
27164 built-ins for vec_xl and vec_xst with short and char pointer
27167 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
27169 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
27170 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
27171 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
27172 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
27173 (_mm_maskz_min_round_ss): New intrinsics.
27174 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
27175 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
27176 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
27177 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
27178 (__builtin_ia32_minss_mask_round): New builtins.
27179 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27180 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27181 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
27183 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
27184 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
27186 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
27189 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
27191 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
27192 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
27193 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
27194 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
27195 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
27196 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
27197 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
27198 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27199 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
27200 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
27201 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
27202 (__builtin_ia32_mulss_mask_round): New builtins.
27203 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27204 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27205 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
27207 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
27208 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
27210 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
27213 2017-05-10 Julia Koval <julia.koval@intel.com>
27215 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
27216 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
27217 (_mm256_setr_m128i): New intrinsics.
27219 2017-05-10 Julia Koval <julia.koval@intel.com>
27221 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
27222 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
27223 (_mm_maskz_rcp14_ss): New intrinsics.
27224 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
27225 (__builtin_ia32_rcp14ss_mask): New builtins.
27226 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
27228 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
27230 PR tree-optimization/51513
27231 * tree-cfg.c (gimple_seq_unreachable_p): New function.
27232 (assert_unreachable_fallthru_edge_p): Use it.
27233 (group_case_labels_stmt): Likewise.
27234 * tree-cfg.h: Prototype it.
27235 * stmt.c: Include cfghooks.h and tree-cfg.h.
27236 (emit_case_dispatch_table) <gap_label>: New local variable.
27237 Use it to fill dispatch table gaps.
27238 Test for default_label before updating probabilities.
27239 (expand_case) <default_label>: Remove unneeded initialization.
27240 Test for unreachable default case statement and remove its edge.
27241 Set default_label accordingly.
27242 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
27244 2017-05-10 Carl Love <cel@us.ibm.com>
27246 * config/rs6000/rs6000-c: Add support for built-in functions
27247 vector signed char vec_neg (vector signed char)
27248 vector signed short int vec_neg (vector short int)
27249 vector signed int vec_neg (vector signed int)
27250 vector signed long long vec_neg (vector signed long long)
27251 vector float vec_neg (vector float)
27252 vector double vec_neg (vector double)
27253 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
27255 * config/rs6000/altivec.h: Add define for vec_neg
27256 * doc/extend.texi: Update the built-in documentation for the
27257 new built-in functions.
27259 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27261 PR tree-optimization/77644
27262 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
27264 2017-05-10 Nathan Sidwell <nathan@acm.org>
27266 * dumpfile.h (TDI_lang_all): New.
27267 (TDF_KIND): New. Renumber others
27268 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
27270 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
27272 (get_dump_file_name): Adjust suffix generation.
27273 (dump_enable_all): Use TDF_KIND.
27274 * doc/invoke.texi (-fdump-lang-all): Document.
27276 * dumpfile.h: Tabify.
27278 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
27281 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
27282 Move member access before delete.
27284 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
27286 * tree-inline.c (expand_call_inline): Split block at stmt
27289 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
27292 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
27293 are now unused after splitting mov{sf,sd}_hardfloat.
27294 (f32_lr2): Likewise.
27295 (f32_lm): Likewise.
27296 (f32_lm2): Likewise.
27297 (f32_li): Likewise.
27298 (f32_li2): Likewise.
27299 (f32_lv): Likewise.
27300 (f32_sr): Likewise.
27301 (f32_sr2): Likewise.
27302 (f32_sm): Likewise.
27303 (f32_sm2): Likewise.
27304 (f32_si): Likewise.
27305 (f32_si2): Likewise.
27306 (f32_sv): Likewise.
27307 (f32_dm): Likewise.
27308 (f32_vsx): Likewise.
27309 (f32_av): Likewise.
27310 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
27311 For movsf, order stores so the VSX stores occur before the GPR
27312 store which encourages the register allocator to use a traditional
27313 FPR instead of a GPR. For movsd, order the stores so that the GPR
27314 store comes before the VSX stores to allow the power6 to work.
27315 This is due to the power6 not having a 32-bit integer store
27316 instruction from a FPR.
27317 (movsf_hardfloat): Likewise.
27318 (movsd_hardfloat): Likewise.
27320 2017-05-09 Martin Sebor <msebor@redhat.com>
27322 PR translation/80280
27323 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
27326 PR translation/80280
27327 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
27328 data member added in r247778.
27329 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
27331 2017-05-09 Nathan Sidwell <nathan@acm.org>
27333 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
27335 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
27338 2017-05-09 Marek Polacek <polacek@redhat.com>
27340 * doc/invoke.texi: Fix typo.
27342 2017-05-09 Richard Biener <rguenther@suse.de>
27344 * tree-vrp.c (vrp_val_is_max): Adjust comment.
27345 (vrp_val_is_min): Likewise.
27346 (set_value_range_to_value): Likewise.
27347 (set_value_range_to_nonnegative): Likewise.
27348 (gimple_assign_nonzero_p): Likewise.
27349 (gimple_stmt_nonzero_p): Likewise.
27350 (vrp_int_const_binop): Likewise. Remove unreachable case.
27351 (adjust_range_with_scev): Adjust comments.
27352 (compare_range_with_value): Likewise.
27353 (extract_range_from_phi_node): Likewise.
27354 (test_for_singularity): Likewise.
27356 2017-05-09 Richard Biener <rguenther@suse.de>
27358 * tree-vrp.c (get_single_symbol): Add assert that we don't
27359 get overflowed constants as invariant part.
27360 (compare_values_warnv): Add comment before the TREE_NO_WARNING
27361 checks. Use wi::cmp instead of recursing for integer constants.
27362 (compare_values): Just ignore whether we assumed undefined
27363 overflow instead of failing the compare.
27364 (extract_range_for_var_from_comparison_expr): Add comment before the
27365 TREE_NO_WARNING sets.
27366 (test_for_singularity): Likewise.
27367 (extract_range_from_comparison): Do not disable optimization
27368 when we assumed undefined overflow.
27369 (extract_range_basic): Remove init of unused var.
27371 2017-05-09 Richard Biener <rguenther@suse.de>
27373 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
27374 (extract_range_from_multiplicative_op_1): Adjust.
27375 (extract_range_from_binary_expr_1): Use int_const_binop.
27377 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
27380 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
27381 rs6000_store_data_bypass_p in seven define_bypass directives and
27382 in several comments.
27383 * config/rs6000/rs6000-protos.h: Add prototype for
27384 rs6000_store_data_bypass_p function.
27385 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
27386 function implements slightly different (rs6000-specific) semantics
27387 than store_data_bypass_p, returning false rather than aborting
27388 with assertion error when arguments do not satisfy the
27389 requirements of store data bypass.
27390 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
27391 rs6000_store_data_bypass_p.
27393 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
27395 * config/xtensa/xtensa-protos.h
27396 (xtensa_initial_elimination_offset): New declaration.
27397 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
27398 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
27399 macro definition, add case for FRAME_POINTER_REGNUM when
27400 FRAME_GROWS_DOWNWARD.
27401 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
27402 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
27403 xtensa_initial_elimination_offset.
27405 2017-05-08 Nathan Sidwell <nathan@acm.org>
27407 * doc/invoke.texi: Alphabetize -fdump options.
27409 2017-05-08 Martin Sebor <msebor@redhat.com>
27411 PR translation/80280
27412 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
27414 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
27416 * target.def (compute_frame_layout): New optional target hook.
27417 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
27418 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
27419 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
27421 * reload1.c (verify_initial_elim_offsets): Likewise.
27422 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
27423 (use_simple_return_p): Call arm_compute_frame_layout if needed.
27424 (arm_get_frame_offsets): Split up into this ...
27425 (arm_compute_frame_layout): ... and this function.
27427 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
27429 * config/aarch64/constraints.md (Usa): New constraint.
27430 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
27432 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
27434 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
27435 with_multilib_list after it has been checked.
27437 2017-05-08 Richard Biener <rguenther@suse.de>
27439 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
27440 (bitmap_set_subtract_values): Likewise.
27442 2017-05-08 Richard Biener <rguenther@suse.de>
27444 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
27445 (gimple_assign_nonzero): ... this and remove strict_overflow_p
27447 (gimple_stmt_nonzero_warnv_p): Rename to ...
27448 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
27450 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
27451 (extract_range_basic): Adjust, do not disable propagation on
27452 strict overflow sensitive simplification.
27453 (vrp_visit_cond_stmt): Likewise.
27455 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
27457 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
27458 body size unconditionally.
27460 2017-05-07 Jeff Law <law@redhat.com>
27463 2017-05-06 Jeff Law <law@redhat.com>
27464 PR tree-optimization/78496
27465 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
27468 PR tree-optimization/78496
27469 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
27470 (simplify_stmt_using_ranges): Call it.
27471 (vrp_dom_walker::before_dom_children): Extract equivalences
27472 from an ASSERT_EXPR with an equality comparison against a
27475 2017-05-06 Jeff Law <law@redhat.com>
27477 PR tree-optimization/78496
27478 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
27481 PR tree-optimization/78496
27482 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
27483 (simplify_stmt_using_ranges): Call it.
27484 (vrp_dom_walker::before_dom_children): Extract equivalences
27485 from an ASSERT_EXPR with an equality comparison against a
27488 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
27490 * lra-constraints.c (lra_copy_reg_equiv): New function.
27491 (split_reg): Use it to copy equivalence information from the
27492 original register to the spill register.
27494 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
27496 PR rtl-optimization/75964
27497 * simplify-rtx.c (simplify_const_relational_operation): Remove
27498 invalid handling of comparisons of integer ABS.
27500 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
27502 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
27503 initialize to zero.
27504 (init_regs): Remove declaration.
27505 (function_arg_advance_32): Initialize error_p as boolean variable.
27507 2017-05-05 Nathan Sidwell <nathan@acm.org>
27509 * store-motion.c (remove_reachable_equiv_notes): Reformat long
27510 lines. Use for (;;).
27512 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27514 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
27515 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
27516 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
27517 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
27518 VF=2 that require versioning.
27520 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27522 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
27525 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27527 * diagnostic.h (diagnostic_override_option_index): Convert from
27528 macro to inline function.
27530 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27532 * diagnostic.c (last_module_changed_p): New function.
27533 (set_last_module): New function.
27534 (diagnostic_report_current_module): Convert macro usage to
27535 the above functions.
27536 * diagnostic.h (diagnostic_context::last_module): Strengthen
27537 from const line_map * to const line_map_ordinary *.
27538 (diagnostic_last_module_changed): Delete macro.
27539 (diagnostic_set_last_module): Delete macro.
27541 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27543 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
27544 with diagnostic_report_diagnostic.
27545 (diagnostic_n_impl_richloc): Likewise.
27546 * diagnostic.h (report_diagnostic): Delete macro.
27547 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
27548 with diagnostic_report_diagnostic.
27549 * substring-locations.c (format_warning_va): Likewise.
27551 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27553 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
27554 save/restor of format_spec. Move option-printing code to...
27555 (print_option_information): ...this new function, and
27556 reimplement by simply printing to the pretty_printer,
27557 rather than appending to the format string.
27559 2017-05-05 David Malcolm <dmalcolm@redhat.com>
27561 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
27562 handling logic into...
27563 (update_effective_level_from_pragmas): ...this new function.
27565 2017-05-04 Andrew Waterman <andrew@sifive.com>
27567 * config/riscv/riscv.opt (mstrict-align): New option.
27568 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
27569 (SLOW_UNALIGNED_ACCESS): Define.
27570 (riscv_slow_unaligned_access): Declare.
27571 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
27573 (riscv_slow_unaligned_access): New variable.
27574 (rocket_tune_info): Set slow_unaligned_access to true.
27575 (optimize_size_tune_info): Set slow_unaligned_access to false.
27576 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
27577 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
27578 (riscv_option_override): Set riscv_slow_unaligned_access.
27579 * doc/invoke.texi: Add -mstrict-align to RISC-V.
27581 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
27583 * config/riscv/riscv.md: Unify indentation.
27585 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27590 * config/rs6000/rs6000.md (u code attribute): Add FIX and
27592 (extendsi<mode>2): Add support for doing sign extension via
27593 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
27594 don't have ISA 3.0 instructions.
27595 (extendsi<mode>2 splitter): Likewise.
27596 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
27597 generate the normal insns since SImode can now go in vector
27598 registers. Disallow the special UNSPECs needed for previous
27599 machines to hide SImode being used. Add new insns
27600 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
27601 (fix_trunc<mode>si2_stfiwx): Likewise.
27602 (fix_trunc<mode>si2_internal): Likewise.
27603 (fixuns_trunc<mode>si2): Likewise.
27604 (fixuns_trunc<mode>si2_stfiwx): Likewise.
27605 (fctiw<u>z_<mode>_smallint): Likewise.
27606 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
27607 of floating point to 32-bit integer from doing a direct move to
27608 the GPR registers to do a store.
27609 (fctiwz_<mode>): Break long line.
27611 2017-05-05 Bin Cheng <bin.cheng@arm.com>
27613 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
27614 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
27615 (addr_list, addr_offset_valid_p): New.
27616 (split_address_groups): Check offset validity with above function.
27617 (gt-tree-ssa-loop-ivopts.h): Include header file.
27619 2017-05-05 Nathan Sidwell <nathan@acm.org>
27621 * config.gcc (arm*-*-*): Add missing 'fi'.
27623 2017-05-05 Steve Ellcey <sellcey@cavium.com>
27625 * doc/invoke.texi (-fopt-info): Explicitly say order of options
27626 included in -fopt-info does not matter.
27627 * doc/optinfo.texi (-fopt-info): Fix description of default
27628 behavour. Explicitly say order of options included in -fopt-info
27631 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
27633 * config.gcc: Allow combinations of aprofile and rmprofile values for
27634 --with-multilib-list.
27635 * config/arm/t-multilib: New file.
27636 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
27637 variables. Remove setting of ISA and floating-point ABI in
27638 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
27639 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
27640 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
27641 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
27643 * config/arm/t-rmprofile: Likewise except for the matches changes.
27644 * doc/install.texi (--with-multilib-list): Document the combination of
27645 aprofile and rmprofile values and warn about pitfalls in doing that.
27647 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
27649 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
27650 (movdi_aarch64): Likewise.
27652 2017-05-05 Jakub Jelinek <jakub@redhat.com>
27654 PR tree-optimization/80632
27655 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
27657 (build_arrays): Initialize it for virtual phis.
27658 (fix_phi_nodes): Use it for virtual phis.
27660 PR tree-optimization/80558
27661 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
27662 [x, y] op z into [x op, y op z] for op & or | if conditions
27665 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
27666 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27669 * config/arm/arm.md (use_literal_pool): Remove.
27670 (64-bit immediate split): No longer takes cost into consideration
27671 if arm_disable_literal_pool is enabled.
27672 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
27673 used when arm_disable_literal_pool is enabled.
27674 (arm_max_const_double_inline_cost): Remove use of
27675 arm_disable_literal_pool.
27676 (push_minipool_fix): Add assert.
27677 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
27678 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
27679 (no_literal_pool_sf_immediate): New.
27681 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27683 PR tree-optimization/80613
27684 * tree-ssa-dce.c (propagate_necessity): Remove cases for
27685 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
27687 2017-05-05 Richard Biener <rguenther@suse.de>
27689 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
27691 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
27693 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
27694 of this flag from insn conditions due to removal from r247495.
27696 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
27698 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
27700 (arm_early_store_addr_dep_ptr): Likewise.
27701 * config/arm/aarch-common-protos.h
27702 (arm_early_load_addr_dep_ptr): Add prototype.
27703 (arm_early_store_addr_dep_ptr): Likewise.
27704 * config/arm/cortex-a53.md: Add new bypasses.
27706 2017-05-05 Jakub Jelinek <jakub@redhat.com>
27708 * tree.c (next_type_uid): Change type to unsigned.
27709 (type_hash_canon): Decrement back next_type_uid if
27710 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
27711 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
27714 2017-05-04 Martin Sebor <msebor@redhat.com>
27716 * builtins.c: Fix a trivial typo in a comment.
27718 PR middle-end/79234
27719 * builtins.c (check_sizes): Adjust to handle reading past the end.
27720 Avoid printing excessive upper bound of ranges. Use %E to print
27721 tree nodes instead of converting them to %wu.
27722 (expand_builtin_memchr): New function.
27723 (compute_dest_size): Rename...
27724 (compute_objsize): ...to this.
27725 (expand_builtin_memcpy): Adjust.
27726 (expand_builtin_mempcpy): Adjust.
27727 (expand_builtin_strcat): Adjust.
27728 (expand_builtin_strcpy): Adjust.
27729 (check_strncat_sizes): Adjust.
27730 (expand_builtin_strncat): Adjust.
27731 (expand_builtin_strncpy): Adjust and simplify.
27732 (expand_builtin_memset): Adjust.
27733 (expand_builtin_bzero): Adjust.
27734 (expand_builtin_memcmp): Adjust.
27735 (expand_builtin): Handle memcmp.
27736 (maybe_emit_chk_warning): Check strncat just once.
27738 2017-05-04 Martin Sebor <msebor@redhat.com>
27740 PR preprocessor/79214
27741 PR middle-end/79222
27742 PR middle-end/79223
27743 * builtins.c (check_sizes): Add inlining context and issue
27744 warnings even when -Wno-system-headers is set.
27745 (check_strncat_sizes): Same.
27746 (expand_builtin_strncat): Same.
27747 (expand_builtin_memmove): New function.
27748 (expand_builtin_stpncpy): Same.
27749 (expand_builtin): Handle memmove and stpncpy.
27751 2017-05-04 Bin Cheng <bin.cheng@arm.com>
27753 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
27754 which is not used any more.
27756 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27758 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
27760 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27762 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
27763 (cortexa53_tunings): Likewise.
27764 (cortexa57_tunings): Likewise.
27765 (cortexa72_tunings): Likewise.
27766 (cortexa73_tunings): Likewise.
27768 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27770 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
27771 Set loop alignment to 8.
27773 2017-05-04 Martin Sebor <msebor@redhat.com>
27775 PR translation/80280
27776 * builtins.c (expand_builtin_object_size): Add missing quoting to
27777 %D and like directives.
27778 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
27779 (hsa_type_for_tree_type): Same.
27780 (verify_function_arguments): Same.
27781 * symtab.c (symbol_table::change_decl_assembler_name): Same.
27782 * varasm.c (get_section): Same.
27785 2017-05-04 Martin Sebor <msebor@redhat.com>
27787 PR translation/80280
27788 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
27790 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
27792 * config/aarch64/aarch64.c (generic_addrcost_table):
27793 Change HI/TI mode setting.
27795 2017-05-04 Martin Jambor <mjambor@suse.cz>
27797 PR tree-optimization/80622
27798 * tree-sra.c (comes_initialized_p): New function.
27799 (build_accesses_from_assign): Only set write lazily when
27800 comes_initialized_p is false.
27801 (analyze_access_subtree): Use comes_initialized_p.
27802 (propagate_subaccesses_across_link): Assert !comes_initialized_p
27803 instead of testing for PARM_DECL.
27805 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27807 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
27808 constraint on operand 0 to allow more general addressing modes.
27809 Adjust output template.
27810 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
27812 * config/aarch64/aarch64-protos.h
27813 (aarch64_address_valid_for_prefetch_p): Declare prototype.
27814 * config/aarch64/constraints.md (Dp): New address constraint.
27815 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
27818 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
27820 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
27821 update use of estimate_ipcp_clone_size_and_time.
27822 (estimate_local_effects): Update use of
27823 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
27824 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
27825 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
27826 Return nonspecialized time.
27828 2017-05-04 Richard Biener <rguenther@suse.de>
27830 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
27831 for the last VUSE which def dominates the PHI. Directly call
27833 (get_continuation_for_phi_1): Remove.
27835 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
27837 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
27838 to explain the use of truncating division. Cap the number of
27839 iterations to the maximum given by nb_iterations_upper_bound,
27842 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
27844 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
27845 * configure: Regenerate.
27846 * config.in: Regenerate.
27847 * config/i386/driver-mingw32.c: new file.
27848 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
27849 * config.host: Link driver-mingw32.o on MinGW host.
27850 * doc/install.texi: Document new --enable-mingw-wildcard configure
27853 2017-05-04 Marek Polacek <polacek@redhat.com>
27855 PR tree-optimization/80612
27856 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
27858 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27859 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
27861 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
27862 (movt splitter): Likewise.
27863 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
27864 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
27865 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
27866 block for Thumb-1 with MOVT.
27867 (thumb2_legitimate_address_p): Move code block ...
27868 (can_avoid_literal_pool_for_label_p): ... into this new function.
27869 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
27871 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
27872 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
27873 "M-profile targets with the MOVT instruction".
27875 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27877 * config/arm/arm-builtins.c (arm_init_builtins): Rename
27878 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
27879 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
27881 2017-05-04 Martin Liska <mliska@suse.cz>
27883 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
27884 variable cond_code.
27886 2017-05-04 Richard Biener <rguenther@suse.de>
27888 * tree.c (array_at_struct_end_p): Handle arrays at struct
27889 end with flexarrays more conservatively. Refactor and treat
27890 arrays of arrays or aggregates more strict. Fix
27891 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
27892 * tree.c (array_at_struct_end_p): Adjust prototype.
27893 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
27894 * gimple-fold.c (get_range_strlen): Likewise.
27895 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
27897 2017-05-04 Richard Biener <rguenther@suse.de>
27899 PR tree-optimization/31130
27900 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
27902 (supports_overflow_infinity): Likewise.
27903 (is_negative_overflow_infinity): Likewise.
27904 (is_positive_overflow_infinity): Likewise.
27905 (is_overflow_infinity): Likewise.
27906 (stmt_overflow_infinity): Likewise.
27907 (overflow_infinity_range_p): Likewise.
27908 (usable_range_p): Remove as always returning true.
27909 (make_overflow_infinity): Remove.
27910 (negative_overflow_infinity): Likewise.
27911 (positive_overflow_infinity): Likewise.
27912 (avoid_overflow_infinity): Likewise.
27913 (set_value_range): Adjust accordingly.
27914 (set_value_range_to_nonnegative): Likewise, remove now unused
27915 overflow_infinity arg.
27916 (vrp_operand_equal_p): Adjust.
27917 (update_value_range): Likewise.
27918 (range_int_cst_singleton_p): Likewise.
27919 (operand_less_p): Likewise.
27920 (compare_values_warnv): Likewise.
27921 (extract_range_for_var_from_comparison_expr): Likewise.
27922 (vrp_int_const_binop): Likewise.
27923 (zero_nonzero_bits_from_vr): Likewise.
27924 (extract_range_from_multiplicative_op_1): Likewise.
27925 (extract_range_from_binary_expr_1): Likewise.
27926 (extract_range_from_unary_expr): Likewise.
27927 (extract_range_from_comparison): Likewise.
27928 (extract_range_basic): Likewise.
27929 (adjust_range_with_scev): Likewise.
27930 (compare_ranges): Likewise.
27931 (compare_range_with_value): Likewise.
27932 (dump_value_range): Likewise.
27933 (test_for_singularity): Likewise, remove strict_overflow_p parameter
27935 (simplify_cond_using_ranges): Adjust.
27937 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
27939 * brig-builtins.def: Added a builtin for class_f64.
27940 * builtin-types.def: Added a builtin type needed by class_f64.
27942 2017-05-03 Jason Merrill <jason@redhat.com>
27944 * timevar.def: Add TV_CONSTEXPR.
27946 2017-05-03 David Malcolm <dmalcolm@redhat.com>
27948 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
27950 2017-05-03 Martin Jambor <mjambor@suse.cz>
27952 * ipa-prop.c (ipa_update_after_lto_read): Removed.
27953 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
27954 * ipa-cp.c (ipcp_propagate_stage): Do not call
27955 ipa_update_after_lto_read.
27956 * ipa-inline.c (ipa_inline): Likewise.
27958 2017-05-03 Martin Jambor <mjambor@suse.cz>
27960 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
27961 tag. Added a default constructor and a destructor.
27962 (ipa_edge_args_sum_t): New class;
27963 (ipa_edge_args_sum): Declare.
27964 (ipa_edge_args_vector): Remove declaration.
27965 (IPA_EDGE_REF): Use ipa_edge_args_sum.
27966 (ipa_free_edge_args_substructures): Remove declaration.
27967 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
27968 (ipa_edge_args_info_available_for_edge_p): Likewise.
27969 * ipa-prop.c (ipa_edge_args_vector): Removed.
27970 (edge_removal_hook_holder): Likewise.
27971 (edge_duplication_hook_holder): Likewise.
27972 (ipa_edge_args_sum): New variable.
27973 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
27974 ipa_edge_args_vector.
27975 (ipa_free_edge_args_substructures): Likewise.
27976 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
27977 ipa_edge_args_vector.
27978 (ipa_edge_removal_hook): Turned into method
27979 ipa_edge_args_sum_t::remove.
27980 (ipa_edge_duplication_hook): Turned into method
27981 ipa_edge_args_sum_t::duplicate.
27982 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
27983 registering edge hooks.
27984 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
27985 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
27986 ipa_edge_args_sum instead of ipa_edge_args_vector.
27987 * ipa-profile.c (ipa_profile): Likewise.
27989 2017-05-03 Martin Jambor <mjambor@suse.cz>
27991 * symbol-summary.h (function_summary): New method exists.
27992 (function_summary::symtab_removal): Deallocate through release.
27993 (call_summary): New class.
27994 (gt_ggc_mx): New overload.
27995 (gt_pch_nx): Likewise.
27996 (gt_pch_nx): Likewise.
27998 2017-05-03 Jeff Law <law@redhat.com>
28000 PR tree-optimization/78496
28001 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
28002 from simplify_cond_using_ranges. Split off code to walk
28003 backwards through casts into ...
28004 (simplify_cond_using_ranges_2): New function.
28005 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
28006 (execute_vrp): After identifying jump threads, call
28007 simplify_cond_using_ranges_2.
28009 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
28012 * ipa-inline.h (inline_summary): Add ctor.
28013 (create_ggc): Do not use ggc_cleared_alloc.
28015 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
28016 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28018 * gcc.c (handle_braces): Support escaping in switch matching
28020 * doc/invoke.texi (Spec Files): Document it.
28021 Remove superfluous @code markup in items.
28023 2017-05-03 David Malcolm <dmalcolm@redhat.com>
28025 * diagnostic-show-locus.c (struct column_range): New struct.
28026 (get_affected_columns): New function.
28027 (get_printed_columns): New function.
28028 (struct correction): New struct.
28029 (correction::ensure_capacity): New function.
28030 (correction::ensure_terminated): New function.
28031 (struct line_corrections): New struct.
28032 (line_corrections::~line_corrections): New dtor.
28033 (line_corrections::add_hint): New function.
28034 (layout::print_trailing_fixits): Reimplement in terms of the new
28036 (selftest::test_overlapped_fixit_printing): New function.
28037 (selftest::diagnostic_show_locus_c_tests): Call it.
28039 2017-05-03 Nathan Sidwell <nathan@acm.org>
28041 Canonicalize canonical type hashing
28042 * tree.h (type_hash_canon_hash): Declare.
28043 * tree.c (type_hash_list, attribute_hash_list): Move into
28044 type_hash_canon_hash.
28045 (build_type_attribute_qual_variant): Break out hash code calc into
28046 type_hash_canon_hash.
28047 (type_hash_canon_hash): New. Generic type hash computation.
28048 (build_range_type_1, build_array_type_1, build_function_type,
28049 build_method_type_directly, build_offset_type, build_complex_type,
28050 make_vector_type): Call it.
28052 2017-05-03 Richard Biener <rguenther@suse.de>
28054 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
28055 When all DRs have unknown misaligned do not always peel
28056 when there is a store but apply the same costing model as if
28057 there were only loads.
28059 2017-05-03 Richard Biener <rguenther@suse.de>
28062 PR tree-optimization/80492
28063 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
28064 compare_base_decls returning dont-know properly.
28066 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
28068 * config/arm/iterators.md (CCSI): New mode iterator.
28069 (arch): New mode attribute.
28070 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
28071 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
28072 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
28073 code iterator for success result mode.
28074 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
28075 the corresponding new insn generators.
28077 2017-05-03 Bin Cheng <bin.cheng@arm.com>
28080 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28081 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28083 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
28085 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
28086 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
28087 (DDR_A): Wrap DDR argument in brackets.
28088 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
28089 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
28090 (DDR_REVERSED_P): Likewise.
28092 2017-05-03 Jakub Jelinek <jakub@redhat.com>
28094 PR tree-optimization/79472
28095 * tree-switch-conversion.c (struct switch_conv_info): Add
28096 contiguous_range and default_case_nonstandard fields.
28097 (collect_switch_conv_info): Compute contiguous_range and
28098 default_case_nonstandard fields, don't clear final_bb if
28099 contiguous_range and only the default case doesn't have the required
28101 (check_all_empty_except_final): Set default_case_nonstandard instead
28102 of failing if contiguous_range and the default case doesn't have empty
28104 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
28105 and only the default case doesn't have the required constants. Skip
28107 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
28108 if default_case_nonstandard.
28109 (build_constructors): Build constant 1 just once. Assert that default
28110 values aren't inserted in between cases if contiguous_range. Skip
28112 (build_arrays): Skip virtual phis.
28113 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
28114 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
28115 Handle virtual phis.
28116 (gen_inbound_check): Handle default_case_nonstandard case.
28117 (process_switch): Adjust check_final_bb caller. Call
28118 gather_default_values with the first non-default case instead of
28119 default case if default_case_nonstandard.
28121 2017-05-02 Nathan Sidwell <nathan@acm.org>
28123 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
28124 check. Fix formatting.
28126 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
28128 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
28129 errors when comparing specialized and unspecialized times.
28131 2017-05-02 David Malcolm <dmalcolm@redhat.com>
28133 * diagnostic-show-locus.c
28134 (layout::should_print_annotation_line_p): Make private.
28135 (layout::print_annotation_line): Make private.
28136 (layout::annotation_line_showed_range_p): Make private.
28137 (layout::show_ruler): Make private.
28138 (layout::print_source_line): Make private. Pass in line and
28139 line_width, rather than calling location_get_source_line. Drop
28141 (layout::print_leading_fixits): New method.
28142 (layout::print_any_fixits): Rename to...
28143 (layout::print_trailing_fixits): ...this, and make private.
28144 Don't print newline fixits.
28145 (diagnostic_show_locus): Move logic for printing one row into...
28146 (layout::print_line): ...this new function. Move the
28147 location_get_source_line call and error-handling from
28148 print_source_line to here. Call print_leading_fixits, and rename
28149 print_any_fixits to print_trailing_fixits.
28150 (selftest::test_fixit_insert_containing_newline): Update now that
28151 newlines are partially supported.
28152 (selftest::test_fixit_insert_containing_newline_2): New test.
28153 (selftest::test_fixit_replace_containing_newline): Update comments.
28154 (selftest::diagnostic_show_locus_c_tests): Call the new test.
28155 * edit-context.c (class added_line): New class.
28156 (class edited_line): Describe newline handling in comment.
28157 (edited_line::actually_edited_p): New method.
28158 (edited_line::print_content): Delete redundant decl.
28159 (edited_line::m_predecessors): New field.
28160 (edited_file::print_content): Call edited_line::print_content.
28161 (edited_file::print_diff): Update to support newlines.
28162 (edited_file::print_diff_hunk): Likewise.
28163 (edited_file::print_run_of_changed_lines): New function.
28164 (edited_file::print_diff_line): Convert to...
28165 (print_diff_line): ...this.
28166 (edited_file::get_effective_line_count): New function.
28167 (edited_line::edited_line): Initialize new field m_predecessors.
28168 (edited_line::~edited_line): Clean up m_predecessors.
28169 (edited_line::apply_fixit): Handle newlines.
28170 (edited_line::get_effective_line_count): New function.
28171 (edited_line::print_content): New function.
28172 (edited_line::print_diff_lines): New function.
28173 (selftest::test_applying_fixits_insert_containing_newline): New
28175 (selftest::test_applying_fixits_replace_containing_newline): New
28177 (selftest::insert_line): New function.
28178 (selftest::test_applying_fixits_multiple_lines): Add example of
28180 (selftest::edit_context_c_tests): Call the new tests.
28182 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28184 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
28185 parameter cand. Update dump information.
28186 (get_computation_cost): Update uses.
28188 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28190 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
28191 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
28192 (get_computation_at, rewrite_use_address): Update use of
28193 get_computation_aff.
28195 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28197 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
28198 (get_computation): Delete.
28199 (get_computation_cost): Implement like get_computation_cost_at.
28200 Use get_computation_at.
28201 (get_computation_cost_at): Delete.
28202 (rewrite_use_nonlinear_expr): Use get_computation_at.
28203 (rewrite_use_compare, remove_unused_ivs): Ditto.
28205 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28207 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
28209 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28211 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
28212 (ivopts_global_cost_for_size): Rename parameter and update uses.
28213 (iv_ca_recount_cost): Update uses.
28214 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
28215 candidates seperately in n_invs and n_cands.
28216 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
28218 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28220 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
28221 (find_inv_vars_cb): New.
28222 (find_depends): Renamed to ...
28223 (find_inv_vars): ... this.
28224 (add_candidate_1, force_var_cost): Call find_inv_vars.
28225 (split_address_cost, determine_group_iv_cost_cond): Ditto.
28227 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28229 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
28230 inv_vars. Add inv_exprs.
28231 (struct iv_cand): Rename depends_on to inv_vars.
28232 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
28233 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
28234 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
28235 (dump_cand): Dump inv_vars.
28236 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
28237 (record_invariant, find_depends, add_candidate_1): Ditto.
28238 (set_group_iv_cost, force_var_cost): Ditto.
28239 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
28240 (get_computation_cost_at, get_computation_cost): Ditto.
28241 (determine_group_iv_cost_generic): Ditto.
28242 (determine_group_iv_cost_address): Ditto.
28243 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
28244 (determine_group_iv_costs): Ditto.
28245 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
28246 (iv_ca_set_remove_invariants): Renamed to ...
28247 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
28248 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
28249 (iv_ca_set_add_invariants): Renamed to ...
28250 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
28251 (iv_ca_set_cp): Use iv_ca_set_add_invs.
28252 (iv_ca_has_deps): Support inv_vars and inv_exprs.
28253 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
28254 (create_new_ivs): Remove useless dump.
28256 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28258 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
28260 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
28261 (iv_ca_set_no_cp, create_new_iv): Ditto.
28263 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28265 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28267 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
28269 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
28270 function tree_check2.
28272 2017-05-02 Martin Liska <mliska@suse.cz>
28274 * doc/gcov.texi: Add missing preposition.
28275 * gcov.c (function_info::function_info): Properly fill up
28276 all member variables.
28278 2017-05-02 Tamar Christina <tamar.christina@arm.com>
28280 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
28282 2017-05-02 Tamar Christina <tamar.christina@arm.com>
28284 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
28286 2017-05-02 Martin Liska <mliska@suse.cz>
28289 * lto-streamer-in.c (lto_read_tree_1): Remove
28290 LTO_STREAMER_DEBUG.
28291 * lto-streamer.c (struct tree_hash_entry): Likewise.
28292 (struct tree_entry_hasher): Likewise.
28293 (tree_entry_hasher::hash): Likewise.
28294 (tree_entry_hasher::equal): Likewise.
28295 (lto_streamer_init): Likewise.
28296 (lto_orig_address_map): Likewise.
28297 (lto_orig_address_get): Likewise.
28298 (lto_orig_address_remove): Likewise.
28299 * lto-streamer.h: Likewise.
28300 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
28301 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
28303 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
28305 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
28306 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
28307 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
28308 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
28309 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
28310 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
28311 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
28312 (mm_maskz_sub_ss): New intrinsics.
28313 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28314 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
28315 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
28316 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
28317 (__builtin_ia32_subss_mask_round): New builtins.
28318 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28319 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28320 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
28322 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
28323 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
28325 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
28328 2017-05-02 Martin Jambor <mjambor@suse.cz>
28330 PR tree-optimization/78687
28331 * tree-sra.c (access): New field parent.
28332 (process_subtree_disqualification): New function.
28333 (disqualify_candidate): Call it.
28334 (build_accesses_from_assign): Reset write flag if creating an
28336 (build_access_subtree): Fill in parent field and also prpagate
28337 down grp_write flag.
28338 (create_artificial_child_access): New parameter set_grp_write, set
28339 grp_write to its value.
28340 (propagate_subaccesses_across_link): Also propagate grp_write flag
28342 (propagate_all_subaccesses): Push the closest parent back to work
28343 queue if add_access_to_work_queue returned true.
28345 2017-05-02 Richard Biener <rguenther@suse.de>
28347 * common.opt (fstrict-overflow): Alias negative to fwrapv.
28348 * doc/invoke.texi (fstrict-overflow): Remove all traces of
28349 -fstrict-overflow documentation.
28350 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
28351 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
28352 flag_strict_overflow.
28353 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
28354 * lto-opts.c (lto_write_options): Do not stream it.
28355 * lto-wrapper.c (merge_and_complain): Do not handle it.
28356 * opts.c (default_options_table): Do not set -fstrict-overflow.
28357 (finish_options): Likewise do not clear it when sanitizing.
28358 * simplify-rtx.c (simplify_const_relational_operation): Do not
28359 test flag_strict_overflow.
28361 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
28363 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
28364 using enabled attribute.
28365 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
28366 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
28367 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
28368 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
28369 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
28370 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
28371 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
28372 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
28373 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
28374 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
28376 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
28378 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
28380 2017-05-02 Richard Biener <rguenther@suse.de>
28382 PR tree-optimization/80591
28384 2017-04-10 Richard Biener <rguenther@suse.de>
28386 * tree-ssa-structalias.c (find_func_aliases): Properly handle
28389 2017-05-02 Richard Biener <rguenther@suse.de>
28391 PR tree-optimization/80549
28392 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
28393 (cleanup_tree_cfg_noloop): Create forwarders to known loop
28394 headers if they do not have a preheader.
28396 2017-05-02 Martin Liska <mliska@suse.cz>
28399 * common.opt: Fix typo.
28400 * doc/invoke.texi: Likewise.
28402 2017-05-01 Jan Beulich <jbeulich@suse.com>
28404 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
28405 swapping, add (x,x,m,x,n) alternative.
28407 2017-05-01 Nathan Sidwell <nathan@acm.org>
28409 * calls.c (combine_pending_stack_adjustment_and_call): Remove
28410 unnecessary unadjusted_alignment check.
28412 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
28415 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
28417 * gimplify.c (gimplify_cilk_detach): New function.
28418 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
28419 * tree-core.h: Document EXPR_CILK_SPAWN.
28420 * tree.h (EXPR_CILK_SPAWN): Define.
28422 2017-05-01 David Malcolm <dmalcolm@redhat.com>
28424 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
28425 to use new fixit_hint representation, using the "replace" logic.
28426 (get_line_span_for_fixit_hint): Likewise.
28427 (layout::print_any_fixits): Likewise.
28428 (selftest::test_one_liner_many_fixits): Rename to...
28429 (selftest::test_one_liner_many_fixits_1): ...this, and update
28430 comment and expected output to reflect that the multiple fix-it
28431 hints are now consolidated into one insertion.
28432 (selftest::test_one_liner_many_fixits_2): New test.
28433 (selftest::test_diagnostic_show_locus_one_liner): Update for
28435 (selftest::test_fixit_consolidation): Update for fix-it API
28437 * diagnostic.c (print_parseable_fixits): Likewise.
28438 * edit-context.c (edited_line::m_line_events): Convert from
28439 auto_vec <line_event *> to auto_vec <line_event>.
28440 (class line_event): Convert from abstract base class to a concrete
28441 class, taking over the role of replace_event.
28442 (class insert_event): Delete.
28443 (class replace_event): Rename to class line_event. Convert to
28445 (edit_context::add_fixits): Reimplement.
28446 (edit_context::apply_insert): Delete.
28447 (edit_context::apply_replace): Rename to...
28448 (edit_context::apply_fixit): ...this. Convert to half-open range.
28449 (edited_file::apply_insert): Delete.
28450 (edited_file::apply_replace): Rename to...
28451 (edited_file::apply_fixit): ...this.
28452 (edited_line::~edited_line): Drop deletion of events.
28453 (edited_line::apply_insert): Delete.
28454 (edited_line::apply_replace): Rename to...
28455 (edited_line::apply_fixit): ...this. Convert to half-open range.
28456 Update for change to type of m_line_events.
28457 * edit-context.h (edit_context::apply_insert): Delete.
28458 (edit_context::apply_replace): Rename to...
28459 (edit_context::apply_fixit): ...this.
28461 2017-05-01 Martin Sebor <msebor@redhat.com>
28463 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
28466 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
28469 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
28470 __get_cpuid_max returns 0.
28471 (__get_cpuid_count): Ditto.
28473 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
28475 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
28476 replacement expression is another instance of one of its arguments.
28478 2017-05-01 Jakub Jelinek <jakub@redhat.com>
28481 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
28482 check for stack push/pop autoinc.
28483 * config/i386/i386.c (ix86_agi_dependent): Return false
28484 if the only reason why modified_in_p returned true is that
28485 addr is SP based and set_insn is a push or pop.
28487 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
28489 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
28492 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
28495 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
28496 (account_size_time): Use two predicates - exec_pred and
28498 (evaluate_conditions_for_known_args): Compute both clause and
28500 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
28501 (inline_summary_t::duplicate): Update.
28502 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
28504 (compute_inline_parameters): Likewise.
28505 (estimate_edge_size_and_time): Update caluclation of time.
28506 (estimate_node_size_and_time): Compute both time and nonspecialized
28508 (estimate_ipcp_clone_size_and_time): Update.
28509 (inline_merge_summary): Update.
28510 (do_estimate_edge_time): Update.
28511 (do_estimate_edge_size): Update.
28512 (do_estimate_edge_hints): Update.
28513 (inline_read_section, inline_write_summary): Stream both new predicates.
28514 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
28516 (compute_inlined_call_time): Cleanup.
28517 (big_speedup_p): Update.
28518 (edge_badness): Update.
28519 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
28520 (size_time_entry): Replace predicate by exec_predicate and
28521 nonconst_predicate.
28522 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
28523 (estimate_edge_time): Return also nonspec_time.
28524 (reset_edge_growth_cache): Update.
28526 2017-04-29 Jakub Jelinek <jakub@redhat.com>
28528 PR rtl-optimization/80491
28529 * ifcvt.c (noce_process_if_block): When looking for x setter
28530 with missing else_bb, don't check only the insn right before
28531 cond_earliest, but look for the last insn that x is modified in
28532 within the same bb.
28534 PR rtl-optimization/80491
28535 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
28537 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
28539 PR tree-optimization/80487
28540 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
28542 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28544 PR tree-optimization/79697
28545 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
28546 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
28547 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
28549 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
28550 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
28552 2017-04-28 Martin Sebor <msebor@redhat.com>
28554 PR tree-optimization/80523
28555 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
28556 (init_target_to_host_charmap, target_to_host, target_strtol10): New
28558 (maybe_warn, format_directive, parse_directive): Use new functions.
28559 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
28561 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
28563 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
28565 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
28567 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
28568 target_header_dir): Set correctly.
28569 * configure: Regenerated.
28570 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
28571 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
28572 instead of SYSTEM_HEADER_DIR.
28574 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
28576 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
28577 (estimate_local_effects): Likewise.
28578 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
28579 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
28580 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
28581 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
28582 do_estimate_edge_time, estimate_edge_time): Likewise.
28583 * ipa-inline-analysis.c (estimate_node_size_and_time,
28584 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
28585 (estimate_time_after_inlining): Remove.
28587 2017-04-28 Martin Liska <mliska@suse.cz>
28589 * doc/gcov.texi: Enhance documentation of gcov.
28591 2017-04-28 Martin Liska <mliska@suse.cz>
28593 * doc/gcov.texi: Sort options in alphabetic order.
28594 * doc/gcov-dump.texi: Likewise.
28595 * doc/gcov-tool.texi: Likewise.
28596 * gcov.c (print_usage): Likewise.
28597 * gcov-dump.c (print_usage): Likewise.
28598 * gcov-tool.c (print_merge_usage_message): Likewise.
28599 (print_rewrite_usage_message): Likewise.
28600 (print_overlap_usage_message): Likewise.
28602 2017-04-28 Martin Liska <mliska@suse.cz>
28604 PR gcov-profile/53915
28605 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
28607 2017-04-28 Martin Liska <mliska@suse.cz>
28609 PR gcov-profile/79891
28610 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
28611 is marked by compiler as living on a line.
28612 (get_cycles_count): Remove usage of the union.
28613 (output_intermediate_file): Likewise.
28614 (find_source): Fix GNU coding style.
28615 (accumulate_line_counts): Remove old non-all block mode.
28616 (output_lines): Remove usage of the union.
28617 * profile.c (output_location): Include all BBs, even if
28618 belonging to a same line (and file) as a previous BB.
28620 2017-04-28 Martin Liska <mliska@suse.cz>
28622 * gcov.c (process_args): Handle new argument 'w'.
28623 (read_graph_file): Assign ID to BBs.
28624 (output_branch_count): Display BB # if verbose flag is set.
28625 (output_lines): Likewise for arcs.
28626 (print_usage): Add '--verbose' option help.
28627 * doc/gcov.texi: Document --verbose (-w) option.
28629 2017-04-28 Martin Liska <mliska@suse.cz>
28631 * gcov.c (struct block_location_info): New struct.
28632 (process_file): Fill up the new structure.
28633 (read_graph_file): Replace usage of encoding by the newly added
28635 (add_line_counts): Likewise.
28636 (accumulate_line_counts): Remove usage of the union.
28637 (function_info::function_info): New function.
28638 (function_info::~function_info): Likewise.
28639 (process_file): Call delete instead of release_function.
28640 (release_function): Release the function.
28641 (release_structures): Call delete instead of release_function.
28642 (solve_flow_graph): Replace usage of num_blocks.
28643 (find_exception_blocks): Likewise.
28644 (output_lines): Fix GNU coding style.
28646 2017-04-28 Martin Liska <mliska@suse.cz>
28649 * coverage.c (coverage_remove_note_file): New function.
28650 * coverage.h: Declare the function.
28651 * toplev.c (finalize): Clean if an error has been seen.
28653 2017-04-28 Martin Liska <mliska@suse.cz>
28655 PR gcov-profile/80031
28656 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
28657 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
28658 * gcov.c (read_graph_file): Read just number of blocks.
28659 * profile.c (branch_prob): Do not stream 0 flags per a basic
28662 2017-04-28 Martin Liska <mliska@suse.cz>
28664 * gcov-dump.c (tag_*): Add new argument to declarations.
28665 (dump_gcov_file): Likewise.
28666 (tag_blocks): Add and use new argument depth.
28667 (tag_arcs): Likewise.
28668 (tag_lines): Likewise.
28669 (tag_counters): Likewise.
28670 (tag_summary): Likewise.
28671 (dump_working_sets): Use depth to do a proper indentation.
28673 2017-04-28 Jakub Jelinek <jakub@redhat.com>
28676 * cgraph.h (symtab_node::debug_symtab): No longer inline.
28677 * symtab.c (symtab_node::debug_symtab): Move definition here.
28679 2017-04-28 Richard Biener <rguenther@suse.de>
28681 * lto-streamer.h (LTO_major_version): Bump to 7.
28683 2017-04-28 Richard Biener <rguenther@suse.de>
28685 * tree-vrp.c (assert_info): New struct.
28686 (add_assert_info): New helper.
28687 (register_edge_assert_for_2): Refactor to add asserts to a vector
28689 (register_edge_assert_for_1): Likewise.
28690 (register_edge_assert_for): Likewise.
28691 (finish_register_edge_assert_for): New helper actually registering
28692 asserts where live on edge.
28693 (find_conditional_asserts): Adjust.
28694 (find_switch_asserts): Likewise.
28695 (evrp_dom_walker::try_find_new_range): Generalize.
28696 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
28698 2017-04-27 Marek Polacek <polacek@redhat.com>
28701 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
28702 arg10 and arg11 to itype.
28704 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
28706 * doc/extend.texi (Object Size Checking): Improve grammar.
28708 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
28711 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
28712 that the logic for permitting reciprocal estimates matches that
28715 2017-04-27 Jakub Jelinek <jakub@redhat.com>
28718 * tree.c (type_cache_hasher::equal): Only compare
28719 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
28720 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
28721 non-aggregate element types.
28722 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
28723 about the flag on ARRAY_TYPEs in the comment, formatting fix.
28725 2017-04-27 Richard Biener <rguenther@suse.de>
28727 PR middle-end/80533
28728 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
28729 stripping ARRAY_REFs from MEM_EXPR make sure we're not
28730 keeping a reference to a trailing array.
28732 2017-04-27 Richard Biener <rguenther@suse.de>
28734 PR middle-end/80539
28735 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
28736 being in loop-closed SSA form conservatively.
28737 (chrec_fold_multiply_poly_poly): Likewise.
28739 2017-04-27 Tamar Christina <tamar.christina@arm.com>
28741 PR middle-end/79665
28742 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
28743 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
28745 2017-04-27 Jakub Jelinek <jakub@redhat.com>
28748 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
28749 (aarch64_function_arg_alignment): Return unsigned int again, but still
28750 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
28751 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
28752 Don't emit -Wpsabi note.
28753 (aarch64_function_arg_boundary): Likewise.
28754 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
28757 2017-04-26 Nathan Sidwell <nathan@acm.org>
28759 * tree.h (crc32_unsigned_n): Declare.
28760 (crc32_unsigned, crc32_unsigned): Make inline.
28761 * tree.c (crc32_unsigned_bits): Replace with ...
28762 (crc32_unsigned_n): ... this.
28763 (crc32_unsigned, crc32_byte): Remove.
28764 (crc32_string): Remove unnecessary braces.
28766 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
28768 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
28769 * ipa-inline-analysis.c (MAX_TIME): Remove.
28770 (account_size_time): Use sreal for time.
28771 (dump_inline_summary): Update.
28772 (estimate_function_body_sizes): Update.
28773 (estimate_edge_size_and_time): Update.
28774 (estimate_calls_size_and_time): Update.
28775 (estimate_node_size_and_time): Update.
28776 (inline_merge_summary): Update.
28777 (inline_update_overall_summary): Update.
28778 (estimate_time_after_inlining): Update.
28779 (inline_read_section): Update.
28780 (inline_write_summary): Update.
28781 * ipa-inline.c (compute_uninlined_call_time): Update.
28782 (compute_inlined_call_time): Update.
28783 (recursive_inlining): Update.
28784 (inline_small_functions): Update.
28785 (dump_overall_stats): Update.
28786 * ipa-inline.h: Include sreal.h.
28787 (size_time_entry): Turn time to sreal.
28788 (inline_summary): Turn self_time nad time to sreal.
28790 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
28792 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
28794 (sreal::stream_out, sreal::stream_in): New.
28795 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
28797 2017-04-25 Jakub Jelinek <jakub@redhat.com>
28799 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
28802 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
28805 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
28807 (dimode_scalar_chain::compute_convert_gain): Ditto.
28808 (dimode_scalar_chain::make_vector_copies): Ditto.
28809 (dimode_scalar_chain::convert_reg): Ditto.
28810 (dimode_scalar_chain::convert_insn): Ditto.
28811 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
28812 (VI248_AVX512BW_1): New mode iterator.
28813 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
28814 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
28817 2017-04-25 Martin Sebor <msebor@redhat.com>
28819 PR tree-optimization/80497
28820 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
28821 constants are representable in HOST_WIDE_INT.
28822 (parse_directive): Ditto.
28824 2017-04-25 Martin Sebor <msebor@redhat.com>
28827 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
28828 (new_zero_array): Adjust signature.
28829 (dom_info::dom_init): Used unsigned rather that size_t.
28830 (dom_info::dom_info): Same.
28832 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28833 Jakub Jelinek <jakub@redhat.com>
28836 * config/arm/arm.c: Include gimple.h.
28837 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
28838 returns negative, increment ncrn only if it returned positive.
28839 (arm_needs_doubleword_align): Return int instead of bool,
28840 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
28841 members, but if there is any such non-FIELD_DECL
28842 > PARM_BOUNDARY aligned decl, return -1 instead of false.
28843 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
28844 returns negative, increment nregs only if it returned positive.
28845 (arm_setup_incoming_varargs): Likewise.
28846 (arm_function_arg_boundary): Emit -Wpsabi note if
28847 arm_needs_doubleword_align returns negative, return
28848 DOUBLEWORD_ALIGNMENT only if it returned positive.
28850 2017-04-25 Marek Polacek <polacek@redhat.com>
28853 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
28854 first argument to type.
28856 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
28859 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
28860 type checks to test for compatibility instead of equality.
28862 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28863 Jakub Jelinek <jakub@redhat.com>
28866 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
28868 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
28869 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
28870 the alignment computation, but return their maximum in warn_alignment.
28871 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
28872 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
28874 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
28875 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
28878 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28880 * config/arc/simdext.md (dmpyh): Fix typo.
28882 2017-04-25 Richard Biener <rguenther@suse.de>
28884 PR tree-optimization/80492
28885 * alias.c (compare_base_decls): Handle registers with asm
28886 specification conservatively.
28887 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
28888 compare_base_decls returning dont-know properly.
28890 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28892 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
28893 (legitimate_offset_address_p): New function.
28894 (arc_legitimate_address_p): Use above function.
28896 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28898 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
28900 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28902 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
28903 ACCH registers whenever they are available.
28905 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28907 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
28908 double regs fix when not used.
28910 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28912 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
28914 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
28915 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
28917 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28919 * config/arc/arc.c (arc_output_addsi): Check for h-register class
28920 when emitting short ADD instructions.
28922 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
28924 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
28926 (cmpsi_cc_c_insn): Likewise.
28927 (cbranchsi4_scratch): Compute proper instruction length using
28928 compact_hreg_operand.
28929 * config/arc/predicates.md (compact_hreg_operand): New predicate.
28931 2017-04-25 Richard Biener <rguenther@suse.de>
28933 PR middle-end/80509
28934 * passes.c (pass_manager::pass_manager): Initialize
28935 m_name_to_pass_map.
28937 2017-04-25 Richard Biener <rguenther@suse.de>
28939 PR tree-optimization/79201
28940 * tree-ssa-sink.c (statement_sink_location): Handle calls.
28942 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28945 * config/s390/vector.md: Split MEM->GPR vector moves for
28946 non-s_operand addresses.
28948 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28951 * config/s390/predicates.md (reload_const_wide_int_operand): New
28953 * config/s390/s390.md ("movti"): Remove d/P alternative.
28954 ("movti_bigconst"): New pattern definition.
28956 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
28959 * s390-protos.h (s390_expand_cs_hqi): Removed.
28960 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
28961 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
28962 modes as well as CCZ1mode and CCZmode.
28963 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
28964 signature of s390_emit_compare_and_swap.
28965 (s390_expand_cs_hqi): Likewise, make static.
28966 (s390_expand_cs_tdsi): Generate an explicit compare before trying
28967 compare-and-swap, in some cases.
28968 (s390_expand_cs): Wrapper function.
28969 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
28971 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
28972 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
28973 patterns for small and large integers. Forbid symref memory operands.
28974 Move expander to s390.c. Require cc register.
28975 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
28976 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
28977 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
28978 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
28979 symref memory operands. Remove CC mode and call s390_match_ccmode
28981 ("atomic_exchange<mode>"): Allow and implement all integer modes.
28983 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
28985 * config/s390/s390.md (define_peephole2): New peephole to help
28986 combining the load-and-test pattern with volatile memory.
28988 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
28990 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
28991 with CCZmode for TARGET_Z196.
28993 2017-04-25 Jakub Jelinek <jakub@redhat.com>
28995 PR rtl-optimization/80501
28996 * combine.c (make_compound_operation_int): Set subreg_code to SET
28997 even for AND with mask of the sign bit of mode.
28999 PR rtl-optimization/80500
29000 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
29001 sum's initial value.
29003 2017-04-25 Julian Brown <julian@codesourcery.com>
29004 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29006 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
29008 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
29010 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
29012 2017-04-25 Julian Brown <julian@codesourcery.com>
29013 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29015 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
29016 (thunderx2t99_sha): New Reservation.
29018 2017-04-25 Julian Brown <julian@codesourcery.com>
29019 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29021 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
29022 type for 1-element load.
29024 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
29026 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
29028 2017-04-24 Martin Jambor <mjambor@suse.cz>
29030 PR tree-optimization/80293
29031 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
29032 char arrays not totally scalarizable if it is false.
29033 (analyze_all_variable_accesses): Pass correct value in the new
29034 parameter. Add a statistics counter.
29036 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
29038 PR middle-end/79931
29039 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
29041 2017-04-24 Richard Biener <rguenther@suse.de>
29043 PR tree-optimization/80494
29044 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
29045 out for complex types.
29047 2017-04-24 Richard Biener <rguenther@suse.de>
29049 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
29050 * tree-ssa-sccvn.c (print_scc): Print SCC size.
29051 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
29052 (DFS): Adjust and never fail.
29053 (sccvn_dom_walker::fail): Remove.
29054 (sccvn_dom_walker::before_dom_children): Adjust.
29055 (run_scc_vn): Likewise and never fail.
29056 * tree-ssa-pre.c (pass_pre::execute): Adjust.
29057 (pass_fre::execute): Likewise.
29059 2017-04-24 Richard Biener <rguenther@suse.de>
29061 PR tree-optimization/79725
29062 * tree-ssa-sink.c (statement_sink_location): Return whether
29063 failure reason was zero uses. Move that check later.
29064 (sink_code_in_bb): Deal with zero uses by removing the stmt
29067 2017-04-24 Richard Biener <rguenther@suse.de>
29070 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
29071 pointer-based references.
29073 2017-04-24 Richard Biener <rguenther@suse.de>
29076 * pass_manager.h (pass_manager::operator new): Remove.
29077 (pass_manager::operator delete): Likewise.
29078 * passes.c (pass_manager::operator new): Remove.
29079 (pass_manager::operator delete): Likewise.
29080 (pass_manager::pass_manager): Zero individual pass members.
29082 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
29085 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
29086 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
29087 Check "XEXP (src, 1)" operand here.
29088 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
29089 Check "XEXP (src, 1)" operand here.
29090 (dimode_scalar_chain::make_vector_copies): Detect count register
29091 of a shift instruction. Zero extend count register from QImode
29092 to DImode to satisfy vector shift pattern count operand predicate.
29093 Substitute vector shift count operand with a DImode copy.
29094 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
29097 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
29099 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
29100 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
29101 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
29102 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
29103 (UNSPEC_NOREX_MEM): Remove definition.
29105 2017-04-21 Richard Biener <rguenther@suse.de>
29107 PR tree-optimization/79547
29108 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
29109 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
29110 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
29111 without any constraints.
29113 2017-04-21 Richard Biener <rguenther@suse.de>
29115 PR tree-optimization/78847
29116 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
29118 2017-04-21 Richard Biener <rguenther@suse.de>
29120 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
29121 (build_distinct_type_copy): Likewise.
29122 (build_variant_type_copy): Likewise.
29123 * tree.c (build_qualified_type): Pass down mem-stat info.
29124 (build_distinct_type_copy): Likewise.
29125 (build_variant_type_copy): Likewise.
29127 2017-04-21 Richard Biener <rguenther@suse.de>
29129 PR tree-optimization/80237
29130 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
29132 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
29133 for a simplified result.
29135 2016-04-21 Richard Biener <rguenther@suse.de>
29137 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
29138 sth as strict as a simple_iv but a chrec without symbols and an
29139 operand defined in the loop we are peeling (and not some subloop).
29140 (propagate_constants_for_unrolling): Propagate all constants.
29142 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
29145 * config/i386/i386.c (print_reg): Remove assert for disalowed
29146 regno values, call output_operand_lossage instead.
29148 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
29151 * config/i386/constraints.md (Yc): New register constraint.
29152 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
29153 Use Yc constraint for alternative 2 of operand 0. Remove
29154 preferred_for_speed attribute.
29156 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
29158 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
29159 lastprivate clauses in SIMT case.
29161 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
29163 * doc/invoke.texi (-Wextra-semi): Document new warning option.
29165 2017-04-20 Richard Biener <rguenther@suse.de>
29167 PR tree-optimization/57796
29168 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
29169 as N scalar stores.
29170 (vect_model_load_cost): Cost gathers as N scalar loads.
29172 2017-04-20 Richard Biener <rguenther@suse.de>
29174 * ggc-page.c (ggc_allocated_p): Rename to ...
29175 (safe_lookup_page_table_entry): ... this and return the lookup
29177 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
29179 2017-04-20 Richard Biener <rguenther@suse.de>
29181 PR tree-optimization/80453
29182 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
29183 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
29184 from the conditions.
29185 (vn_phi_eq): Pass them down.
29186 (vn_phi_lookup): Record them.
29187 (vn_phi_insert): Likewise.
29189 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
29191 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
29192 uninitialized variable warning to avoid buffer overrun.
29194 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
29197 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
29198 is suppressed for '{ 0 }' in C.
29200 2017-04-20 Jakub Jelinek <jakub@redhat.com>
29202 * BASE-VER: Set to 8.0.0.
29204 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
29206 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
29207 priority .init_array and .fini_array section with SECTION_NOTYPE
29210 2017-04-20 Jakub Jelinek <jakub@redhat.com>
29212 PR middle-end/80423
29213 * tree.h (build_array_type): Add typeless_storage default argument.
29214 * tree.c (type_cache_hasher::equal): Also compare
29215 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
29216 (build_array_type): Add typeless_storage argument, set
29217 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
29219 (build_nonshared_array_type): Adjust build_array_type_1 caller.
29220 (build_array_type): Likewise. Add typeless_storage argument.
29222 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
29223 Jakub Jelinek <jakub@redhat.com>
29225 PR tree-optimization/80426
29226 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
29227 operation on symbolic operands, also compute the overflow for the
29228 invariant part when the operation degenerates into a negation.
29230 2017-04-19 Jakub Jelinek <jakub@redhat.com>
29233 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
29234 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
29237 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
29239 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
29242 * config/avr/avr.c (tree.h): Include it.
29243 (cgraph.h): Include it.
29244 (avr_encode_section_info): Don't warn for uninitialized progmem
29245 variable if it's just an alias.
29247 2017-04-19 Richard Biener <rguenther@suse.de>
29250 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
29251 when needed by AutoPGO.
29253 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
29256 * doc/lto.texi: Remove an extra 'that'.
29258 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
29260 PR rtl-optimization/80429
29261 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
29262 are only used in debug insns.
29264 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
29265 Vladimir Makarov <vmakarov@redhat.com>
29267 * config/sparc/predicates.md (input_operand): Add comment. Return
29268 true for any memory operand when LRA is in progress.
29269 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
29271 2017-04-18 Jeff Law <law@redhat.com>
29274 * mips.md ({return,simple_return}_internal): Do not overwrite
29277 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29279 PR tree-optimization/80443
29280 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
29281 instead of adding 1, subtract -1 and similarly instead of subtracting
29284 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
29286 PR rtl-optimization/80357
29287 * haifa-sched.c (tmp_bitmap): New variable.
29288 (model_recompute): Handle duplicate use records.
29289 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
29290 (free_global_sched_pressure_data): Free it.
29292 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
29295 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29296 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29297 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29298 instead of SYSTEM_HEADER_DIR.
29300 2017-04-18 Jeff Law <law@redhat.com>
29302 PR middle-end/80422
29303 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
29304 predecessors after walking up the insn chain.
29306 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29309 * dwarf2out.c (modified_type_die): Try harder not to emit internal
29310 sizetype type into debug info.
29312 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29315 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
29316 unneeded test for TARGET_UPPER_REGS_SF.
29317 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
29319 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29322 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
29323 instead of gsi_after_labels.
29325 2017-04-18 Jeff Law <law@redhat.com>
29327 * regcprop.c (maybe_mode_change): Avoid creating copies of the
29331 2017-04-13 Jeff Law <law@redhat.com>
29332 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
29333 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
29335 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
29338 * config/avr/avr.c (intl.h): Include it.
29339 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
29341 2017-04-18 Martin Liska <mliska@suse.cz>
29343 PR gcov-profile/78783
29344 * gcov-tool.c (gcov_output_files): Validate that destination
29345 file is either removed by the tool or by a user.
29347 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
29348 Guy Benyei <guybe@mellanox.com>
29350 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
29351 block, and do not negate it, the stored id is already negative.
29353 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
29355 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
29357 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29360 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
29361 masks of options that should be turned off if the VSX vector
29362 options are turned off.
29363 (OTHER_P8_VECTOR_MASKS): Likewise.
29364 (OTHER_VSX_VECTOR_MASKS): Likewise.
29365 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
29366 rs6000_disable_incompatible_switches to validate no type switches
29368 (rs6000_incompatible_switch): New function to disallow turning on
29369 other vector options if -mno-vsx, -mno-power8-vector, or
29370 -mno-power9-vector are specified.
29372 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
29374 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
29376 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
29378 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
29379 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
29380 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
29381 (ARG_POINTER_CFA_OFFSET): Likewise.
29383 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
29385 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
29386 conditions to take advantage of various optimizations.
29388 2017-04-13 Jeff Law <law@redhat.com>
29390 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
29391 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
29392 (zero_extendsidi2_dext): Likewise.
29394 2017-04-13 Jakub Jelinek <jakub@redhat.com>
29397 * fold-const.c (fold_ternary_loc): Revert
29398 use op0 instead of fold_convert_loc (loc, type, arg0) part of
29401 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
29403 PR rtl-optimization/80343
29404 * lra-remat.c (update_scratch_ops): Assign original hard reg to
29405 new scratch pseudo.
29407 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
29410 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
29411 to ubsan_encode_value.
29413 2017-04-13 Jeff Law <law@redhat.com>
29415 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
29416 appearing in DEBUG_INSNs.
29418 2017-04-13 Martin Liska <mliska@suse.cz>
29420 PR gcov-profile/80413
29421 * gcov-io.c (gcov_write_string): Copy to buffer just when
29422 allocated size is greater than zero.
29424 2017-04-13 Jakub Jelinek <jakub@redhat.com>
29427 * dwarf2out.c (decls_for_scope): Ignore declarations of
29428 current_function_decl in BLOCK_NONLOCALIZED_VARS.
29430 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
29433 * ipa-visibility.c (non_local_p): Fix typos.
29434 (localize_node): When localizing symbol in same comdat group,
29435 dissolve the group only when we know external symbols are going
29437 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
29439 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29441 PR tree-optimization/79390
29442 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
29443 order does not result in usable sequence, retry with reversed operand
29449 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
29450 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
29451 op0 instead of fold_convert_loc (loc, type, arg0).
29453 2017-04-12 Jeff Law <law@redhat.com>
29455 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
29456 has a delay slot in the generated code.
29458 * config/cris/cris.md (cris_preferred_reload_class): Return
29459 GENNONACR_REGS rather than GENERAL_REGS.
29461 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29464 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
29465 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
29466 signedness of the result type.
29468 2017-04-12 Richard Biener <rguenther@suse.de>
29469 Jeff Law <law@redhat.com>
29471 PR tree-optimization/80359
29472 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
29473 trim stores to TARGET_MEM_REFs.
29475 2017-04-12 Richard Biener <rguenther@suse.de>
29477 PR tree-optimization/79390
29478 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
29479 threading case even more.
29481 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
29484 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
29485 for quad_address_p for TImode, instead of just not indexed_address.
29487 2017-04-12 Richard Biener <rguenther@suse.de>
29488 Bernd Edlinger <bernd.edlinger@hotmail.de>
29490 PR middle-end/79671
29491 * alias.c (component_uses_parent_alias_set_from): Handle
29492 TYPE_TYPELESS_STORAGE.
29493 (get_alias_set): Likewise.
29494 * tree-core.h (tree_type_common): Add typeless_storage flag.
29495 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
29496 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
29497 for types containing members with TYPE_TYPELESS_STORAGE.
29498 (place_field): Likewise.
29499 (layout_type): Likewise for ARRAY_TYPE.
29500 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
29501 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
29502 TYPE_TYPELESS_STORAGE.
29503 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
29505 2017-04-12 Jakub Jelinek <jakub@redhat.com>
29508 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
29509 first argument to type.
29511 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29515 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
29516 CONST0_RTX (mode) rather than const0_rtx where appropriate.
29517 (rs6000_expand_binop_builtin): Likewise.
29518 (rs6000_expand_ternop_builtin): Likewise; also add missing
29519 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
29520 vshasigma built-ins.
29521 * doc/extend.texi: Document that vec_xxpermdi's third argument
29522 must be a constant.
29524 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
29526 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
29527 Use shift_const cost parameter when calculating gain of STV shifts.
29529 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
29531 PR rtl-optimization/70478
29532 * lra-constraints.c (process_alt_operands): Check memory for
29533 disfavoring memory insn operand.
29535 2017-04-11 Jakub Jelinek <jakub@redhat.com>
29537 PR middle-end/80100
29538 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
29539 left shift in unsigned HOST_WIDE_INT type.
29541 PR rtl-optimization/80385
29542 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
29543 (not (neg X)) into (plus X -1) for complex or non-integral modes.
29546 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
29547 if they have any depend clauses.
29549 2017-04-11 Martin Liska <mliska@suse.cz>
29552 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
29553 * ipa-split.c (split_function): Create a local comdat symbol
29554 if caller is in a comdat group.
29556 2017-04-11 Martin Liska <mliska@suse.cz>
29559 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
29562 2017-04-11 Martin Sebor <msebor@redhat.com>
29564 PR middle-end/80364
29565 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
29566 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
29568 (directive::set_width, directive::set_precision, format_character):
29570 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
29573 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
29576 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
29577 conflict, set target->arch_name instead of target->cpu_name.
29579 2017-04-11 Richard Biener <rguenther@suse.de>
29581 PR tree-optimization/80374
29582 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
29583 build_zero_cst, remove fold_convertible_p check again.
29585 2017-04-11 Martin Liska <mliska@suse.cz>
29588 * ubsan.c (instrument_object_size): Do not instrument register
29591 2017-04-11 Jakub Jelinek <jakub@redhat.com>
29594 * config/i386/i386-builtin-types.def
29595 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
29596 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
29597 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
29598 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
29599 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
29600 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
29601 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
29602 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
29603 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
29604 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
29605 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
29606 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
29607 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
29608 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
29609 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
29610 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
29611 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
29612 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
29613 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
29614 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
29615 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
29616 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
29617 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
29618 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
29619 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
29620 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
29621 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
29622 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
29623 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
29624 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
29625 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
29626 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
29627 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
29628 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
29629 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
29630 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
29631 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
29632 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
29633 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
29634 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
29635 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
29636 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
29637 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
29638 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
29639 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
29640 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
29642 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
29643 flag to second_arg_count, handle 4 argument function type _COUNT
29644 aliases, handle second_arg_count on second argument rather than last.
29646 2017-04-10 Jeff Law <law@redhat.com>
29648 PR tree-optimization/80374
29649 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
29650 record anything if we can not convert integer_zero_node to the
29653 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29656 * config/rs6000/rs6000.c (rs6000_option_override_internal):
29657 Enhance special handling given to the TARGET_P9_MINMAX option in
29658 relation to certain other options.
29660 2017-04-10 Bin Cheng <bin.cheng@arm.com>
29662 PR tree-optimization/80153
29663 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
29664 remove POINTER_PLUS_EXPR's base part directly, rather than through
29667 2017-04-10 Richard Biener <rguenther@suse.de>
29668 Bin Cheng <bin.cheng@arm.com>
29670 PR tree-optimization/80153
29671 * tree-affine.c (aff_combination_to_tree): Get base pointer from
29672 the first element of pointer type aff_tree. Build result expr in
29674 (add_elt_to_tree): Convert to type unconditionally. Remove other
29675 fold_convert calls.
29676 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
29677 (rewrite_use_nonlinear_expr): Check invariant using iv information.
29679 2017-04-10 Richard Biener <rguenther@suse.de>
29681 * tree-ssa-structalias.c (find_func_aliases): Properly handle
29684 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
29686 PR rtl-optimization/70478
29687 * lra-constraints.c (curr_small_class_check): New.
29688 (update_and_check_small_class_inputs): New.
29689 (process_alt_operands): Update curr_small_class_check. Disfavor
29690 alternative insn memory operands. Check available regs for small
29693 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
29696 * config/mips/mips.opt (-mvirt): Update description.
29697 * doc/invoke.texi (-mvirt): Likewise.
29699 2017-04-10 Richard Biener <rguenther@suse.de>
29701 PR middle-end/80362
29702 * fold-const.c (fold_binary_loc): Look at unstripped ops when
29703 looking for NEGATE_EXPR in -A / -B to A / B folding.
29705 2017-04-10 Martin Liska <mliska@suse.cz>
29707 PR gcov-profile/80224
29708 * gcov.c (print_usage): Fix usage string.
29709 (get_gcov_intermediate_filename): Remove.
29710 (output_gcov_file): Use both for normal and intermediate format.
29711 (generate_results): Do not initialize special file for
29712 intermediate format.
29714 2017-04-10 Richard Biener <rguenther@suse.de>
29716 PR tree-optimization/80304
29717 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
29720 2017-04-10 Nathan Sidwell <nathan@acm.org>
29723 * config/rs6000/rs6000.c (rs6000_vector_type): New.
29724 (rs6000_init_builtins): Use it.
29726 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29728 * config/arm/arm.md (<mrc>): Add mode to SET source.
29729 (<mrrc>): Likewise.
29731 2017-04-10 Richard Biener <rguenther@suse.de>
29733 PR middle-end/80344
29734 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
29736 2017-04-10 Jakub Jelinek <jakub@redhat.com>
29739 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
29740 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
29741 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
29742 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
29743 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
29744 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
29745 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
29746 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
29747 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
29748 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
29749 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
29750 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
29751 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
29752 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
29753 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
29754 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
29755 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
29756 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
29757 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
29758 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
29759 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
29760 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
29761 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
29763 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
29765 PR rtl-optimization/70478
29766 * lra-constraints.c: Reverse the last patch.
29768 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
29770 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
29771 Add comment for WCHAR_T.
29773 2017-04-08 Martin Liska <mliska@suse.cz>
29776 2017-04-07 Martin Liska <mliska@suse.cz>
29779 * ipa-split.c (split_function): Add function part to a same comdat
29782 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29785 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
29787 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
29789 * rs6000/rs6000.c (vec_load_pendulum): Rename...
29790 (vec_pairing): ...to this.
29791 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
29792 (rs6000_sched_init): Adjust for name change.
29793 (struct rs6000_sched_context): Likewise.
29794 (rs6000_init_sched_context): Likewise.
29795 (rs6000_set_sched_context): Likewise.
29797 2017-04-07 Jakub Jelinek <jakub@redhat.com>
29803 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
29805 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
29806 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
29807 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
29809 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
29811 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
29813 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
29815 PR rtl-optimization/70703
29816 * ira-color.c (update_conflict_hard_regno_costs): Use
29817 int64_t instead of HOST_WIDE_INT.
29819 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
29821 PR rtl-optimization/70478
29822 * lra-constraints.c (process_alt_operands): Disfavor alternative
29823 insn memory operands.
29825 2017-04-07 Jeff Law <law@redhat.com>
29827 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
29828 CALL and NOTE_INSN_CALL_ARG_LOCATION.
29830 2017-04-07 Martin Liska <mliska@suse.cz>
29833 * config/aarch64/aarch64.c (aarch64_process_target_attr):
29834 Show error message instead of an ICE.
29836 2017-04-07 Martin Liska <mliska@suse.cz>
29839 * ipa-split.c (split_function): Add function part to a same comdat
29842 2017-04-07 Richard Biener <rguenther@suse.de>
29844 PR middle-end/80341
29845 * tree.c (get_unwidened): Also handle ! for_type case for
29847 * convert.c (do_narrow): Split out from ...
29848 (convert_to_integer_1): ... here. Do not pass final truncation
29849 type to get_unwidened for TRUNC_DIV_EXPR.
29851 2017-04-07 Richard Biener <rguenther@suse.de>
29853 * tree-affine.c (wide_int_ext_for_comb): Take type rather
29855 (aff_combination_const): Adjust.
29856 (aff_combination_scale): Likewise.
29857 (aff_combination_add_elt): Likewise.
29858 (aff_combination_add_cst): Likewise.
29859 (aff_combination_convert): Likewise.
29860 (add_elt_to_tree): Likewise. Remove unused argument.
29861 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
29863 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
29865 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
29867 * config/arm/arm.c (arm_default_short_enums): Use
29868 ARM_DEFAULT_SHORT_ENUMS.
29869 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
29871 2017-04-06 Jakub Jelinek <jakub@redhat.com>
29874 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
29875 members with redundant out-of-class redeclaration.
29877 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
29880 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
29881 * config/i386/i386.md (*zero_extendsidi2):
29882 Add (?*x,*x) and (?*v,*v) alternatives.
29884 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
29887 * config/i386/i386.c (ix86_expand_builtin)
29888 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
29889 mode from insn data. Convert operands to insn operand mode.
29890 Copy operands that don't satisfy insn predicate to a register.
29892 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
29894 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
29897 2017-04-06 Richard Biener <rguenther@suse.de>
29899 PR tree-optimization/80334
29900 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
29901 preserve alignment of accesses.
29903 2017-04-06 Richard Biener <rguenther@suse.de>
29905 PR tree-optimization/80262
29906 * tree-sra.c (build_ref_for_offset): Preserve address-space
29908 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
29909 Drop useless address-space information on MEM_REF offsets.
29911 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
29913 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
29915 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
29917 PR rtl-optimization/70703
29918 * ira-color.c (update_conflict_hard_regno_costs): Use
29919 HOST_WIDE_INT instead of long.
29921 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
29924 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
29925 not defined for x86_64 target. Add -mmmx target option when __SSE2__
29927 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
29928 for x86_64 target. Handle -m3dnowa option.
29930 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
29932 PR rtl-optimization/70703
29933 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
29934 (update_conflict_hard_regno_costs): Use long instead of unsigned
29935 arithmetic for cost calculation.
29937 2017-04-05 Jakub Jelinek <jakub@redhat.com>
29938 Bernd Edlinger <bernd.edlinger@hotmail.de>
29941 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
29944 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
29947 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
29948 ptr_mode with Pmode throughout.
29949 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
29950 into probe_stack_range and use DImode.
29952 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
29955 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
29956 call_eh_return is true.
29958 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29960 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
29961 Initialize last_match_fntype_index.
29963 2017-04-05 Jakub Jelinek <jakub@redhat.com>
29966 * tree-nvr.c: Include internal-fn.h.
29967 (pass_return_slot::execute): Ignore internal calls without
29970 2017-04-04 Jakub Jelinek <jakub@redhat.com>
29971 Richard Biener <rguenther@suse.de>
29974 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
29975 captures used multiple times, except for the last use.
29976 * generic-match-head.c: Include gimplify.h.
29978 2017-04-04 Jakub Jelinek <jakub@redhat.com>
29980 PR tree-optimization/79390
29981 * target.h (struct noce_if_info): Declare.
29982 * targhooks.h (default_noce_conversion_profitable_p): Declare.
29983 * target.def (noce_conversion_profitable_p): New target hook.
29984 * ifcvt.h (struct noce_if_info): New type, moved from ...
29985 * ifcvt.c (struct noce_if_info): ... here.
29986 (noce_conversion_profitable_p): Renamed to ...
29987 (default_noce_conversion_profitable_p): ... this. No longer
29989 (noce_try_store_flag_constants, noce_try_addcc,
29990 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
29991 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
29992 instead of noce_conversion_profitable_p.
29993 * config/i386/i386.c: Include ifcvt.h.
29994 (ix86_option_override_internal): Don't override
29995 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
29996 (ix86_noce_conversion_profitable_p): New function.
29997 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
29998 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
29999 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
30000 * doc/tm.texi: Regenerated.
30002 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30004 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
30007 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
30010 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
30011 instructions for small multiply cores.
30013 2017-04-04 Jeff Law <law@redhat.com>
30015 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
30017 (mips_expand_vec_perm_const): Initialize elements in orig_perm
30018 that are not set by the loop over the elements.
30020 2017-04-04 Jakub Jelinek <jakub@redhat.com>
30023 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
30024 int mode, convert_modes it to mode as unsigned, otherwise use
30025 lowpart_subreg to mode rather than SImode.
30026 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
30027 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
30028 Use DImode instead of SImode for the shift count operand.
30029 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
30032 2017-04-04 Richard Biener <rguenther@suse.de>
30034 PR middle-end/80281
30035 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
30036 arithmetic done for the negate or the plus. Simplify.
30037 (A - (-B) -> A + B): Likewise.
30038 * fold-const.c (split_tree): Make sure to not negate pointers.
30040 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
30042 PR rtl-optimization/60818
30043 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
30044 a compare of comparisons with the thing compared if this results
30045 in a different machine mode.
30047 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
30049 * alias.c (base_alias_check): Fix typo in comment.
30050 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
30051 * cgraphunit.c (symbol_table::compile): Likewise.
30052 * collect2.c (maybe_run_lto_and_relink): Likewise.
30053 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
30054 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
30055 * config/avr/avr.c (avr_map_op_t): Likewise.
30056 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
30057 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
30058 * config/epiphany/epiphany.md (movcc): Likewise.
30059 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
30060 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
30062 * config/mips/mips.c (mips_save_restore_reg): Likewise.
30063 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
30064 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
30065 * config/sh/sh.c (sh_rtx_costs): Likewise.
30066 * fold-const.c (fold_truth_andor): Likewise.
30067 * genautomata.c (collapse_flag): Likewise.
30068 * gengtype.h (struct type::u::s): Likewise.
30069 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
30070 * input.c (FORMAT_AMOUNT): Likewise.
30071 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
30072 (known_aggs_to_agg_replacement_list): Likewise.
30073 * ipa-inline-analysis.c: Likewise.
30074 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
30075 * ipa-polymorphic-call.c
30076 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
30077 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
30078 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
30080 * modulo-sched.c (apply_reg_moves): Likewise.
30081 * omp-expand.c (build_omp_regions_1): Likewise.
30082 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
30083 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
30084 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
30085 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
30086 * value-prof.c: Likewise.
30087 * var-tracking.c (val_reset): Likewise.
30089 2017-04-03 Richard Biener <rguenther@suse.de>
30091 PR tree-optimization/80275
30092 * fold-const.c (split_address_to_core_and_offset): Handle
30095 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
30097 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
30098 descriptors is at least equal to that of functions.
30100 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
30102 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
30104 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
30107 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
30108 (mov<IMOD4:mode>): New expander.
30109 (*mov<IMOD4:mode>_internal): New insn and split pattern.
30111 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
30113 PR rtl-optimization/79405
30114 * fwprop.c (propagations_left): New variable.
30115 (forward_propagate_into): Decrement it.
30116 (fwprop_init): Initialize it.
30117 (fw_prop): If the variable has reached zero, stop propagating.
30118 (fwprop_addr): Ditto.
30120 2017-03-31 Jakub Jelinek <jakub@redhat.com>
30123 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
30124 a FUNCTION_DECL, pass it as decl instead of origin to
30127 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
30129 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
30132 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
30135 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
30136 TARGET_VSX_SMALL_INTEGER.
30138 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30140 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30141 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
30143 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
30145 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
30146 extraction from odd-numbered MSA register.
30148 2017-03-31 Jakub Jelinek <jakub@redhat.com>
30150 PR middle-end/80173
30151 * expmed.c (store_bit_field_1): Don't attempt to create
30152 a word subreg out of hard registers wider than word if they
30153 have HARD_REGNO_NREGS of 1 for their mode.
30155 PR middle-end/80163
30156 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
30157 conversions to integer types wider than word and pointer.
30160 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
30161 (rtx_equal_for_cselib_p): Pass 0 to it.
30162 * cselib.c (cselib_hasher::equal): Likewise.
30163 (rtx_equal_for_cselib_1): Add depth argument. If depth
30164 is 128, don't look up VALUE locs and punt. Increment
30165 depth in recursive calls when walking VALUE locs.
30167 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
30169 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
30170 (make_gcov_file_name): Use the canonical path name for generating
30172 (read_line): Fix handling of files with ascii null bytes.
30174 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
30176 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
30177 to initialise a vector register instead
30178 of using a const_int.
30180 2017-03-30 Jakub Jelinek <jakub@redhat.com>
30182 PR translation/80189
30183 * gimplify.c (omp_default_clause): Use %qs instead of %s in
30184 diagnostic messages.
30186 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
30189 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
30190 (dfp_diex_<mode>): Update mode of operand 1.
30191 * doc/extend.texi (dxex, dxexq): Document change to return type.
30192 (diex, diexq): Document change to argument type.
30194 2017-03-30 Martin Jambor <mjambor@suse.cz>
30197 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
30198 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
30199 it reflects the signature changes performed at the callee side.
30200 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
30201 to cgraph_build_function_type_skip_args.
30202 (build_function_decl_skip_args): Adjust call to the above function.
30204 2017-03-30 Jakub Jelinek <jakub@redhat.com>
30207 * config/i386/sse.md
30208 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
30209 register as dest whenever it is a MEM not rtx_equal_p to the
30210 corresponding dup operand, and when forcing into reg move the
30211 reg into the memory afterwards.
30212 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
30213 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
30214 for the force_reg mode.
30215 (avx512vl_vextractf128<mode>): Use register as dest either
30216 always when a MEM, or when it is a MEM not rtx_equal_p to the
30217 corresponding dup operand, or even not when it is a CONST_VECTOR
30218 depending on the mode and lo vs. hi.
30219 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
30221 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
30222 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
30223 Likewise. Require that operands[2] is even.
30224 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
30225 Remove extraneous parens. Require that operands[2] is a multiple
30227 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
30228 operands[0] is a MEM if <mask_applied>, the predicates/constraints
30229 disallow memory then.
30231 2017-03-30 Richard Biener <rguenther@suse.de>
30233 PR tree-optimization/77498
30234 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
30235 to non-constants over backedges.
30237 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
30239 PR rtl-optimization/80233
30240 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
30241 as last_combined_insn. Do not test for BARRIER_P separately.
30243 2017-03-29 Andreas Schwab <schwab@suse.de>
30246 * calls.c (prepare_call_address): Convert funexp to Pmode before
30247 copying to temp reg.
30249 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30251 PR tree-optimization/80158
30252 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
30253 Handle possible future case of more than one alternate
30255 (replace_rhs_if_not_dup): Likewise.
30256 (replace_one_candidate): Likewise.
30258 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
30260 PR rtl-optimization/80193
30261 * ira.c (ira): Do not check allocation for LRA.
30263 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
30265 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
30266 (nvptx_output_simt_exit): Declare.
30267 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
30268 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
30269 (init_softstack_frame): Move initialization of crtl->is_leaf to...
30270 (nvptx_declare_function_name): ...here. Emit declaration of local
30271 memory space buffer for omp_simt_enter insn.
30272 (nvptx_output_unisimt_switch): New.
30273 (nvptx_output_softstack_switch): New.
30274 (nvptx_output_simt_enter): New.
30275 (nvptx_output_simt_exit): New.
30276 * config/nvptx/nvptx.h (struct machine_function): New fields
30277 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
30278 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
30279 (UNSPECV_SIMT_EXIT): Ditto.
30280 (omp_simt_enter_insn): New insn.
30281 (omp_simt_enter): New expansion.
30282 (omp_simt_exit): New insn.
30283 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
30285 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
30286 (expand_GOMP_SIMT_ENTER_ALLOC): New.
30287 (expand_GOMP_SIMT_EXIT): New.
30288 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
30289 (GOMP_SIMT_ENTER_ALLOC): Ditto.
30290 (GOMP_SIMT_EXIT): Ditto.
30291 * target-insns.def (omp_simt_enter): New insn.
30292 (omp_simt_exit): Ditto.
30293 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
30295 (lower_rec_simd_input_clauses): Implement SIMT privatization.
30296 (lower_rec_input_clauses): Likewise.
30297 (lower_lastprivate_clauses): Handle SIMT privatization.
30299 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
30300 (ompdevlow_adjust_simt_enter): New.
30301 (find_simtpriv_var_op): New.
30302 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
30303 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
30305 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
30306 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
30307 (copy_decl_for_dup_finish): Ditto.
30309 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
30311 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
30314 * config/i386/i386.c (ix86_option_override_internal): Always
30315 allow -mpreferred-stack-boundary=3 for 64-bit targets.
30317 2017-03-28 Bin Cheng <bin.cheng@arm.com>
30319 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
30321 2017-03-28 Bin Cheng <bin.cheng@arm.com>
30323 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
30324 mark new edge's irreducible flag accordign to it.
30325 (vect_do_peeling): Check loop preheader edge's irreducible flag
30326 and pass it to function slpeel_add_loop_guard.
30328 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
30330 PR tree-optimization/80218
30331 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
30332 Update block frequencies and counts.
30334 2017-03-28 Richard Biener <rguenther@suse.de>
30336 PR tree-optimization/78644
30337 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
30338 of a simplification result we may not use it at all.
30340 2017-03-28 Richard Biener <rguenther@suse.de>
30343 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
30344 without arguments, generate default definition of a SSA name.
30346 2017-03-28 Richard Biener <rguenther@suse.de>
30348 PR middle-end/80222
30349 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
30350 TYPE_REF_CAN_ALIAS_ALL references.
30351 * fold-const.c (fold_indirect_ref_1): Likewise.
30353 2017-03-28 Martin Liska <mliska@suse.cz>
30356 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
30357 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
30359 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
30360 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
30362 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
30363 (EXTRA_SPECS): Define.
30364 (SUBTARGET_EXTRA_SPECS): Likewise.
30365 (SUBTARGET_CPP_SPEC): Likewise.
30366 * config/arc/elf.h (EXTRA_SPECS): Renamed to
30367 SUBTARGET_EXTRA_SPECS.
30368 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
30370 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
30372 * config/arc/simdext.md (vst64_insn): Update pattern.
30373 (vld32wh_insn): Likewise.
30374 (vld32wl_insn): Likewise.
30375 (vld64_insn): Likewise.
30376 (vld32_insn): Likewise.
30378 2017-03-28 Marek Polacek <polacek@redhat.com>
30381 * fold-const.c (fold_comparison): Use protected_set_expr_location
30382 instead of SET_EXPR_LOCATION.
30384 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
30386 * tree.c (add_expr): Avoid name lookup warning.
30388 2017-03-27 Jeff Law <law@redhat.com>
30390 PR tree-optimization/80216
30391 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
30392 function name. Limit recursion depth.
30393 (record_temporary_equivalences): Corresponding changes.
30395 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
30397 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
30400 2017-03-27 Jakub Jelinek <jakub@redhat.com>
30403 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
30405 * cfgcleanup.c (reg_note_cfa_p): New array.
30406 (insns_have_identical_cfa_notes): New function.
30407 (old_insns_match_p): Don't cross-jump in between /f
30408 and non-/f instructions. If both i1 and i2 are frame related,
30409 verify all CFA notes, their order and content.
30411 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
30414 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
30415 HImode and SImode with zero extend to DImode to one insn.
30416 (bswap<mode>2_extenddi): Likewise.
30417 (bswapsi2_extenddi): Likewise.
30418 (bswaphi2_extendsi): Likewise.
30419 (bswaphi2): Combine bswap HImode and SImode into one insn.
30420 Separate memory insns from swapping register.
30421 (bswapsi2): Likewise.
30422 (bswap<mode>2): Likewise.
30423 (bswaphi2_internal): Delete, no longer used.
30424 (bswapsi2_internal): Likewise.
30425 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
30426 store, and gpr<-gpr swap insns.
30427 (bswap<mode>2_store): Likewise.
30428 (bswaphi2_reg): Register only splitter, combine with the splitter.
30429 (bswaphi2 splitter): Likewise.
30430 (bswapsi2_reg): Likewise.
30431 (bswapsi2 splitter): Likewise.
30432 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
30433 the insns into load, store, and register/register insns.
30434 (bswapdi2_ldbrx): Likewise.
30435 (bswapdi2_load): Likewise.
30436 (bswapdi2_store): Likewise.
30437 (bswapdi2_reg): Likewise.
30439 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
30441 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
30442 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
30444 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
30447 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
30449 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30450 special handling for target option conflicts between dform
30451 options (-mpower9-dform, -mpower9-dform-vector,
30452 -mpower9-dform-scalar) and -mno-direct-move.
30454 2017-03-27 Richard Biener <rguenther@suse.de>
30456 PR tree-optimization/80181
30457 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
30459 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30461 * config/arc/predicates.md (move_double_src_operand): Replace the
30462 call to move_double_src_operand with a call to address_operand.
30464 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30466 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
30467 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
30468 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
30470 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30472 * config/arc/predicates.md (long_immediate_loadstore_operand):
30473 Consider scaled addresses cases.
30475 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
30477 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
30478 restored when in interrupt.
30479 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
30480 doesn't have delay slot.
30482 2017-03-27 Richard Biener <rguenther@suse.de>
30485 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
30486 inlined thunk clones.
30488 2017-03-27 Jakub Jelinek <jakub@redhat.com>
30491 * asan.c (instrument_derefs): Copy over last operand from
30492 original COMPONENT_REF to the new COMPONENT_REF with
30493 DECL_BIT_FIELD_REPRESENTATIVE.
30494 * ubsan.c (instrument_object_size): Likewise.
30496 2017-03-27 Richard Biener <rguenther@suse.de>
30498 PR tree-optimization/80170
30499 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
30500 sure DR/SCEV didnt fold in constants we do not see when looking
30501 at the reference base alignment.
30503 2017-03-27 Richard Biener <rguenther@suse.de>
30505 PR middle-end/80171
30506 * gimple-fold.c (fold_ctor_reference): Properly guard against
30507 NULL return value from canonicalize_constructor_val.
30509 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
30512 * config/i386/i386.c (ix86_expand_builtin)
30513 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
30514 flags reg setting and flags reg using instructions.
30515 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
30516 clobbering instructions to zero extend op2.
30518 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
30520 * doc/install.texi (Configuration) <--with-aix-soname>:
30521 Update link to AIX ld.
30523 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
30525 PR rtl-optimization/80160
30526 PR rtl-optimization/80159
30527 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
30528 reg_alternate_class into account.
30530 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
30533 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
30534 to consider in curr_insn_transform.
30536 2017-03-24 Jakub Jelinek <jakub@redhat.com>
30538 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
30539 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
30540 and emit_mode_inner.
30542 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30544 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
30545 argument to the overloaded builtin variants. Use the new flag to
30546 deprecate certain builtin variants.
30547 * config/s390/s390-builtin-types.def: Add new builtin types.
30548 * config/s390/s390-builtins.h: Support new flags field for
30549 overloaded builtins.
30550 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
30551 (s390_macro_to_expand): Enable vector float data type.
30552 (s390_cpu_cpp_builtins_internal): Indicate support of the new
30553 builtins by incrementing the __VEC__ version number.
30554 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
30556 (s390_resolve_overloaded_builtin): Emit error messages depending
30557 on the builtin flags.
30558 * config/s390/s390.c (s390_expand_builtin): Support additional
30559 flags argument. Change error message to match the messages
30560 emitted in s390-c.c.
30561 * config/s390/s390.md: New UNSPEC_* constants.
30562 (op_type): Add new instruction types.
30563 * config/s390/vecintrin.h: Add new builtins and test data class
30565 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
30566 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
30567 (VEC_INEXACT, VEC_NOINEXACT): New constants.
30568 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
30569 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
30570 ("vec_mergel<mode>"): V_HW -> VEC_HW.
30572 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
30573 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
30574 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
30575 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
30577 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
30578 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
30579 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
30580 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
30582 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
30583 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
30584 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
30585 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
30586 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
30587 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
30588 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
30590 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
30591 ("vec_scatter_element<V_HW_4:mode>_DI")
30592 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
30593 ("vec_fpint<mode>", "vflls")
30594 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
30595 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
30596 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
30597 ("*vec_cmphe<mode>_cc"): ... these.
30599 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
30600 mode constant instead of magic value.
30602 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30604 * config/s390/s390.c (s390_expand_vec_compare): Support other
30605 vector floating point modes than just V2DF.
30606 (s390_expand_vcond): Likewise.
30607 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
30608 (s390_cannot_change_mode_class): Prevent mode changes between TF
30609 and V1TF in vector registers.
30610 * config/s390/s390.md (DF, SF): New mode attributes.
30611 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
30612 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
30613 SFmode support for VRs.
30614 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
30616 (VFT, VF_HW): New mode iterators.
30617 (vw, sdx): New mode attributes.
30618 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
30619 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
30620 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
30621 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
30622 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
30623 also the new vector floating point modes. Renaming to ...
30625 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
30626 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
30627 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
30628 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
30629 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
30630 ("vec_unordered<mode>"): ... these.
30632 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
30633 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
30634 ("*vec_extendv2df"): New insn definitions.
30636 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30638 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
30639 ("mulditi3_2", "*muldi3_sign"): New patterns.
30640 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
30641 rename the pattern definition.
30643 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30645 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
30647 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
30649 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30651 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
30652 instruction if possible.
30653 * config/s390/vector.md (vec_halfnumelts): New mode
30655 ("*vec_vllezlf<mode>"): New pattern.
30657 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30659 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
30660 ("popcountv4si2", "popcountv2di2"): Rename to ...
30661 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
30662 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
30664 ("popcount<mode>2_vxe"): New pattern.
30666 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30668 * common/config/s390/s390-common.c (processor_flags_table): Add
30670 * config.gcc: Add arch12.
30671 * config/s390/driver-native.c (s390_host_detect_local_cpu):
30672 Default to arch12 for unknown CPU model numbers.
30673 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
30674 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
30675 PROCESSOR_max sanity check.
30676 * config/s390/s390-opts.h (enum processor_type): Add
30678 * config/s390/s390.c (processor_table): Add arch12.
30679 (s390_expand_builtin): Add check for B_VXE flag.
30680 (s390_issue_rate): Add PROCESSOR_ARCH12.
30681 (s390_get_sched_attrmask): Likewise.
30682 (s390_get_unit_mask): Likewise.
30683 (s390_sched_score): Enable z13 scheduling for arch12.
30684 (s390_sched_reorder): Likewise.
30685 (s390_sched_variable_issue): Likewise.
30686 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
30688 (s390_tune_attr): Use z13 scheduling also for arch12.
30689 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
30690 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
30691 (TARGET_VXE_P): New macros.
30692 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
30693 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
30694 * config/s390/s390.opt: Add arch12 as processor_type.
30696 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30698 * config/s390/s390.md
30699 ("fixuns_truncdddi2", "fixuns_trunctddi2")
30700 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
30701 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
30703 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
30704 Rename expanders to ...
30706 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
30707 ("fixuns_truncdddi2_emu"): ... these.
30709 ("fixuns_trunc<mode>si2_emu"): New expander.
30711 ("*fixuns_truncdfdi2_z13"): Rename to ...
30712 ("*fixuns_truncdfdi2_vx"): ... this.
30714 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30716 * config/s390/2964.md: Remove the single element vector compare
30717 instructions which are no longer used.
30718 * config/s390/s390.c (s390_select_ccmode): Remove handling of
30720 (s390_canonicalize_comparison): Remove handling of DFmode
30722 (s390_expand_vec_compare_scalar): Remove function.
30723 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
30724 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
30726 ("*cmp<mode>_ccs"): Add wfcdb instruction.
30728 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30730 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
30732 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
30733 will anyway by matched by mov<mode>_64dfp.
30735 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30737 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
30738 vlef/vstef. Add missing operand to vleif.
30740 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30742 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
30743 pair for all vector types with 64 bit elements.
30744 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
30745 * config/s390/vector.md (V_HW_64): ... here.
30746 (V_128_NOSINGLE): New mode iterator.
30747 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
30748 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
30749 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
30750 ("*vec_load_pairv2di"): Change to ...
30751 ("*vec_load_pair<mode>"): ... this one.
30753 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30755 * config/s390/constraints.md: Add comments.
30756 (jKK): Reject element sizes > 8 bytes.
30757 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
30759 * config/s390/s390.md: Add the s_operand checks formerly in
30760 s390_split_ok_p to various splitters where they are still
30762 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
30763 for 128 bit vectors. Plus two splitters.
30765 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30767 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
30770 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30773 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
30774 error if the boundary argument is not constant.
30776 2017-03-24 Jakub Jelinek <jakub@redhat.com>
30778 PR rtl-optimization/80112
30779 * loop-doloop.c (doloop_condition_get): Don't check condition
30780 if cmp isn't SET with IF_THEN_ELSE src.
30782 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30784 PR tree-optimization/80158
30785 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
30786 replacing a candidate statement, also replace it for the
30787 candidate's alternate interpretation.
30788 (replace_rhs_if_not_dup): Likewise.
30789 (replace_one_candidate): Likewise.
30791 2017-03-24 Richard Biener <rguenther@suse.de>
30793 PR tree-optimization/80167
30794 * graphite-isl-ast-to-gimple.c
30795 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
30797 (translate_isl_ast_to_gimple::get_rename): Likewise.
30799 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
30801 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
30802 handling of certain combinations of target options, including the
30803 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
30804 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
30806 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30809 * config/arm/arm.md (*load_multiple): Add reload_completed to
30810 matching condition.
30812 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30813 Richard Biener <rguenth@suse.de>
30815 PR tree-optimization/79908
30816 PR tree-optimization/80136
30817 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
30818 been cast away, gimplify_and_add suffices.
30820 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
30822 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
30824 2017-03-23 Richard Biener <rguenther@suse.de>
30826 PR tree-optimization/80032
30827 * gimplify.c (gimple_push_cleanup): Forced unconditional
30828 cleanups still have to go to the conditional_cleanups
30831 2017-03-22 Jakub Jelinek <jakub@redhat.com>
30833 PR tree-optimization/80072
30834 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
30836 (next_operand_entry_id): Change type to unsigned int.
30837 (sort_by_operand_rank): Make sure to return the right return value
30838 even if unsigned fields are bigger than INT_MAX.
30839 (struct oecount): Change cnt and id type to unsigned int.
30840 (oecount_hasher::equal): Formatting fix.
30841 (oecount_cmp): Make sure to return the right return value
30842 even if unsigned fields are bigger than INT_MAX.
30843 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
30846 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
30847 TREE_READONLY on result if writing it more than once.
30850 * doc/invoke.texi (-fsanitize=thread): Document that with
30851 -fnon-call-exceptions atomics are not able to throw
30855 * tsan.c: Include tree-eh.h.
30856 (instrument_builtin_call): Call maybe_clean_eh_stmt or
30857 maybe_clean_or_replace_eh_stmt where needed.
30858 (instrument_memory_accesses): Add cfg_changed argument.
30859 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
30860 if it returned true.
30861 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
30863 PR rtl-optimization/63191
30864 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
30865 wrapper function, moved the whole old content into ...
30866 (ix86_delegitimize_address_1): ... this. New inline function.
30867 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
30868 true as last argument instead of ix86_delegitimize_address.
30870 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
30872 * config/aarch64/aarch64.c (generic_branch_cost): Copy
30873 cortexa57_branch_cost.
30875 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
30877 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
30879 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30882 * doc/md.texi (Constraints): Document wA constraint.
30883 * config/rs6000/constraints.md (wA): New.
30884 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
30885 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
30886 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
30887 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
30889 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
30892 * gimplify.c (is_oacc_declared): New function.
30893 (oacc_default_clause): Use it to set default flags for acc declared
30894 variables inside parallel regions.
30895 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
30896 declared variables.
30897 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
30898 declare attribute to any decl as necessary.
30900 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
30903 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
30904 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
30905 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
30906 (arm_arch_lpae): This.
30907 * config/arm/arm.c (arm_arch7ve): Rename into ...
30908 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
30909 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
30912 2017-03-22 Martin Liska <mliska@suse.cz>
30915 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
30916 error message instead of an ICE.
30918 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30920 * doc/extend.texi (6.11 Additional Floating Types): Revise.
30922 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
30924 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30926 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30929 2017-03-21 Martin Sebor <msebor@redhat.com>
30931 * doc/extend.texi: Use "cannot" instead of "can't."
30932 * doc/hostconfig.texi: Same.
30933 * doc/install.texi: Same.
30934 * doc/invoke.texi: Same.
30935 * doc/loop.texi: Same.
30936 * doc/md.texi: Same.
30937 * doc/objc.texi: Same.
30938 * doc/rtl.texi: Same.
30939 * doc/tm.texi: Same.
30940 * doc/tm.texi.in: Same.
30941 * doc/trouble.texi: Same.
30943 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
30946 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
30947 (collect_checksum_attributes): Set it.
30948 (die_checksum_ordered): Use it.
30950 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30952 PR tree-optimization/79908
30953 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
30954 change: For a VA_ARG whose LHS has been cast away, use
30955 force_gimple_operand to construct the side effects.
30957 2017-03-21 David Malcolm <dmalcolm@redhat.com>
30959 PR translation/80001
30960 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
30961 more amenable to translation.
30962 (oacc_loop_auto_partitions): Likewise.
30964 2017-03-21 Marek Polacek <polacek@redhat.com>
30965 Martin Sebor <msebor@redhat.com>
30967 PR tree-optimization/80109
30968 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
30969 on INTEGRAL_TYPE_P.
30971 2017-03-21 Jakub Jelinek <jakub@redhat.com>
30972 Segher Boessenkool <segher@kernel.crashing.org>
30975 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
30976 check reg_used_between_p between insn and one of succ or succ2
30977 depending on if succ is artificial insn not inserted into insn
30980 2017-03-21 Martin Liska <mliska@suse.cz>
30982 PR gcov-profile/80081
30983 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
30984 * doc/gcc.texi: Include gcov-dump stuff.
30985 * doc/gcov-dump.texi: New file.
30987 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
30989 PR rtl-optimization/79150
30990 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
30991 conditional jump, if the jump is the last insn of the loop.
30993 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30994 Richard Biener <rguenth@suse.de>
30996 PR tree-optimization/79908
30997 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
30998 been cast away, use force_gimple_operand to construct the side
31001 2017-03-21 Martin Liska <mliska@suse.cz>
31003 PR libfortran/79956
31004 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
31007 2017-03-21 Brad Spengler <spender@grsecurity.net>
31010 * plugin.c (htab_hash_plugin): New function.
31011 (add_new_plugin): Use it and adjust.
31012 (parse_plugin_arg_opt): Adjust.
31013 (init_one_plugin): Likewise.
31015 2017-03-21 Richard Biener <rguenther@suse.de>
31017 PR tree-optimization/80032
31018 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
31019 if set force the cleanup to happen unconditionally.
31020 (gimplify_target_expr): Push inserted clobbers with force_uncond
31021 to avoid them being removed by control-dependent DCE.
31023 2017-03-21 Richard Biener <rguenther@suse.de>
31025 PR tree-optimization/80122
31026 * tree-inline.c (copy_bb): Do not expans va-arg packs or
31027 va_arg_pack_len when the inlined call stmt requires pack
31029 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
31031 2017-03-21 Jakub Jelinek <jakub@redhat.com>
31034 * tsan.c (instrument_builtin_call): If the memory model argument
31035 is not a constant, assume it is valid.
31038 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
31041 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
31043 PR rtl-optimization/79910
31044 * combine.c (can_combine_p): Do not allow combining an I0 or I1
31045 if its dest is used by an insn before I2 (other than the combined
31046 insns themselves, which are properly handled already).
31048 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
31051 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
31053 * combine.c (record_used_regs): New static function.
31054 (try_combine): Handle situations where there is an additional
31055 instruction between I2 and I3 which needs to have a LOG_LINK
31059 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
31061 * combine.c (try_combine): Delete redundant i1 test. Call
31062 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
31064 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31067 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
31068 alternatives 13/14.
31070 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31072 PR tree-optimization/80054
31073 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
31074 the optimization if a PHI or any of its arguments is not dominated
31075 by the candidate's basis. Use gphi* rather than gimple* as
31077 (replace_profitable_candidates): Clean up a gimple* variable that
31078 should be a gphi* variable.
31080 2017-03-20 Martin Sebor <msebor@redhat.com>
31083 * doc/extend.texi (attribute constructor): Document present limitation.
31085 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
31088 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
31089 __POWER9_VECTOR__ #ifdef control, change template definition to
31090 use Power9-specific built-in function.
31091 (vec_any_eq): Likewise.
31092 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
31093 to control outcomes from this test.
31094 (vector_ae_<mode>p): For VEC_F modes, likewise.
31096 2017-03-20 Ian Lance Taylor <iant@google.com>
31098 * config/i386/i386.c (ix86_function_regparm): Save an extra
31099 register for -fsplit-stack with DECL_STATIC_CHAIN.
31101 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
31104 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
31105 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
31107 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
31109 * config/riscv/riscv.c (riscv_print_operand): Use "fence
31111 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
31114 2017-03-20 Marek Polacek <polacek@redhat.com>
31117 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
31119 2017-03-20 Richard Biener <rguenther@suse.de>
31121 PR tree-optimization/80113
31122 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
31123 allocate extra SSA name for PHI def.
31124 (add_close_phis_to_outer_loops): Likewise.
31125 (add_close_phis_to_merge_points): Likewise.
31126 (copy_loop_close_phi_args): Likewise.
31127 (copy_cond_phi_nodes): Likewise.
31129 2017-03-20 Martin Liska <mliska@suse.cz>
31131 PR middle-end/79753
31132 * tree-chkp.c (chkp_build_returned_bound): Do not build
31133 returned bounds for a LHS that's not a BOUNDED_P type.
31135 2017-03-20 Martin Liska <mliska@suse.cz>
31139 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
31140 COMPLEX_CST and VECTOR_CST.
31142 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31145 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
31146 target operand. A new splitter adds the clobber statement in case
31147 the target operand is dead anyway.
31149 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
31151 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
31152 to age-old versions of binutils and glibc.
31154 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
31156 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
31158 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
31160 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
31162 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
31164 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
31165 requirement for binutils 2.13.
31167 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
31169 * combine.c (try_combine): Delete redundant i1 test. Call
31170 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
31172 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
31174 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
31175 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
31177 <riscv64-*-elf>: Re-arrange section
31178 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
31179 <riscv32-*-linux>: Likewise.
31180 <riscv64-*-elf>: Likewise
31181 <riscv64-*-linux>: Likewise.
31183 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
31186 * aarch64.opt(verbose-cost-dump): Fix typo.
31188 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
31191 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
31192 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
31194 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
31196 * reload.c (find_reloads): When reloading a nonoffsettable address,
31197 use RELOAD_OTHER for it and its address reloads.
31199 PR rtl-optimization/79910
31200 * combine.c (record_used_regs): New static function.
31201 (try_combine): Handle situations where there is an additional
31202 instruction between I2 and I3 which needs to have a LOG_LINK
31205 2017-03-17 Jeff Law <law@redhat.com>
31207 PR tree-optimization/71437
31208 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
31209 conditional in the hash table first.
31210 (vrp_dom_walker::before_dom_children): Extract condition from
31211 ASSERT_EXPR. Record condition, its inverion and any implied
31212 conditions as well.
31214 2017-03-17 Marek Polacek <polacek@redhat.com>
31215 Markus Trippelsdorf <markus@trippelsdorf.de>
31217 PR tree-optimization/80079
31218 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
31221 2017-03-17 Richard Biener <rguenther@suse.de>
31223 PR middle-end/80075
31224 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
31225 Properly verify the LHS before the RHS possibly claims to be
31227 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
31230 2017-03-17 Martin Jambor <mjambor@suse.cz>
31232 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
31233 (List of -O2 options): Likewise.
31234 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
31237 2017-03-17 Tom de Vries <tom@codesourcery.com>
31239 * gcov-dump.c (print_usage): Print bug_report_url.
31241 2017-03-17 Richard Biener <rguenther@suse.de>
31243 PR middle-end/80050
31244 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
31245 (parser::peek): Likewise.
31247 2017-03-17 Richard Biener <rguenther@suse.de>
31249 PR tree-optimization/80048
31250 * sese.c (free_sese_info): Properly release rename_map and
31251 copied_bb_map elements.
31253 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
31255 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
31256 Add linked-list forward and backlinks. Insert on
31257 construction, remove on destruction.
31258 (class pass_store_merging): Add m_stores_head field.
31259 (pass_store_merging::terminate_and_process_all_chains):
31260 Iterate over m_stores_head list.
31261 (pass_store_merging::terminate_all_aliasing_chains):
31263 (pass_store_merging::execute): Check for debug stmts first.
31264 Push new chains onto the m_stores_head stack.
31266 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
31269 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
31270 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
31271 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
31273 2017-03-16 Jeff Law <law@redhat.com>
31275 PR tree-optimization/71437
31276 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
31277 member function. Implementation moved into after_dom_children
31278 member function and into the threader's thread_outgoing_edges
31280 (dom_opt_dom_walker::after_dom_children): Simplify by moving
31281 some code into new thread_outgoing_edges.
31282 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
31283 definition. Simplify marker handling (do it here). Assume we always
31284 have the available expression and the const/copies tables.
31285 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
31287 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
31288 * tree-vrp.c (equiv_stack): No longer file scoped.
31289 (vrp_dom_walker): New class.
31290 (vrp_dom_walker::before_dom_children): New member function.
31291 (vrp_dom_walker::after_dom_children): Likewise.
31292 (identify_jump_threads): Setup domwalker. Use it rather than
31293 walking edges in a random order by hand. Simplify setup/finalization.
31294 (finalize_jump_threads): Remove.
31295 (vrp_finalize): Do not call identify_jump_threads here.
31296 (execute_vrp): Do it here instead and call thread_through_all_blocks
31299 PR tree-optimization/71437
31300 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
31302 (simplify_stmt_for_jump_threading): Add basic_block argument. All
31304 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
31305 (dom_opt_dom_walker::thread_across_edge): Remove
31306 handle_dominating_asserts argument. All callers changed.
31307 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
31308 changes. Remove calls to lhs_of_dominating_assert. Other
31309 uses of handle_dominating_asserts turn into unconditional code
31310 (simplify_control_stmt_condition_1): Likewise.
31311 (simplify_control_stmt_condition): Likewise.
31312 (thread_through_normal_block, thread_across_edge): Likewise.
31313 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
31314 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
31315 object if it is not an SSA_NAME.
31316 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
31317 before calling into the VRP specific simplifiers.
31318 (identify_jump_threads): Remove handle_dominating_asserts
31321 2017-03-16 Jakub Jelinek <jakub@redhat.com>
31324 * tree-diagnostic.c (default_tree_printer): No longer static.
31325 * tree-diagnostic.h (default_tree_printer): New prototype.
31327 2017-03-16 Tamar Christina <tamar.christina@arm.com>
31329 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
31330 Change ins into fmov.
31332 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31334 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
31335 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
31336 Use h_con constraint for operand 1.
31337 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
31338 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
31340 2017-03-15 Jeff Law <law@redhat.com>
31342 PR tree-optimization/71437
31343 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
31344 (record_temporary_equivalences): Use it.
31346 PR tree-optimization/71437
31347 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
31348 tree-ssa-scopedtables.
31349 (lookup_avail_expr, build_and_record_new_cond): Likewise.
31350 (record_conditions, record_cond, vuse_eq): Likewise.
31351 (record_edge_info): Adjust to API tweak of record_conditions.
31352 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
31353 (record_temporary_equivalences, optimize_stmt): Likewise.
31354 (eliminate_redundant_computations): Likewise.
31355 (record_equivalences_from_stmt): Likewise.
31356 * tree-ssa-scopedtables.c: Include options.h and params.h.
31357 (vuse_eq): New function, moved from tree-ssa-dom.c
31358 (build_and_record_new_cond): Likewise.
31359 (record_conditions): Likewise. Accept vector of conditions rather
31360 than edge_equivalence structure for first argument.
31361 for the first argument.
31362 (avail_exprs_stack::lookup_avail_expr): New member function, moved
31363 from tree-ssa-dom.c.
31364 (avail_exprs_stack::record_cond): Likewise.
31365 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
31366 from tree-ssa-dom.c.
31367 (avail_exprs_stack): Add new member functions lookup_avail_expr
31369 (record_conditions): Declare.
31371 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
31374 * lra-constraints.c (process_alt_operands): Increase reject for
31375 reloading an input/output operand.
31377 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
31380 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
31381 insns to convert from signed/unsigned char/short to IEEE 128-bit
31383 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
31385 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
31388 * config/i386/i386.c (ix86_vector_duplicate_value): Create
31389 subreg of inner mode for values already in registers.
31391 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
31393 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
31394 iteration reg is used after the loop.
31396 2017-03-14 Martin Sebor <msebor@redhat.com>
31398 PR tree-optimization/79800
31399 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
31400 precision in negative-positive range.
31401 (format_floating): Call non-const overload with adjusted precision.
31403 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
31406 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
31409 2017-03-14 Martin Sebor <msebor@redhat.com>
31411 PR middle-end/80020
31412 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
31413 * builtins.def (aligned_alloc): Use it.
31416 * Makefile.in (GTFILES): Add calls.c.
31417 * calls.c: Include "gt-calls.h".
31419 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
31421 PR rtl-optimization/79728
31422 * regs.h (struct target_regs): New field
31423 x_contains_allocatable_regs_of_mode.
31424 (contains_allocatable_regs_of_mode): New macro.
31425 * reginfo.c (init_reg_sets_1): Initialize it, and change
31426 contains_reg_of_mode so it includes global regs as well.
31427 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
31428 rather than contains_regs_of_mode.
31430 2017-03-14 Martin Liska <mliska@suse.cz>
31432 * doc/invoke.texi: Document options that can't be combined with
31433 -fcheck-pointer-bounds.
31435 2017-03-14 Martin Liska <mliska@suse.cz>
31437 PR middle-end/79831
31438 * doc/invoke.texi (-Wchkp): Document the option.
31440 2017-03-14 Martin Liska <mliska@suse.cz>
31442 * Makefile.in: Install gcov-dump.
31444 2017-03-14 Martin Liska <mliska@suse.cz>
31446 * multiple_target.c (expand_target_clones): Bail out for
31447 an invalid attribute.
31449 2017-03-14 Richard Biener <rguenther@suse.de>
31451 * alias.c (struct alias_set_entry): Pack properly.
31452 * cfgloop.h (struct loop): Likewise.
31453 * cse.c (struct set): Likewise.
31454 * ipa-utils.c (struct searchc_env): Likewise.
31455 * loop-invariant.c (struct invariant): Likewise.
31456 * lra-remat.c (struct cand): Likewise.
31457 * recog.c (struct change_t): Likewise.
31458 * rtl.h (struct address_info): Likewise.
31459 * symbol-summary.h (function_summary): Likewise.
31460 * tree-loop-distribution.c (struct partition): Likewise.
31461 * tree-object-size.c (struct object_size_info): Likewise.
31462 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
31463 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
31464 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
31465 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
31466 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
31467 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
31468 (struct _stmt_vec_info): Likewise.
31470 2017-03-14 Martin Liska <mliska@suse.cz>
31473 * multiple_target.c (create_dispatcher_calls): Check that
31474 a target can create a function dispatcher.
31476 2017-03-14 Martin Liska <mliska@suse.cz>
31479 * multiple_target.c (expand_target_clones): Drop local.local
31480 flag for default implementation.
31482 2017-03-14 Richard Biener <rguenther@suse.de>
31484 PR tree-optimization/80030
31485 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
31487 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
31489 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
31490 gcc_fallthrough() instead of __attribute__((fallthrough));
31492 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
31494 * doc/gcc.texi: Remove "up" link to (DIR).
31495 * doc/gccint.texi: Ditto.
31497 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
31499 * doc/install.texi (Specific) <avr>: Remove reference to
31502 2017-03-13 Jeff Law <law@redhat.com>
31504 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
31505 attribute rather than comments.
31507 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
31508 match_scratch operand is highest.
31510 2017-03-13 Martin Liska <mliska@suse.cz>
31512 PR middle-end/78339
31513 * ipa-pure-const.c (warn_function_noreturn): If the declarations
31514 is a CHKP clone, use original declaration.
31516 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31518 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
31519 (arc_conditional_register_usage): Use a different allocation order
31520 when optimizing for size.
31521 * common/config/arc/arc-common.c (arc_option_optimization_table):
31522 Section anchors default on when optimizing for size.
31524 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31526 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
31528 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31530 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
31531 * config/arc/arc.md (cpu_facility): Add cd variant.
31532 (*movqi_insn): Add code density variant.
31533 (*movhi_insn): Likewise.
31534 (*movqi_insn): Likewise.
31535 (*addsi3_mixed): Likewise.
31536 (subsi3_insn): Likewise.
31538 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31540 * config/arc/arc.md (movsi_cond_exec): Update constraint.
31542 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
31544 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
31545 expressions with MINUS and UNARY ops.
31547 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31550 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
31552 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
31553 between vec_select and vector argument.
31554 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
31555 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
31556 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
31557 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
31558 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
31559 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
31561 2017-03-13 Richard Biener <rguenther@suse.de>
31564 * params.def (vect-max-peeling-for-alignment): Fix typo.
31566 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
31568 * doc/install.texi (Specific) <mips-*-*>: Remove description of
31569 issue that only occurred with binutils below 2.18.
31571 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
31573 * doc/install.texi (Specific) <cris-axis-elf>: No longer
31574 refer to binutils 2.11/2.12 minimum.
31576 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
31578 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
31579 ftp.kernel.org and simplify binutils requirement.
31581 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
31583 * doc/invoke.texi (Warning Options): Fix spelling of link-time
31585 (Optimize Options): Ditto. Also remove redundancy.
31587 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31589 PR translation/79848
31590 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
31592 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
31593 to G_ to avoid double translation.
31595 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31597 PR translation/79923
31598 * auto-profile.c (get_combined_location): Convert leading
31599 character of diagnostics to lower case and remove trailing period.
31600 (read_profile): Likewise for various diagnostics.
31601 * config/arm/arm.c (arm_option_override): Remove trailing period
31602 from various diagnostics.
31603 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
31604 (msp430_expand_delay_cycles): Likewise.
31606 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31609 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
31610 full command-line argument, rather than just "str".
31611 (aarch64_validate_march): Likewise.
31612 (aarch64_validate_mtune): Likewise.
31614 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
31616 PR rtl-optimization/78911
31617 * lra-assigns.c (must_not_spill_p): New function.
31618 (spill_for): Use it.
31620 2017-03-10 Jakub Jelinek <jakub@redhat.com>
31622 PR tree-optimization/79981
31623 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
31624 ATOMIC_COMPARE_EXCHANGE ifn result.
31625 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
31626 IFN_ATOMIC_COMPARE_EXCHANGE.
31628 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31631 * opts.c (parse_sanitizer_options): Add missing question mark to
31632 "did you mean" message.
31634 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31636 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
31638 (VMULEUH_UNS): Likewise.
31639 (VMULOUB_UNS): Likewise.
31640 (VMULOUH_UNS): Likewise.
31641 * config/rs6000/rs6000.c (builtin_function_type): Remove
31642 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
31644 2017-03-10 David Malcolm <dmalcolm@redhat.com>
31647 * read-rtl-function.c (function_reader::read_rtx_operand): Update
31648 x with result of extra_parsing_for_operand_code_0.
31649 (function_reader::extra_parsing_for_operand_code_0): Convert
31650 return type from void to rtx, returning x. When reading
31651 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
31652 larger size containing struct block_symbol.
31654 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
31656 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
31657 -mfloat128-hardware without -m64.
31659 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
31662 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
31663 entries to the case statement that marks unsigned arguments to
31664 overloaded functions.
31666 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
31668 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
31669 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
31671 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
31674 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
31675 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
31677 2017-03-10 Martin Liska <mliska@suse.cz>
31681 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
31682 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
31685 2017-03-10 Olivier Hainque <hainque@adacore.com>
31687 * tree-switch-conversion (array_value_type): Start by resetting
31688 candidate type to it's main variant.
31690 2017-03-10 Jakub Jelinek <jakub@redhat.com>
31692 PR rtl-optimization/79909
31693 * combine.c (try_combine): Use simplify_replace_rtx on individual
31694 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
31695 of the whole CALL_INSN_FUNCTION_USAGE.
31697 PR tree-optimization/79972
31698 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
31699 get_range_info on SSA_NAMEs. Formatting fixes.
31701 2017-03-10 Richard Biener <rguenther@suse.de>
31702 Jakub Jelinek <jakub@redhat.com>
31704 PR tree-optimization/77975
31705 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
31706 edge to be constant.
31707 (get_val_for): For constant x return it. Formatting fix.
31708 (loop_niter_by_eval): Avoid pointless looping if the next iteration
31709 would use the same bases as the current one.
31711 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31713 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
31714 instead of vec_select for V1TImode.
31715 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
31717 (VSX_LE_128): Add V1TI to this mode iterator.
31718 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
31719 (*vsx_le_perm_store_<mode>): Likewise.
31720 (pre-reload splitter for VSX stores): Likewise.
31721 (post-reload splitter for VSX stores): Likewise.
31722 (*vsx_xxpermdi2_le_<mode>): Likewise.
31723 (*vsx_lxvd2x2_le_<mode>): Likewise.
31724 (*vsx_stxvd2x2_le_<mode>): Likewise.
31726 2017-03-09 Michael Eager <eager@eagercon.com>
31728 Correct failures with --enable-checking=yes,rtl.
31730 * config/microblaze/microblaze.c (microblaze_expand_shift):
31731 Replace GET_CODE test with CONST_INT_P and INTVAL test with
31732 test for const0_rtx.
31733 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
31734 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
31736 2017-03-09 Richard Biener <rguenther@suse.de>
31738 PR tree-optimization/79977
31739 * graphite-scop-detection.c (scop_detection::merge_sese):
31740 Handle the case of extra exits to blocks dominating the entry.
31742 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
31744 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
31747 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
31749 PR rtl-optimization/79949
31750 * lra-constraints.c (process_alt_operands): Check memory when
31751 trying to predict a cycle. Print about the overall increase.
31753 2017-03-09 Richard Biener <rguenther@suse.de>
31755 PR middle-end/79971
31756 * gimple-expr.c (useless_type_conversion_p): Preserve
31757 TYPE_SATURATING for fixed-point types.
31759 2017-03-09 Richard Biener <rguenther@suse.de>
31762 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
31763 alignment of BLKmode params.
31765 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31768 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
31769 (VALL_NO_V2Q): Likewise.
31771 * config/aarch64/aarch64-simd.md
31772 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
31774 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
31775 VALL_NO_V2Q mode iterator.
31776 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
31778 2017-03-09 Martin Liska <mliska@suse.cz>
31780 PR tree-optimization/79631
31781 * tree-chkp-opt.c (chkp_is_constant_addr): Call
31782 tree_int_cst_sign_bit just for INTEGER constants.
31784 2017-03-09 Martin Liska <mliska@suse.cz>
31788 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
31791 2017-03-09 Marek Polacek <polacek@redhat.com>
31794 * tree.c (inchash::add_expr): Handle TREE_VEC.
31796 2017-03-09 Martin Liska <mliska@suse.cz>
31799 (chkp_narrow_size_and_offset): New function.
31800 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
31801 (void chkp_parse_bit_field_ref): New function.
31802 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
31803 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
31805 2017-03-09 Martin Liska <mliska@suse.cz>
31808 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
31809 (chkp_find_bounds_1): Remove gcc_unreachable.
31811 2017-03-09 Jakub Jelinek <jakub@redhat.com>
31814 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
31815 BUILT_IN_SYNC*, determine the access type from the size suffix and
31816 always build a MEM_REF with that type. Handle forgotten
31817 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
31820 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
31821 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
31822 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
31823 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
31824 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
31825 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
31826 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
31827 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
31828 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
31829 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
31830 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
31831 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
31832 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
31833 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
31834 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
31835 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
31836 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
31837 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
31838 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
31839 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
31840 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
31841 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
31842 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
31843 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
31844 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
31845 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
31846 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
31847 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
31848 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
31849 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
31850 definitions outside of __OPTIMIZE__ guarded section.
31853 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
31854 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
31855 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
31856 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
31859 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31861 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
31862 ("vfenez<mode>"): Add missing constraints.
31864 2017-03-08 Martin Sebor <msebor@redhat.com>
31867 * config/nds32/nds32.c (nds32_option_override):
31868 Fix misspelled diagnostic.
31870 2017-03-08 Jakub Jelinek <jakub@redhat.com>
31873 * gimplify.c (gimplify_omp_for): Replace index var in outer
31874 taskloop statement with an artificial variable and add
31875 OMP_CLAUSE_PRIVATE clause for it.
31877 2017-03-08 Richard Biener <rguenther@suse.de>
31879 PR tree-optimization/79955
31880 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
31881 for accesses that are completely outside of the variable.
31883 2017-03-08 Andrew Haley <aph@redhat.com>
31885 PR tree-optimization/79943
31886 * tree-ssa-loop-split.c (compute_new_first_bound): When
31887 calculating the new upper bound, (END-BEG) should be added, not
31890 2017-03-08 Jakub Jelinek <jakub@redhat.com>
31892 * config/avr/avr.md (setmemhi): Make sure match_dup
31893 operand number comes before match_scratch.
31895 2017-03-08 Richard Biener <rguenther@suse.de>
31897 PR tree-optimization/79920
31898 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
31899 with ncopies == 1 to ...
31900 (vect_transform_slp_perm_load): ... here. Properly compute
31901 all element loads by iterating VF times over the group. Do
31902 not handle ncopies (computed in a broken way) in
31903 vect_create_mask_and_perm.
31905 2017-03-08 Jakub Jelinek <jakub@redhat.com>
31908 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
31909 is a uniform vector, use uniform_vector_p return value instead of
31910 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
31912 2017-03-07 Marek Polacek <polacek@redhat.com>
31914 PR middle-end/79809
31915 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
31916 (alloca_call_type): Likewise.
31918 2017-03-07 Martin Liska <mliska@suse.cz>
31920 * gcov.c (process_args): Put comment to correct location.
31922 2017-03-07 Martin Liska <mliska@suse.cz>
31924 PR middle-end/68270
31925 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
31926 Use array_at_struct_end_p instead of DECL_CHAIN (field).
31927 (chkp_narrow_bounds_for_field): Likewise.
31928 (chkp_parse_array_and_component_ref): Pass one more argument to
31931 2017-03-07 Richard Biener <rguenther@suse.de>
31933 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
31936 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
31938 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
31939 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
31941 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31944 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
31945 to end of description.
31946 (PARAM_MAX_STORES_TO_MERGE): Likewise.
31948 2017-03-07 Jakub Jelinek <jakub@redhat.com>
31950 PR rtl-optimization/79901
31951 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
31953 (*avx512f_<code><mode>3<mask_name>): ... this.
31954 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
31955 iterator instead of VI8_AVX2_AVX512BW.
31957 PR rtl-optimization/79901
31958 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
31959 min/max expander, expand it using expand_vec_cond_expr.
31962 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
31965 2017-03-06 Jakub Jelinek <jakub@redhat.com>
31968 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
31969 to void * for PCH reasons.
31970 * dwarf2out.c (output_loc_operands, output_die): Cast
31971 v.val_vec.array to unsigned char *.
31973 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
31976 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
31979 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
31981 PR rtl-optimization/79571
31982 * lra-constraints.c (process_alt_operands): Calculate static
31983 reject and subtract it from overall when only addresses will be
31986 2017-03-06 Julia Koval <julia.koval@intel.com>
31989 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
31990 incoming stack boundary to 128 for 64-bit targets.
31992 2017-03-06 Richard Biener <rguenther@suse.de>
31994 PR tree-optimization/79894
31995 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
31996 to NULL after folding it.
31998 2017-03-06 Richard Biener <rguenther@suse.de>
32000 PR tree-optimization/79824
32001 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
32002 check disabling peeling for gaps.
32004 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
32006 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
32007 attributes): Document gettimeofday.
32009 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
32011 * config/s390/s390.c (s390_option_override_internal): Set
32012 PARAM_MIN_VECT_LOOP_BOUND
32014 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
32016 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
32017 * config/s390/s390.md: Likewise.
32019 2017-03-06 Jakub Jelinek <jakub@redhat.com>
32022 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
32023 (<avx2_avx512>_perm<mode>): Rename to ...
32024 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
32026 (<avx512>_perm<mode>_mask): Rename to ...
32027 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
32029 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
32030 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
32031 instead of VI8F_256_512.
32032 (avx512f_perm<mode>): New define_expand.
32033 (avx512f_perm<mode>_mask): Likewise.
32034 (avx512f_perm<mode>_1<mask_name>): New define_insn.
32035 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
32037 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32039 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
32040 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
32042 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
32044 2017-03-06 Martin Liska <mliska@suse.cz>
32047 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
32048 when having a SSA NAME w/o VAR_DECL assigned to it.
32050 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32052 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
32053 msa_dpsub_<su>_d): Fix MODE for vec_select.
32055 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32057 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
32059 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
32061 2017-03-06 Richard Biener <rguenther@suse.de>
32063 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
32064 * plugin.c (register_plugin_info): Likewise.
32065 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
32067 2017-03-05 Jakub Jelinek <jakub@redhat.com>
32069 * config/i386/sse.md (sse_storehps, sse_storelps,
32070 avx_<castmode><avxsizesuffix>_<castmode>,
32071 avx512f_<castmode><avxsizesuffix>_<castmode>,
32072 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
32073 in condition that at least one operand is not a MEM.
32075 2017-03-03 Jakub Jelinek <jakub@redhat.com>
32077 PR middle-end/79805
32078 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
32079 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
32081 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
32082 gimple_call_nothrow_p flag based on whether original builtin can throw.
32083 If it can, emit following stmts on the fallthrough edge.
32084 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
32085 don't create new bb if inserting just debug stmts on the edge, try to
32086 insert them on the fallthru bb or just reset debug stmts.
32088 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
32091 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
32092 restore recog_data (including the operand rtxes inside it) around
32093 the call to get_insn_template.
32095 2017-03-03 Martin Sebor <msebor@redhat.com>
32097 PR tree-optimization/79699
32098 * context.c (context::~context): Free MPFR caches to avoid
32099 a memory leak on program exit.
32101 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32103 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
32104 Use wide_int::ulow () instead of .elt (0).
32106 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
32108 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
32109 (*pushxf): Limit oF constraint to 32bit targets and add oC
32110 constraint for 64bit targets.
32111 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
32112 (*pushdf): Change rmF constraint to rmC.
32114 2017-03-03 Martin Liska <mliska@suse.cz>
32116 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
32117 Remove unused variable.
32119 2017-03-03 Jakub Jelinek <jakub@redhat.com>
32122 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
32123 is a memory operand, increase num_memory.
32124 (ix86_expand_args_builtin): Likewise.
32126 2017-03-03 Jan Hubicka <jh@suse.cz>
32129 * ipa-devirt.c (maybe_record_node): Properly handle
32130 __cxa_pure_virtual visibility.
32132 2017-03-03 Martin Liska <mliska@suse.cz>
32134 PR tree-optimization/79803
32135 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
32137 (pass_loop_prefetch::execute): Disabled optimization if an
32138 assumption about L1 cache size is not met.
32140 2017-03-03 Martin Liska <mliska@suse.cz>
32142 PR rtl-optimization/79574
32143 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
32144 (hash_scan_set): Likewise.
32145 (dump_hash_table): Likewise.
32146 (hoist_code): Likewise.
32148 2017-03-03 Richard Biener <rguenther@suse.de>
32150 * fixed-value.c (fixed_from_string): Restore use of elt (1)
32151 in place of uhigh ().
32152 (fixed_convert_from_real): Likewise.
32154 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
32157 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
32159 2017-03-03 Richard Biener <rguenther@suse.de>
32161 PR middle-end/79818
32162 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
32163 TYPE_OVERFLOW_UNDEFINED check.
32165 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32167 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
32169 (vector_ae_<mode>_p): Likewise.
32170 (vector_nez_<mode>_p): Likewise.
32171 (vector_ne_v2di_p): Likewise.
32172 (vector_ae_v2di_p): Likewise.
32173 (vector_ne_<mode>_p): Likewise.
32174 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
32176 (vsx_tsqrt<mode>2_fe): Likewise.
32178 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
32181 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
32183 2017-03-02 Jakub Jelinek <jakub@redhat.com>
32185 PR rtl-optimization/79780
32186 * cprop.c (one_cprop_pass): When second and further conditional trap
32187 in a single basic block is turned into an unconditional trap, turn it
32188 into a deleted note to avoid RTL verification failures.
32190 2017-03-02 Richard Biener <rguenther@suse.de>
32192 * fold-const.c (const_binop): Use ulow () instead of elt (0).
32194 2017-03-02 Richard Biener <rguenther@suse.de>
32196 PR tree-optimization/79345
32198 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
32199 param and abort the walk, returning -1 if it is hit.
32200 (walk_aliased_vdefs): Take a limit param and pass it on.
32201 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
32202 defaulting to 0 and return a signed int.
32203 * tree-ssa-uninit.c (struct check_defs_data): New struct.
32204 (check_defs): New helper.
32205 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
32206 about uninitialized memory.
32207 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
32208 bogus uninitialized warning.
32209 (fixed_convert_from_real): Likewise.
32211 2017-03-02 Bin Cheng <bin.cheng@arm.com>
32213 PR tree-optimization/66768
32214 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
32215 iv_use if base object can't be determined.
32217 2017-03-02 Jakub Jelinek <jakub@redhat.com>
32219 PR tree-optimization/79345
32220 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
32221 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
32222 (get_pattern_stats): Initialize it.
32223 * genemit.c (gen_expand): Verify match_scratch numbers come after
32224 match_operand/match_dup numbers.
32225 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
32226 match_scratch numbers.
32227 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
32229 * config/s390/s390.md (trunctdsd2): Likewise.
32231 2017-03-02 Richard Biener <rguenther@suse.de>
32233 * wide-int.h (wide_int_storage::operator=): Implement in terms
32236 2017-03-02 Richard Biener <rguenther@suse.de>
32238 PR tree-optimization/79777
32239 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
32240 the to insert expression to sth existing.
32242 2017-03-01 Martin Sebor <msebor@redhat.com>
32244 PR middle-end/79692
32245 * gimple-ssa-sprintf.c
32246 (directive::known_width_and_precision): New function.
32247 (format_integer): Use it.
32248 (get_mpfr_format_length): Consider the full range of precision
32249 when computing %g output with the # flag. Set the likely byte
32250 count to 3 rather than 1 when precision is indeterminate.
32251 (format_floating): Correct the lower bound of precision.
32253 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32255 * doc/invoke.texi: Document default code model for 64-bit Linux.
32257 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32260 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
32261 udiv rather than div since input pattern is unsigned.
32263 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
32265 * config/i386/i386.c (print_reg): Warn for values of
32266 unsupported size in integer register.
32268 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
32271 * config/rs6000/predicates.md (current_file_function_operand): Do
32272 not allow self calls to be local if the function is replaceable.
32274 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
32277 * config/rs6000/altivec.h (vec_ctz and others): Change the
32278 preprocessor macro that controls conditional compilation from
32279 _ARCH_PWR9 to __POWER9_VECTOR__.
32280 (vec_all_ne): Change parameterization of __altivec_scalar_pred
32281 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
32282 control (instead of _ARCH_PWR9 control) so that template
32283 definition uses power9-specific function.
32284 (vec_any_eq): Likewise.
32285 (vec_all_ne): Change macro definition to use a power9-specific
32286 expansion under #ifdef __POWER9_VECTOR__ control (instead of
32287 _ARCH_PWR9 control).
32288 (vec_any_eq) Likewise.
32289 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
32290 expansion for CMPNEF to remove support for xvcmpnesp instruction.
32291 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
32292 support for xvcmpnedp instruction.
32293 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
32294 macro expansion so that Power9 implementation of vec_all_ne does
32295 not use the AltiVec predicate framework.
32296 (VCMPNEH_P): Likewise.
32297 (VCMPNEW_P): Likewise.
32298 (VCMPNED_P): Likewise.
32299 (VCMPNEFP_P): Likewise.
32300 (VCMPNEDP_P): Likewise.
32301 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
32302 implementation of vec_any_eq to not use AltiVec predicate
32304 (VCMPAEH_P): Likewise.
32305 (VCMPAEW_P): Likewise.
32306 (VCMPAED_P): Likewise.
32307 (VCMPAEFP_P): Likewise.
32308 (VCMPAEDP_P): Likewise.
32309 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
32310 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
32311 not use the AltiVec predicate framework.
32312 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
32313 of vec_any_eq to not use AltiVec predicate framework.
32314 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
32315 support for predefined __POWER9_VECTOR__ macro to indicate that
32316 Power9 instruction selection is enabled.
32317 (altivec_overloaded_builtins): Remove extraneous
32318 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
32319 function argument types RS6000_BTI_bool_V16QI and
32320 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
32321 entry for overloaded function argument types RS6000_BTI_bool_V4SI
32322 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
32323 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
32324 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
32325 Power9 for implementations of vec_cmpne. Change the signature for
32326 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
32327 (representing vec_all_ne) to remove the previously described first
32328 argument of type RS6000_BTI_INTSI, as this was an artifact of
32329 reliance on the AltiVec predicate framework, which is no longer
32330 used in the implementation of these functions. Add
32331 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
32332 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
32333 since, unlike the AltiVec predicate framework implementation, we
32334 do not share function descriptors between vec_alle and vec_anyeq.
32335 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
32336 set of modes that receive special treatment even when
32337 TARGET_P9_VECTOR is true. The special treatment emits code that
32338 does not depend on Power9 instructions.
32339 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
32340 define_expand to not rely on AltiVec predicate framework.
32341 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
32343 (vector_ne_v2di_p): Change this define_expand to not rely on
32344 AltiVec predicate framework.
32345 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
32347 (vector_ne_<mode>_p): Change this define_expand to not rely on
32348 AltiVec predicate framework.
32349 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
32351 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
32352 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
32353 define_insn pattern.
32354 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
32355 define_insn pattern because the xvcmpne<VSs>. instruction is not
32357 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
32358 instruction is not supported.
32360 2017-03-01 Jakub Jelinek <jakub@redhat.com>
32362 * config/nvptx/nvptx.c: Include intl.h.
32364 2017-03-01 Martin Jambor <mjambor@suse.cz>
32367 * ipa-prop.h (ipa_bits): Removed field known.
32368 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
32369 to pointers. Adjusted their comments to warn about their sharing.
32370 (ipcp_transformation_summary): Change bits to a vector of pointers.
32371 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
32372 (ipa_get_ipa_bits_for_value): Declare.
32373 * tree-vrp.h (value_range): Mark as GTY((for_user)).
32374 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
32375 (ipa_bits_hash_table): Likewise.
32376 (ipa_vr_ggc_hash_traits): Likewise.
32377 (ipa_vr_hash_table): Likewise.
32378 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
32379 being pointers and vr_known being removed.
32380 (ipa_set_jf_unknown): Likewise.
32381 (ipa_get_ipa_bits_for_value): New function.
32382 (ipa_set_jfunc_bits): Likewise.
32383 (ipa_get_value_range): New overloaded functions.
32384 (ipa_set_jfunc_vr): Likewise.
32385 (ipa_compute_jump_functions_for_edge): Use the above functions to
32386 construct bits and vr parts of jump functions.
32387 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
32388 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
32390 (ipcp_grow_transformations_if_necessary): Also allocate
32391 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
32393 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
32394 them. Fix too long lines.
32395 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
32396 vr_known being removed.
32397 (ipa_read_jump_function): Use new setter functions to construct bits
32398 and vr parts of jump functions or set them to NULL.
32399 (write_ipcp_transformation_info): Adjust for bits being pointers.
32400 (read_ipcp_transformation_info): Likewise.
32401 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
32403 Include gt-ipa-prop.h.
32404 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
32406 (ipcp_store_bits_results): Likewise.
32407 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
32408 Do not write to existing jump functions but use a temporary instead.
32410 2017-03-01 Jakub Jelinek <jakub@redhat.com>
32413 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
32414 attempt to use its first operand as BIT_FIELD_REF base.
32416 2017-03-01 Richard Biener <rguenther@suse.de>
32418 PR middle-end/79721
32419 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
32420 interpolating formula in wrapping arithmetic.
32421 (chrec_apply): Convert chrec_evaluate return value to wanted type.
32423 2017-03-01 Jakub Jelinek <jakub@redhat.com>
32425 PR tree-optimization/79734
32426 * tree-vect-generic.c (expand_vector_condition): Optimize
32427 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
32428 Handle VEC_COND_EXPR where comparison has different inner width from
32429 type's inner width.
32431 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
32433 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
32434 markup, and similar issues. Remove @opindex entries for things
32435 that aren't options. Add missing -mmpy-option entries.
32437 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32439 PR tree-optimization/79737
32440 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
32441 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
32442 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
32443 instead of byte_size. Formatting fix.
32444 (shift_bytes_in_array_right): Formatting fix.
32446 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
32449 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
32450 condition on optimize for the leaf function test.
32452 2017-02-28 Martin Liska <mliska@suse.cz>
32455 * read-rtl-function.c (function_reader::handle_unknown_directive):
32456 Bail out when one uses -flto.
32458 2017-02-28 Martin Liska <mliska@suse.cz>
32460 * common.opt: Replace space with tabular for options of <number>
32462 * config/i386/i386.opt: Show <number> value for
32463 -mlarge-data-threshold.
32464 * opts.c (print_filtered_help): Do not display number in hexadecimal
32467 2017-02-28 Martin Liska <mliska@suse.cz>
32469 * common.opt: Fix --help=option -Q for options which are of
32472 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
32474 * config/i386/i386.c (print_reg): Error out for values
32475 of 8-bit size in invalid integer register.
32477 2017-02-28 Martin Sebor <msebor@redhat.com>
32479 PR tree-optimization/79691
32480 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
32482 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32485 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
32486 gcc_unreachable with output_operand_lossage.
32488 2017-02-28 Richard Biener <rguenther@suse.de>
32490 PR tree-optimization/79740
32491 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
32493 (visit_nary_op): Insert the nary into the hashtable if we
32494 pattern-matched sth.
32495 * tree-ssa-pre.c (eliminate_insert): Robustify.
32497 2017-02-28 Richard Biener <rguenther@suse.de>
32499 PR middle-end/79731
32500 * fold-const.c (decode_field_reference): Reject out-of-bound
32503 2017-02-28 Jakub Jelinek <jakub@redhat.com>
32505 * config/i386/i386.c: Include intl.h.
32506 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
32507 instead of just cond ? "..." : "...".
32508 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
32509 * coverage.c (read_counts_file): Likewise.
32510 * omp-offload.c: Include intl.h.
32511 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
32512 of just cond ? "..." : "...".
32513 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
32514 of just cond ? "..." : "...".
32516 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
32519 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
32521 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
32522 'tune for' CPU name.
32523 * config/arm/arm-cpu-data.h: Regenerated.
32525 2017-02-28 Richard Biener <rguenther@suse.de>
32527 PR tree-optimization/79732
32528 * tree-inline.c (expand_call_inline): Do not shadow var.
32530 2017-02-28 Richard Biener <rguenther@suse.de>
32532 PR tree-optimization/79723
32533 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
32534 address-space properly.
32536 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
32538 * doc/optinfo.texi (Optimization groups): Fix option used for
32540 * doc/invoke.texi (-fopt-info): Document "omp".
32541 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
32542 (OPTGROUP_ALL): Add OPTGROUP_OMP.
32543 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
32544 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
32545 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
32547 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
32549 * dumpfile.c (optgroup_options): Instead of "openmp", associate
32550 OPTGROUP_OMP with "omp".
32552 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
32555 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
32556 for arithmetic shift of unsigned V2DI.
32558 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
32560 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
32561 arc/linux.h headers.
32562 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
32563 (LINK_SPEC): Likewise.
32564 (ARC_TLS_EXTRA_START_SPEC): Likewise.
32565 (EXTRA_SPECS): Likewise.
32566 (STARTFILE_SPEC): Likewise.
32567 (ENDFILE_SPEC): Likewise.
32568 (LIB_SPEC): Likewise.
32569 (TARGET_SDATA_DEFAULT): Likewise.
32570 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
32571 (MULTILIB_DEFAULTS): Likewise.
32572 (DWARF2_UNWIND_INFO): Likewise.
32573 * config/arc/big.h: New file.
32574 * config/arc/elf.h: Likewise.
32575 * config/arc/linux.h: Likewise.
32576 * config/arc/t-uClibc: Remove.
32578 2017-02-27 Bin Cheng <bin.cheng@arm.com>
32580 PR tree-optimization/77536
32581 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
32582 (tree_transform_and_unroll_loop): Use above function to compute the
32583 estimated niter of unrolled loop and use it when scaling profile.
32584 Also use count info rather than frequency if it's non-zero.
32585 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
32586 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
32587 (vect_transform_loop): Call above function.
32589 2017-02-27 Richard Biener <rguenther@suse.de>
32591 PR tree-optimization/45397
32592 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
32593 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
32594 (visit_nary_op): Add pattern matching for CSEing sign-changed
32595 or truncated operations with wider ones.
32597 2017-02-27 Richard Biener <rguenther@suse.de>
32599 PR tree-optimization/79690
32600 * tree-vect-stmts.c (vectorizable_store): Use vector type
32601 built from the DR with address-space.
32603 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
32605 * doc/invoke.texi (Optimize Options): Refine the description
32606 of asan-use-after-return.
32608 2017-02-25 Alan Modra <amodra@gmail.com>
32610 PR rtl-optimization/79584
32611 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
32612 base, not ad->base_term, the reg within base. Remove assertion
32613 that ad->base == ad->base_term. Replace gen_int_mode using
32614 bogus mode with const0_rtx.
32616 2017-02-25 Jakub Jelinek <jakub@redhat.com>
32618 PR middle-end/79396
32619 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
32620 FMA_EXPR like tcc_binary or tcc_unary.
32622 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
32625 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
32627 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
32628 (output_loc_operands): Handle DW_OP_call_ref and
32629 DW_OP_GNU_variable_value.
32630 (struct variable_value_struct): New type.
32631 (struct variable_value_hasher): Likewise.
32632 (variable_value_hash): New variable.
32633 (string_types): Remove.
32634 (copy_loc_descr): New function.
32635 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
32636 (prepend_loc_descr_to_each): New function.
32637 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
32638 instead of add_loc_descr_to_each if the first argument is single
32639 location list and the second has multiple.
32640 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
32641 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
32642 when looking for variable value which doesn't have other location info.
32643 (loc_list_from_tree): Formatting fix.
32644 (gen_array_type_die): Simplify DW_AT_string_length handling.
32645 (adjust_string_types): Remove.
32646 (gen_subprogram_die): Don't call adjust_string_types nor test/set
32647 string_types. Call resolve_variable_values.
32648 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
32649 (resolve_addr_in_expr): Likewise. Add A argument.
32650 (copy_deref_exprloc): Remove deref argument. Adjust for the
32651 original expression being DW_OP_GNU_variable_value with optionally
32652 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
32653 optionally after it.
32654 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
32655 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
32656 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
32657 (variable_value_hasher::hash, variable_value_hasher::equal): New
32659 (resolve_variable_value_in_expr, resolve_variable_value,
32660 resolve_variable_values, note_variable_value_in_expr,
32661 note_variable_value): New functions.
32662 (dwarf2out_early_finish): Call note_variable_value on all toplevel
32665 2017-02-24 Jakub Jelinek <jakub@redhat.com>
32668 * opts.h (handle_generated_option): Add GENERATED_P argument.
32669 * opts-common.c (handle_option): Adjust function comment.
32670 (handle_generated_option): Add GENERATED_P argument, pass it to
32672 (control_warning_option): Pass false to handle_generated_option
32674 * opts.c (maybe_default_option): Pass true to handle_generated_option
32676 * optc-gen.awk: Likewise.
32678 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32680 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
32681 a REG, look at the REG it is a SUBREG of.
32682 (splitter for cmpeqsi_t): Ditto.
32684 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32686 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
32687 the special USEs with the pattern of the insn, not the insn itself.
32689 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
32692 * doc/invoke.texi: Document -mload-store-pairs.
32694 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32695 Sandra Loosemore <sandra@codesourcery.com>
32697 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
32698 argument isn't a CONST_INT.
32699 (nios2_alternate_compare_const): Assert op is a CONST_INT.
32700 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
32701 (nios2_validate_compare): Bypass alternate compare logic if *op2
32702 is not a CONST_INT.
32703 (ldstwm_operation_p): Return false if first_base is not a REG or
32704 if first_offset is not a CONST_INT.
32706 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32708 * config/cris/cris.md: Use correct operand in a define_peephole2.
32710 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32712 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
32714 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
32716 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
32717 this_insn if it is an INSN or JUMP_INSN.
32718 (force_offsettable): Look at base, not at addr.
32719 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
32720 on things that aren't necessarily CONST_INTs.
32722 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
32724 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
32725 -mfpmath=sse is the default also for x86-32 targets with SSE2
32726 instruction set when @option{-ffast-math} is enabled
32728 2017-02-24 Jeff Law <law@redhat.com>
32730 PR rtl-optimizatoin/79286
32731 * ira.c (update_equiv_regs): Drop may_trap_p exception to
32734 2017-02-24 Richard Biener <rguenther@suse.de>
32736 PR tree-optimization/79389
32737 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
32740 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
32742 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
32743 function comment to reflect reality.
32744 (loop_exits_before_overflow): Fix typo in function description.
32746 2017-02-24 Richard Biener <rguenther@suse.de>
32748 PR tree-optimization/79389
32749 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
32750 properly that a threading opportunity exists. Detect conditional
32751 copy/constant propagation opportunities.
32753 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
32755 * config/visium/visium.md (type): Add trap.
32756 (b): New mode attribute.
32757 (*btst): Rename into...
32758 (*btst<mode>): ...this and adjust.
32759 (*cbranchsi4_btst_insn): Rename into...
32760 (*cbranch<mode>4_btst_insn): ...this and adjust.
32761 (trap): New define_insn.
32763 2017-02-23 Jakub Jelinek <jakub@redhat.com>
32765 PR tree-optimization/79389
32766 * ifcvt.c (struct noce_if_info): Add rev_cond field.
32767 (noce_reversed_cond_code): New function.
32768 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
32769 reversed_comparison_code. Formatting fix.
32770 (noce_try_store_flag): Test rev_cond != NULL in addition to
32771 reversed_comparison_code.
32772 (noce_try_store_flag_constants): Likewise.
32773 (noce_try_store_flag_mask): Likewise.
32774 (noce_try_addcc): Use rev_cond if non-NULL instead of
32775 reversed_comparison_code.
32776 (noce_try_cmove_arith): Likewise. Formatting fixes.
32777 (noce_try_minmax, noce_try_abs): Clear rev_cond.
32778 (noce_find_if_block): Initialize rev_cond.
32779 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
32780 instead of false as last argument never attempt to reverse it
32783 2017-02-23 Bin Cheng <bin.cheng@arm.com>
32785 PR tree-optimization/79663
32786 * tree-predcom.c (combine_chains): Process refs in reverse order
32787 only for ZERO length chains, and add explaining comment.
32789 2017-02-23 Jeff Law <law@redhat.com>
32791 PR tree-optimization/79578
32792 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
32793 in call to operand_equal_p.
32795 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
32798 * config/i386/cpuid.h: Fix another undefined behavior.
32800 2017-02-23 Richard Biener <rguenther@suse.de>
32802 PR tree-optimization/79683
32803 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
32804 vector types for data-refs.
32806 2017-02-23 Martin Liska <mliska@suse.cz>
32808 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
32810 2017-02-23 Jakub Jelinek <jakub@redhat.com>
32812 PR middle-end/79665
32813 * internal-fn.c (get_range_pos_neg): Moved to ...
32814 * tree.c (get_range_pos_neg): ... here. No longer static.
32815 * tree.h (get_range_pos_neg): New prototype.
32816 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
32817 are known to be in between 0 and signed maximum inclusive, try to
32818 expand both unsigned and signed divmod and use the cheaper one from
32821 2017-02-22 Jeff Law <law@redhat.com>
32823 PR tree-optimization/79578
32824 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
32825 to compare base operands.
32827 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
32830 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
32831 gpc_reg_operand instead of fpr_reg_operand.
32833 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
32835 * config/mips/mips.c (mips_return_in_memory): Force FP
32836 vector types to be returned in memory for o32 ABI.
32838 2017-02-22 Jakub Jelinek <jakub@redhat.com>
32840 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
32841 instead of DW_TAG_member for static data member declarations and don't
32842 set no_linkage_name for static inline data members.
32843 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
32846 2017-02-22 Martin Liska <mliska@suse.cz>
32848 * doc/invoke.texi: Replace inequality signs with square brackets
32851 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
32854 * lra-constraints.c (simplify_operand_subreg): Handle
32855 WORD_REGISTER_OPERATIONS targets.
32857 2017-02-22 Jakub Jelinek <jakub@redhat.com>
32860 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
32861 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
32862 elimination by swapping fld*.
32864 2017-02-22 Richard Biener <rguenther@suse.de>
32866 PR tree-optimization/79673
32867 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
32868 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
32869 irrelevant address-space qualifiers and avoiding a
32870 ADDR_SPACE_CONVERT_EXPR from fold_convert.
32872 2017-02-22 Richard Biener <rguenther@suse.de>
32874 PR tree-optimization/79666
32875 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
32876 to not symbolically negate if that may introduce undefined
32879 2017-02-22 Martin Liska <mliska@suse.cz>
32882 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
32883 * data-streamer-out.c (streamer_write_gcov_count_stream):
32885 * value-prof.c (stream_out_histogram_value): Make assert more
32886 precise based on type of counter.
32888 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
32891 * config/i386/i386.md (standard_x87sse_constant_load splitter):
32892 Use nonimmediate_operand instead of memory_operand for operand 1.
32893 (float-extend standard_x87sse_constant_load splitter): Ditto.
32895 2017-02-21 Jeff Law <law@redhat.com>
32897 PR tree-optimization/79621
32898 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
32899 blocks with edges to themselves.
32901 2017-02-21 Jakub Jelinek <jakub@redhat.com>
32904 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
32905 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
32906 Use gimple_call_builtin_p.
32909 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
32910 on temporarily removed DEBUG_INSNs.
32912 PR tree-optimization/79649
32913 * tree-loop-distribution.c (classify_partition): Give up on
32914 non-generic address space loads/stores.
32916 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
32918 * doc/loop.texi (Loop manipulation): Remove nonexistent
32919 tree_ssa_loop_version from the documentation.
32920 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
32922 2017-02-21 Jakub Jelinek <jakub@redhat.com>
32925 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
32926 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
32927 * config/rs6000/rs6000.c: Include except.h.
32928 (rs6000_expand_split_stack_prologue): Call
32929 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
32931 2017-02-21 Martin Jambor <mjambor@suse.cz>
32934 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
32935 have been analyzed.
32937 2017-02-21 Martin Jambor <mjambor@suse.cz>
32939 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
32940 for backward compatibility only.
32941 * doc/invoke.texi (Option Summary): Remove all references to
32942 -fipa-cp-alignment.
32944 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
32948 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
32950 * lra-constraints.c (curr_insn_transform): Handle
32951 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
32953 2017-02-21 Martin Liska <mliska@suse.cz>
32955 * config/i386/i386.opt: Replace -masm-dialect with -masm.
32957 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
32959 PR translation/79638
32960 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
32962 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
32965 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
32966 (arm_function_ok_for_sibcall): Return false for an indirect call by
32967 descriptor if all the argument registers are used.
32968 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
32969 alignment of the function.
32971 2017-02-21 Jakub Jelinek <jakub@redhat.com>
32973 PR tree-optimization/61441
32974 * simplify-rtx.c (simplify_const_unary_operation): For
32975 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
32976 the sNaN unmodified.
32978 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
32980 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32981 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32982 instead of SYSTEM_HEADER_DIR.
32984 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
32985 Martin Liška <mliska@suse.cz>
32987 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
32988 Fix typos and grammar, use active voice, and clarify.
32990 2017-02-20 Marek Polacek <polacek@redhat.com>
32992 PR middle-end/79537
32993 * gimplify.c (gimplify_expr): Handle unused *&&L;.
32996 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
32998 2017-02-20 Jakub Jelinek <jakub@redhat.com>
33001 * config/i386/i386.c (ix86_expand_builtin): Handle
33002 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
33003 ix86_builtins_isa[fcode].isa as a requirement of those
33004 flags and any other flag in the bitmask.
33005 (ix86_init_mmx_sse_builtins): Use 0 instead of
33006 ~OPTION_MASK_ISA_64BIT as mask.
33007 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
33008 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
33009 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
33010 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
33012 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33015 * lra-constraints.c (split_reg): Check requested split mode
33016 is supported by the register.
33018 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33020 * lra-constraints.c (simplify_operand_subreg): Remove early
33023 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33026 * lra-constraints.c (curr_insn_transform): Tighten condition
33027 for converting SUBREG reloads from OP_OUT to OP_INOUT.
33029 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33032 * lra-constraints.c (curr_insn_transform): Handle
33033 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
33035 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
33038 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
33040 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
33042 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
33045 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
33048 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33050 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
33053 2017-02-18 Jakub Jelinek <jakub@redhat.com>
33055 * final.c (last_columnnum, override_columnnum): New variables.
33056 (final_start_function): Set last_columnnum, pass it to begin_prologue
33057 hook and pass 0 to dwarf2out_begin_prologue.
33058 (final_scan_insn): Update override_columnnum. Pass last_columnnum
33059 to source_line debug hook.
33060 (notice_source_line): Compute last_columnnum and for debug_column_info
33061 return true on column changes.
33062 * debug.h (struct gcc_debug_hooks): Add column argument to
33063 source_line and begin_prologue hooks.
33064 (debug_nothing_int_charstar_int_bool): Remove prototype.
33065 (debug_nothing_int_int_charstar,
33066 debug_nothing_int_int_charstar_int_bool): New prototypes.
33067 (dwarf2out_begin_prologue): Add column argument.
33068 * debug.c (do_nothing_debug_hooks): Adjust source_line and
33069 begin_prologue hooks.
33070 (debug_nothing_int_charstar_int_bool): Remove.
33071 (debug_nothing_int_int_charstar,
33072 debug_nothing_int_int_charstar_int_bool): New functions.
33073 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
33074 through to dwarf2out_source_line.
33075 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
33076 (dwarf2out_source_line): Add column argument, emit it if requested.
33077 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
33079 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
33080 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
33081 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
33082 through to dwarf2out_begin_prologue.
33083 (vmsdbgout_source_line): Add column argument, pass it through to
33084 dwarf2out_source_line.
33085 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
33086 dbxout_source_line caller.
33087 (dbxout_source_line): Add column argument.
33089 * common.opt (gno-column-info, gcolumn-info): New options.
33090 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
33091 (check_die): Also test for multiple DW_AT_decl_column attributes.
33092 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
33093 DW_AT_decl_column if requested.
33094 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
33096 (gen_variable_die): Likewise.
33097 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
33098 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
33101 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
33102 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
33103 (ix86_handle_option): Handle OPT_m3dnowa.
33104 * doc/invoke.texi (-m3dnowa): Document.
33105 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
33106 -m3dnowa instead of -m3dnow -march=athlon.
33109 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
33110 instead of gcc_assert for K, r and R code checks. Formatting fixes.
33112 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33115 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
33116 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
33117 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
33118 generator for vsx_xxpermdi_<mode>_be.
33119 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
33120 force big-endian semantics.
33121 (vsx_xxpermdi_<mode>_be): New define_expand with same
33122 implementation as previous version of vsx_xxpermdi_<mode>.
33124 2017-02-17 Jakub Jelinek <jakub@redhat.com>
33126 PR tree-optimization/79327
33127 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
33128 variable, its initialization and use.
33130 2017-02-17 Julia Koval <julia.koval@intel.com>
33132 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
33133 (OPTION_MASK_ISA_PKU_UNSET): New.
33134 (ix86_handle_option): Handle -mrdpid.
33135 * config/i386/cpuid.h (bit_RDPID): New.
33136 * config/i386/driver-i386.c (host_detect_local_cpu):
33137 Detect RDPID feature.
33138 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
33139 * config/i386/i386-c.c (ix86_target_macros_internal):
33141 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
33142 (ix86_valid_target_attribute_inner_p): Add "rdpid".
33143 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
33144 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
33145 * config/i386/i386.md (define_insn "rdpid"): New.
33146 * config/i386/i386.opt Add -mrdpid.
33147 * config/i386/immintrin.h (_rdpid_u32): New.
33149 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
33151 PR rtl-optimization/79541
33152 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
33153 instead of transforming it into USE.
33155 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
33157 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
33158 If HONOR_SNANS (SFmode) force the input to a register.
33159 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
33160 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
33161 an frsp or similar insn.
33163 2017-02-17 Martin Liska <mliska@suse.cz>
33165 PR rtl-optimization/79577
33166 * params.def (selsched-max-sched-times): Increase minimum to 1.
33168 2017-02-17 Martin Liska <mliska@suse.cz>
33170 PR rtl-optimization/79574
33171 * gcse.c (want_to_gcse_p): Prevent integer overflow.
33173 2017-02-17 Martin Liska <mliska@suse.cz>
33175 PR tree-optimization/79529
33176 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
33177 ssa_defined_default_def_p to handle cases which are implicitly
33179 * tree-ssa.c (ssa_defined_default_def_p): New function.
33180 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
33181 which are implicitly defined.
33182 * tree-ssa.h (ssa_defined_default_def_p): Declare.
33184 2017-02-17 Richard Biener <rguenther@suse.de>
33186 PR middle-end/79576
33187 * params.def (max-ssa-name-query-depth): Limit to 10.
33189 2017-02-17 Richard Biener <rguenther@suse.de>
33191 PR tree-optimization/79552
33192 * tree-ssa-structalias.c (visit_loadstore): Properly verify
33195 2017-02-17 Richard Biener <rguenther@suse.de>
33198 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
33200 2017-02-17 Marek Polacek <polacek@redhat.com>
33202 PR middle-end/79536
33203 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
33204 (fold_negate_expr): New wrapper.
33206 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
33208 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
33209 Correct terminology and de-emphasize pre-standard behavior.
33211 2017-02-16 Alan Modra <amodra@gmail.com>
33213 PR rtl-optimization/79286
33214 * ira.c (def_dominates_uses): New function.
33215 (update_equiv_regs): Don't create an equivalence for insns that
33216 may trap where the register def does not dominate the use.
33218 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
33220 PR rtl-optimization/78127
33221 * lra.c (lra): Call lra_eliminate before finish the loop after
33224 2017-02-16 Richard Biener <rguenther@suse.de>
33226 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
33228 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
33229 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
33230 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
33231 (isl_val_int_from_wi): New function.
33232 (extract_affine_gmp): Rename to ...
33233 (extract_affine_wi): ... this, take a widest_int.
33234 (extract_affine_int): Just wrap extract_affine_wi.
33235 (add_param_constraints): Use isl_val_int_from_wi.
33236 (add_loop_constraints): Likewise, and extract_affine_wi.
33238 2017-02-15 Jeff Law <law@redhat.com>
33240 PR middle-end/79521
33241 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
33242 ira_init_register_move_cost_if_necessary.
33244 2017-02-15 Martin Sebor <msebor@redhat.com>
33246 PR middle-end/32003
33247 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
33248 removed in a prior commit.
33250 2017-02-15 Bin Cheng <bin.cheng@arm.com>
33252 PR tree-optimization/79347
33253 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
33254 counters during peeling.
33256 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
33258 * Makefile.in (site.exp): Remove "set ISLVER".
33260 2017-02-15 Jakub Jelinek <jakub@redhat.com>
33263 * real.c (real_from_integer): Call real_convert even for decimal.
33265 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
33268 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
33270 2017-02-14 Andrew Pinski <apinski@cavium.com>
33272 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
33273 cores and change the partno/implementer to be correct.
33274 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
33275 the 'B" as the implementer.
33276 * config/aarch64/aarch64-tune.md: Regenerate.
33278 2017-02-14 Carl Love <cel@us.ibm.com>
33280 * config/rs6000/rs6000.c: Add case statement entry to make the
33281 xvcvuxdsp built-in argument unsigned.
33282 * config/rs6000/vsx.md: Fix the source and return operand types so they
33283 match the instruction definitions from the ISA document. Fix typo
33284 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
33287 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
33290 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
33291 member early_clobber_alts.
33292 * lra-lives.c (reg_early_clobber_p): New.
33293 (process_bb_lives): Use it.
33294 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
33295 (debug_operand_data): Initialize early_clobber_alts.
33296 (setup_operand_alternative): Set up early_clobber_alts.
33297 (collect_non_operand_hard_regs): Ditto. Pass early clobber
33298 alternatives to new_insn_reg.
33299 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
33301 (lra_update_insn_regno_info): Pass the new arg.
33303 2017-02-14 Jakub Jelinek <jakub@redhat.com>
33305 PR middle-end/79505
33306 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
33307 (new_oacc_loop_raw): Don't clear already cleared fields.
33310 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
33311 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
33312 _mm512_prefetch_i64gather_ps): New inline functions and macros.
33314 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
33317 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
33319 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
33322 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
33323 the extra instruction to the right place to store 128-bit constant
33326 2017-02-14 Martin Sebor <msebor@redhat.com>
33328 PR middle-end/79448
33329 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
33330 warning for strings of unknown length.
33332 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
33334 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
33336 2017-02-14 Jeff Law <law@redhat.com>
33339 * ira-costs.c (scan_one_insn): Initialize register move costs
33340 for pseudos seen in USE/CLOBBER insns.
33342 PR tree-optimization/79095
33343 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
33344 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
33345 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
33346 if the operands are known to be not equal, then the resulting range
33348 (intersect_ranges): If the new range is ~[0,0] and the old range is
33349 wide, then prefer ~[0,0].
33350 * tree-vrp.c (overflow_comparison_p_1): New function.
33351 (overflow_comparison_p): New function.
33352 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
33353 if NAME is used in an overflow test.
33354 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
33355 overflow check that can be expressed as an equality test, then adjust
33356 ops to be that equality test.
33358 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33360 * config/s390/s390-builtin-types.def: Remove flags argument.
33361 * config/s390/s390.c (s390_init_builtins): Likewise.
33363 2017-02-14 Martin Liska <mliska@suse.cz>
33365 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
33366 vector. Fix trailing white spaces.
33368 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
33370 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
33373 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33375 PR rtl-optimization/68664
33376 * config/arm/arm.c (arm_sched_can_speculate_insn):
33377 New function. Declare prototype.
33378 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
33380 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33382 PR rtl-optimization/68664
33383 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
33385 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
33387 2017-02-14 Amit Pawar <amit.pawar@amd.com>
33389 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
33390 max skip bytes for function, loop and jump.
33392 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33394 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
33395 ABS_EXPR for gimple dump.
33397 2017-02-14 Jakub Jelinek <jakub@redhat.com>
33400 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
33402 PR tree-optimization/79408
33403 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
33404 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
33405 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
33406 also if rhs1 is INTEGER_CST.
33408 2017-02-14 Richard Biener <rguenther@suse.de>
33410 PR middle-end/79432
33411 * tree-into-ssa.c (insert_phi_nodes): When the function can
33412 have abnormal edges rewrite SSA names with broken use-def
33413 dominance out of SSA and register them for PHI insertion.
33415 2017-02-13 Martin Sebor <msebor@redhat.com>
33417 PR middle-end/79496
33418 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
33419 clearing info.nowrite flag when snprintf size argument is a range.
33421 2017-02-13 Jakub Jelinek <jakub@redhat.com>
33423 * cprop.c (cprop_jump): Add missing space in string literal.
33424 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
33425 (get_constraint_for_component_ref): Likewise.
33426 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
33427 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
33428 * lra-constraints.c (process_alt_operands): Likewise.
33429 * ipa-inline.c (inline_small_functions): Likewise.
33430 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
33431 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
33432 * trans-mem.c (diagnose_tm_1_op): Likewise.
33433 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
33434 (grid_parallel_clauses_gridifiable): Likewise.
33436 * config/nvptx/mkoffload.c (process): Add space in between
33439 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
33440 "MOD4_SSE_REGS" and "ALL_REGS".
33442 * spellcheck.c (test_data): Add , in between "foo" and "food".
33444 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33447 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
33448 boundary crossing check and subsequent code generation agree.
33450 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33452 * config/aarch64/aarch64.c (has_memory_op): Delete.
33453 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
33456 2017-02-13 Jakub Jelinek <jakub@redhat.com>
33458 PR rtl-optimization/79388
33459 PR rtl-optimization/79450
33460 * combine.c (distribute_notes): When removing TEM_INSN for which
33461 corresponding dest has last value recorded, invalidate that last
33464 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33466 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
33467 of explicit '@'. Add missing assembly comment marker on branch costs
33470 2017-02-13 Nathan Sidwell <nathan@acm.org>
33472 * gengtype-lex.l (<in_struct>): Add '/'.
33474 2017-02-13 Martin Liska <mliska@suse.cz>
33477 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
33479 2017-02-13 Richard Biener <rguenther@suse.de>
33481 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
33483 * configure: Re-generate.
33484 * config.in: Likewise.
33485 * graphite-dependences.c: Simplify as if
33486 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
33487 * graphite-isl-ast-to-gimple.c: Likewise.
33488 * graphite-optimize-isl.c: Likewise.
33489 * graphite-poly.c: Likewise.
33490 * graphite-sese-to-poly.c: Likewise.
33491 * graphite.h: Likewise.
33492 * toplev.c: Include isl/version.h and use isl_version () for
33493 printing the ISL version.
33494 * doc/install.texi: Update ISL requirement.
33496 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
33498 * doc/standards.texi (Standards): Update reference to
33501 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
33503 * doc/extend.texi (Named Address Spaces): sourceware.org now
33505 * doc/install.texi (Binaries): Ditto.
33508 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
33510 * doc/cpp.texi: Replace "stringify"/"stringification" with C
33511 standard terminology "stringize"/"stringizing" throughout.
33512 * doc/cppinternals.texi: Likewise.
33514 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
33516 * doc/extend.texi: Fix some spelling mistakes and typos.
33517 * doc/invoke.texi: Likewise.
33519 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
33522 * params.def (inline-min-speedup) Change from 10 to 8.
33524 2017-02-11 Jakub Jelinek <jakub@redhat.com>
33526 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
33529 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
33532 * ipa-inline-analysis.c (get_minimal_bb): New function.
33533 (record_modified): Use it.
33534 (remap_edge_change_prob): Handle also ancestor functions.
33536 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
33538 * doc/contrib.texi (Contributors): Remove broken link into
33539 the Mauve CVS repository.
33541 2017-02-11 Jakub Jelinek <jakub@redhat.com>
33543 PR middle-end/79454
33544 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
33545 result computation whenever lhs doesn't have vector mode, not
33546 just when it has BLKmode.
33548 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
33550 * doc/makefile.texi (profiledbootstrap): Refer to the
33551 installation instructions only in textual form.
33553 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33556 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
33558 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
33560 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
33561 (Specific): Update mingw-w64 reference.
33563 (Specific): Remove broken link to Renesas RX processor.
33565 2017-02-10 Richard Biener <rguenther@suse.de>
33567 * toplev.c (process_options): Do not mention obsolete graphite
33568 options when printing sorry message about missing graphite support.
33569 Mention -floop-nest-optimize.
33571 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
33573 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
33574 (vtst_p16): Likewise.
33575 (vtstq_p8): Likewise.
33576 (vtstq_p16): Likewise.
33578 (vtstq_p64): Likewise.
33579 * config/arm/arm_neon.h (vgetq_lane_p64): New.
33580 (vset_lane_p64): New.
33581 (vsetq_lane_p64): New.
33583 2017-02-10 Jakub Jelinek <jakub@redhat.com>
33585 PR tree-optimization/79411
33586 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
33587 stmt operands are SSA_NAMEs used in abnormal phis.
33588 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
33591 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
33594 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
33597 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
33599 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
33601 2017-02-09 Jakub Jelinek <jakub@redhat.com>
33603 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
33607 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
33608 not arbitrary TREE_CONSTANT.
33611 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
33612 "omp declare target link" attribute unless is_global_var.
33613 * omp-offload.c (find_link_var_op): Likewise.
33615 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
33616 Chung-Lin Tang <cltang@codesourcery.com>
33618 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
33620 (gimplify_adjust_omp_clauses): Don't delete TILE.
33621 (gimplify_omp_for): Deal with TILE.
33622 * internal-fn.c (expand_GOACC_TILE): New function.
33623 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
33625 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
33626 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
33628 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
33629 avoid DIV for outermost collapse var.
33630 (expand_oacc_for): Insert tile element loop as needed. Adjust.
33631 Remove out of date comments, fix whitespace.
33632 * omp-general.c (omp_extract_for_data): Deal with tiling.
33633 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
33634 adjust OLF_DIM_BASE value.
33635 (struct omp_for_data): Add tiling field.
33636 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
33637 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
33638 for auto loops. Remove default auto determining, moved to
33639 oacc_loop_fixed_partitions.
33640 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
33641 stmts, add e_mask field.
33642 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
33643 (oacc_thread_numbers): Use oacc_dim_call.
33644 (oacc_xform_tile): New.
33645 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
33646 (finish_oacc_loop): Adjust for ifns vector.
33647 (oacc_loop_discover_walk): Append loop abstraction sites to list,
33648 add case for GOACC_TILE fns.
33649 (oacc_loop_xform_loop): Delete.
33650 (oacc_loop_process): Iterate over call list directly, and add
33651 handling for GOACC_TILE fns.
33652 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
33654 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
33655 vector partitioning to outer loops. Assign 2 partitions to loops
33656 when available. Add TILE handling.
33657 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
33658 (execite_oacc_device_lower): Process GOACC_TILE fns,
33659 ignore unknown specs.
33660 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
33661 * tree.c (omp_clause_num_ops): Adjust TILE ops.
33662 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
33664 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
33666 * configure.ac (ACX_BUGURL): Update.
33667 * configure: Regenerate.
33669 2017-02-09 Richard Biener <rguenther@suse.de>
33671 PR tree-optimization/69823
33672 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
33673 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
33675 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
33677 * config/arc/arc-c.def: Add __NPS400__ definition.
33678 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
33679 (TARGET_NPS400): Define.
33681 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
33683 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
33685 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
33686 pointer, arch_info.
33687 (arc_cpu_types): Fill the arch_info field with a pointer into the
33688 arc_arch_types table.
33689 (arc_selected_cpu): Declare.
33690 * config/arc/arc.c (arc_selected_cpu): Make global.
33691 (arc_selected_arch): Delete.
33692 (arc_base_cpu): Delete.
33693 (arc_override_options): Remove references to deleted variables,
33694 update access to arch information.
33695 (ARC_OPT): Update access to arch information.
33696 (ARC_OPTX): Likewise.
33697 * config/arc/arc.h (arc_base_cpu): Remove declaration.
33698 (TARGET_ARC600): Update access to arch information.
33699 (TARGET_ARC601): Likewise.
33700 (TARGET_ARC700): Likewise.
33701 (TARGET_EM): Likewise.
33702 (TARGET_HS): Likewise.
33703 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
33706 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
33709 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
33710 condition/operands for integer GE/LE/GEU/LEU operations.
33712 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
33714 PR translation/79397
33715 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
33718 2017-02-08 Martin Jambor <mjambor@suse.cz>
33721 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
33722 whether allocation happened.
33723 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
33724 nothing was allocated.
33726 2017-02-08 Jakub Jelinek <jakub@redhat.com>
33728 PR tree-optimization/79408
33729 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
33730 constant, but SSA_NAME with a known integer range, use the minimum
33731 of that range instead of op1 to determine if modulo can be replaced
33732 with its first operand.
33734 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33736 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
33738 2017-02-08 Richard Biener <rguenther@suse.de>
33740 PR tree-optimization/71824
33741 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
33742 Check all loops contained in the merged region.
33744 2017-02-07 Andrew Pinski <apinski@cavium.com>
33746 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
33748 2017-02-07 Andrew Pinski <apinski@cavium.com>
33750 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
33751 (thunderxt88): Likewise.
33752 (thunderxt81): Disable LSE and change v8.1 to v8.
33753 (thunderxt83): Likewise.
33755 2017-02-07 Jakub Jelinek <jakub@redhat.com>
33756 Richard Biener <rguenther@suse.de>
33758 PR middle-end/79399
33759 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
33760 type from int to size_t.
33761 * ira-costs.c (struct_costs_size): Change type from int to size_t.
33763 2017-02-07 Jakub Jelinek <jakub@redhat.com>
33765 PR rtl-optimization/79386
33766 * cprop.c (bypass_conditional_jumps): Initialize
33767 bypass_last_basic_block already before splitting bbs after
33768 unconditional traps...
33769 (bypass_conditional_jumps): ... rather than here.
33772 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
33773 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
33774 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
33775 fix -masm=intel patterns.
33777 2017-02-07 Richard Biener <rguenther@suse.de>
33779 PR tree-optimization/79256
33780 PR middle-end/79278
33781 * builtins.c (get_object_alignment_2): Use min_align_of_type
33782 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
33783 and ADJUST_FIELD_ALIGN.
33785 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
33787 * doc/tm.texi: Regenerate.
33788 * stor-layout.c (layout_decl): Adjust.
33789 (update_alignment_for_field): Likewise.
33790 (place_field): Likewise.
33791 (min_align_of_type): Likewise.
33792 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
33793 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
33794 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
33795 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
33796 * config/frv/frv.c (frv_adjust_field_align): Likewise.
33797 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
33798 * config/i386/i386.c (x86_field_alignment): Likewise.
33799 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
33800 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
33801 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
33802 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
33803 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
33804 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
33808 2017-01-30 Richard Biener <rguenther@suse.de>
33810 PR tree-optimization/79256
33811 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
33812 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
33815 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
33817 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
33818 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
33819 builtins to SImode and emit a zero-extend, if necessary.
33821 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
33823 * docs/invoke.texi (RISC-V Options): Alphabetize.
33825 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
33827 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
33830 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
33832 * config/riscv/riscv.c: New file.
33833 * common/config/riscv/riscv-common.c: Likewise.
33834 * config.gcc: Likewise.
33835 * config/riscv/constraints.md: Likewise.
33836 * config/riscv/elf.h: Likewise.
33837 * config/riscv/generic.md: Likewise.
33838 * config/riscv/linux.h: Likewise.
33839 * config/riscv/multilib-generator: Likewise.
33840 * config/riscv/peephole.md: Likewise.
33841 * config/riscv/pic.md: Likewise.
33842 * config/riscv/predicates.md: Likewise.
33843 * config/riscv/riscv-builtins.c: Likewise.
33844 * config/riscv/riscv-c.c: Likewise.
33845 * config/riscv/riscv-ftypes.def: Likewise.
33846 * config/riscv/riscv-modes.def: Likewise.
33847 * config/riscv/riscv-opts.h: Likewise.
33848 * config/riscv/riscv-protos.h: Likewise.
33849 * config/riscv/riscv.h: Likewise.
33850 * config/riscv/riscv.md: Likewise.
33851 * config/riscv/riscv.opt: Likewise.
33852 * config/riscv/sync.md: Likewise.
33853 * config/riscv/t-elf-multilib: Likewise.
33854 * config/riscv/t-linux: Likewise.
33855 * config/riscv/t-linux-multilib: Likewise.
33856 * config/riscv/t-riscv: Likewise.
33857 * configure.ac: Likewise.
33858 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
33859 Waterman as RISC-V maintainers.
33860 * doc/install.texi: Add RISC-V entries.
33861 * doc/invoke.texi: Add RISC-V options section.
33862 * doc/md.texi: Add RISC-V constraints section.
33863 * configure: Regenerated.
33865 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
33868 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
33869 false values to be constant vectors with all 0 or all 1 bits set.
33870 (vcondu<mode><mode>): Likewise.
33871 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
33873 (fpmask_comparison_operator): Update comment.
33874 (vecint_comparison_operator): New predicate.
33875 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
33876 vector conditionals when the true and false values are constant
33877 vectors with all 0 bits or all 1 bits set.
33879 2017-02-06 Martin Sebor <msebor@redhat.com>
33881 PR tree-optimization/79376
33882 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
33884 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
33886 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
33887 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
33888 to simplify split condition.
33890 2017-02-06 Jakub Jelinek <jakub@redhat.com>
33892 * omp-expand.c (oxpand_omp_atomic_fetch_op,
33893 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
33896 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
33898 PR rtl-optimization/68664
33899 * target.def (can_speculate_insn): New hook.
33900 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
33901 * doc/tm.texi: Regenerate.
33902 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
33903 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
33904 (rs6000_sched_can_speculate_insn): New function.
33906 2017-02-06 Jakub Jelinek <jakub@redhat.com>
33908 PR tree-optimization/79284
33909 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
33910 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
33911 vectorizable_mask_load_store, vectorizable_operation,
33912 vect_is_simple_cond, get_same_sized_vectype): Use it instead
33913 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
33914 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
33915 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
33916 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
33917 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
33918 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
33919 is_gimple_assign (stmt). Replace another such test with
33920 is_gimple_assign (stmt).
33922 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
33925 * config/avr/avr.c (rtl-iter.h): Include it.
33926 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
33927 (avr_legitimate_combined_insn): ...and implementation.
33929 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
33931 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
33932 * config/s390/s390.c (s390_const_operand_ok)
33933 (s390_canonicalize_comparison, s390_extract_part)
33934 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
33935 (s390_contiguous_bitmask_p, s390_rtx_costs)
33936 (legitimize_pic_address): Likewise.
33937 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
33938 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
33939 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
33940 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
33941 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
33943 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
33945 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
33946 REGNO($0) == REGNO($1).
33948 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33950 * config/s390/linux.h(SIZE_TYPE): Add comment.
33952 2017-02-06 Julian Brown <julian@codesourcery.com>
33953 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33954 Virendra Pathak <virendra.pathak@broadcom.com>
33956 * config/aarch64/aarch64-cores.def: Change the scheduler
33958 * config/aarch64/aarch64.md: Include thunderx2t99.md.
33959 * config/aarch64/thunderx2t99.md: New file.
33961 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
33963 * doc/standards.texi (Go Language): Update link to language
33966 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
33968 * tree-eh.c (lower_resx): Sanitize profile.
33969 (cleanup_empty_eh_move_lp): Likewise.
33971 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
33974 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
33976 * cfgloopmanip.h (loop_version): Update prototype.
33977 * modulo-sched.c (sms_schedule): Update call of loop_version.
33978 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
33979 * tree-parloops.c (gen_parallel_loop): Likewise.
33980 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
33981 * tree-ssa-loop-split.c (split_loop): Likewise.
33982 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
33983 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
33985 2017-02-05 Martin Liska <mliska@suse.cz>
33988 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
33990 (print_operand_address): Initialize a struct to zero.
33992 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
33994 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
33995 garbage collector only in textual form.
33997 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
33999 * doc/extend.texi (x86 specific memory model extensions for
34000 transactional memory): Simplify a phrase.
34002 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
34005 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
34006 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
34007 (atomic_storedi_1): Likewise.
34009 2017-02-04 Jakub Jelinek <jakub@redhat.com>
34011 PR tree-optimization/79338
34012 * tree-parloops.c (gather_scalar_reductions): Don't call
34013 vect_analyze_loop_form for loop->inner before destroying loop's
34016 2017-02-03 Martin Sebor <msebor@redhat.com>
34018 PR tree-optimization/79327
34019 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
34020 when precision has resulted in leading zeros.
34021 (format_integer): Adjust the likely counter to assume an unknown
34022 argument that may be zero is non-zero.
34024 2017-02-03 Jason Merrill <jason@redhat.com>
34027 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
34028 avoid copying non-taken branch.
34030 2017-02-03 Jakub Jelinek <jakub@redhat.com>
34032 PR tree-optimization/79340
34033 * tree-vect-loop.c (vectorizable_reduction): Release
34034 vec_defs elements after safe_splicing them into other vectors.
34037 PR tree-optimization/79327
34038 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
34039 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
34041 (format_integer): Use wide_int_to_tree instead of build_int_cst
34042 + to_?hwi. If argmin is NULL, just set argmin and argmax to
34043 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
34044 of shortest and longest sequence.
34046 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
34048 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
34049 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
34051 2017-02-03 Walter Lee <walt@tilera.com>
34054 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
34055 after initial stackframe link reg save.
34056 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
34058 2017-02-03 Jakub Jelinek <jakub@redhat.com>
34061 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
34062 wu for stxssp alternative.
34064 2017-02-03 Martin Sebor <msebor@redhat.com>
34066 PR tree-optimization/79352
34067 * gimple-fold.c (get_range_strlen): Add argument.
34068 (get_range_strlen): Change return type to bool.
34069 (get_maxval_strlen): Pass in a dummy argument.
34070 * gimple-fold.h (get_range_strlen): Change return type to bool.
34071 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
34072 * tree.h (array_at_struct_end_p): Add argument.
34073 * tree.c (array_at_struct_end_p): Handle it.
34075 2017-02-03 Martin Liska <mliska@suse.cz>
34078 * multiple_target.c (create_dispatcher_calls): Redirect edge
34079 from a caller of a dispatcher.
34080 (expand_target_clones): Make the clones local.
34081 (ipa_target_clone): Do both target clones and resolvers.
34082 (ipa_dispatcher_calls): Remove the pass.
34083 (pass_dispatcher_calls::gate): Likewise.
34084 (make_pass_dispatcher_calls): Likewise.
34085 * passes.def (pass_target_clone): Put as very first IPA early
34088 2017-02-03 Martin Liska <mliska@suse.cz>
34090 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
34091 in case of a function with ifunc attribute.
34093 2017-02-03 Martin Liska <mliska@suse.cz>
34095 * cgraph.c (cgraph_node::dump): Dump function version info.
34096 * symtab.c (symtab_node::dump_base): Add missing new line.
34098 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
34100 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
34101 (ifcombine_ifandif): Use it.
34103 2017-02-03 Martin Liska <mliska@suse.cz>
34105 * doc/invoke.texi: Document default value for
34106 use-after-scope-direct-emission-threshold.
34108 2017-02-03 Martin Liska <mliska@suse.cz>
34110 PR tree-optimization/79339
34111 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
34112 (format_floating): Likewise.
34114 2017-02-03 Martin Liska <mliska@suse.cz>
34117 * ipa-prop.c (ipa_node_params_t::insert): Remove current
34119 (ipa_node_params_t::remove): Likewise.
34120 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
34121 initialization from removed ipa_node_params_t::insert.
34122 (ipa_node_params::~ipa_node_params): Move from removed
34123 ipa_node_params_t::release.
34124 * symbol-summary.h (symbol_summary::m_released): New member.
34125 Do not release a summary twice. Do not allow to call finalizer
34126 for types of a summary that live in GGC memory.
34128 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34130 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
34133 2017-02-02 Martin Sebor <msebor@redhat.com>
34135 PR middle-end/79275
34136 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
34137 (format_string): Tighten up the range of output for non-constant
34138 strings and correct the expected range for wide non-constant strings.
34140 2017-02-02 Martin Sebor <msebor@redhat.com>
34142 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
34144 PR middle-end/32003
34145 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
34147 (-fdump-tree-@var): Add to index and document how to come up
34148 with pass-specific option and dump file names.
34149 (-fdump-passes): Clarify where to look for output.
34151 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
34153 PR middle-end/77445
34154 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
34155 statistics of the analyzed path; allow threading for speed when
34156 any of BBs along the path are optimized for speed.
34158 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
34160 PR middle-end/78468
34161 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
34162 settings of the virtual registers.
34165 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
34167 * explow.c (get_dynamic_stack_size): Take known alignment of stack
34168 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
34171 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34173 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
34174 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
34176 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34178 * config/s390/s390.md: Add missing comments with the expanded
34180 * config/s390/vector.md: Likewise.
34181 * config/s390/vx-builtins.md: Likewise.
34183 2017-02-02 Jakub Jelinek <jakub@redhat.com>
34186 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
34187 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
34188 conditions on a single line.
34190 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34192 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
34193 __S390_VX__ to __VX__.
34195 2017-02-01 Andrew Pinski <apinski@cavium.com>
34197 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
34198 stmt_info to record_stmt_cost.
34199 (vect_get_known_peeling_cost): Pass stmt_info if known to
34201 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
34202 cpu_vector_cost field into
34203 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
34204 field into vec_int_stmt_cost and vec_fp_stmt_cost.
34205 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
34206 splitting of scalar_stmt_cost and vec_stmt_cost.
34207 (thunderx_vector_cost): Likewise.
34208 (cortexa57_vector_cost): LIkewise.
34209 (exynosm1_vector_cost): Likewise.
34210 (xgene1_vector_cost): Likewise.
34211 (thunderx2t99_vector_cost): Improve after the splitting of the two
34213 (aarch64_builtin_vectorization_cost): Update for the splitting of
34214 scalar_stmt_cost and vec_stmt_cost.
34216 2017-02-01 Torvald Riegel <triegel@redhat.com>
34217 Richard Henderson <rth@redhat.com>
34219 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
34220 conditional on existance of a fast atomic load.
34221 * optabs-query.c (can_atomic_load_p): New function.
34222 * optabs-query.h (can_atomic_load_p): Declare it.
34223 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
34224 no fast atomic load is available for the particular size of access.
34225 (expand_atomic_compare_and_swap): Likewise.
34226 (expand_atomic_load): Likewise.
34227 (expand_atomic_store): Likewise.
34228 (expand_atomic_fetch_op): Likewise.
34229 * testsuite/lib/target-supports.exp
34230 (check_effective_target_sync_int_128): Remove x86 because it provides
34231 no fast atomic load.
34232 (check_effective_target_sync_int_128_runtime): Likewise.
34234 2017-02-01 Richard Biener <rguenther@suse.de>
34236 * graphite.c: Include tree-vectorizer.h for find_loop_location.
34237 (graphite_transform_loops): Provide opt-info for optimized nests.
34238 * tree-parloop.c (parallelize_loops): Provide opt-info for
34239 parallelized loops.
34241 2017-02-01 Richard Biener <rguenther@suse.de>
34243 PR middle-end/79315
34244 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
34245 was not set before.
34247 2017-02-01 Richard Biener <rguenther@suse.de>
34249 PR tree-optimization/71824
34250 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
34251 Verify the loops are valid in the merged SESE region.
34252 (scop_detection::can_represent_loop_1): Check analyzing the
34253 evolution of the number of iterations in the region succeeds.
34255 2017-01-31 Ian Lance Taylor <iant@golang.org>
34257 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
34258 REG_ARGS_SIZE note to 32-bit push insns and call insn.
34260 2017-01-31 David Malcolm <dmalcolm@redhat.com>
34262 PR preprocessor/79210
34263 * input.c (get_substring_ranges_for_loc): Replace line_width
34264 assertion with error-handling.
34266 2017-01-31 Richard Biener <rguenther@suse.de>
34268 PR tree-optimization/77318
34269 * graphite-sese-to-poly.c (extract_affine): Fix assert.
34270 (create_pw_aff_from_tree): Take loop parameter.
34271 (add_condition_to_pbb): Pass loop of the condition to
34272 create_pw_aff_from_tree.
34274 2017-01-31 Jakub Jelinek <jakub@redhat.com>
34276 * config/s390/s390.c (s390_asan_shadow_offset): New function.
34277 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
34279 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
34283 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
34285 (convert_int_to_float128): Likewise.
34286 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
34287 (convert_int_to_float128): Likewise.
34288 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
34289 (UNSPEC_IEEE128_CONVERT): Likewise.
34290 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
34291 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
34292 Use local variables for IBM extended format.
34293 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
34294 (fix_trunc<mode>si2_fprs): Likewise.
34295 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
34296 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
34297 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
34298 to know that we can now have integers of all sizes in vector
34300 (fix<uns>_<mode>di2_hw): Likewise.
34301 (float<uns>_<mode>si2_hw): Likewise.
34302 (fix_<mode>si2_hw): Likewise.
34303 (fixuns_<mode>si2_hw): Likewise.
34304 (float<uns>_<mode>di2_hw): Likewise.
34305 (float_<mode>di2_hw): Likewise.
34306 (float_<mode>si2_hw): Likewise.
34307 (floatuns_<mode>di2_hw): Likewise.
34308 (floatuns_<mode>si2_hw): Likewise.
34309 (xscvqp<su>wz_<mode>): Delete, no longer used.
34310 (xscvqp<su>dz_<mode>): Likewise.
34311 (xscv<su>dqp_<mode>): Likewise.
34312 (ieee128_mfvsrd_64bit): Likewise.
34313 (ieee128_mfvsrd_32bit): Likewise.
34314 (ieee128_mfvsrwz): Likewise.
34315 (ieee128_mtvsrw): Likewise.
34316 (ieee128_mtvsrd_64bit): Likewise.
34317 (ieee128_mtvsrd_32bit): Likewise.
34319 2017-01-31 Martin Liska <mliska@suse.cz>
34322 * ipa-prop.c (ipa_free_all_node_params): Call release method
34323 instead of ~sumbol_summary to not to trigger double times
34326 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
34328 PR tree-optimization/71691
34329 * bitmap.h (class auto_bitmap): New.
34330 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
34331 is_maybe_undefined instead of ssa_undefined_value_p.
34333 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34335 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
34336 __S390_ARCH_LEVEL__ to __ARCH__.
34338 2017-01-31 Jakub Jelinek <jakub@redhat.com>
34340 PR tree-optimization/79267
34341 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
34342 if should_remove_lhs_p is true.
34344 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
34347 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
34348 (add_alignment_attribute): New.
34349 (base_type_die): Add alignment attribute.
34350 (subrange_type_die): Likewise.
34351 (modified_type_die): Likewise.
34352 (gen_array_type_die): Likewise.
34353 (gen_descr_array_type_die: Likewise.
34354 (gen_enumeration_type_die): Likewise.
34355 (gen_subprogram_die): Likewise.
34356 (gen_variable_die): Likewise.
34357 (gen_field_die): Likewise.
34358 (gen_ptr_to_mbr_type_die): Likewise.
34359 (gen_struct_or_union_type_die): Likewise.
34360 (gen_subroutine_type_die): Likewise.
34361 (gen_typedef_die): Likewise.
34362 (base_type_cmp): Compare alignment attribute.
34364 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34367 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
34368 (setb_unsigned) New pattern for setb with CCUNS.
34369 * config/rs6000/rs6000.c (expand_block_compare): Use a different
34370 subfc./subfe sequence to avoid overflow problems. Generate a
34371 shorter sequence with cmpld/setb for power9.
34372 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
34373 for generating subfc. instruction.
34374 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
34375 now uses this instruction.
34377 2017-01-30 Ian Lance Taylor <iant@google.com>
34380 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
34381 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
34383 2017-01-30 Martin Sebor <msebor@redhat.com>
34385 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
34386 Move constant to the right of a relational operator.
34387 (get_mpfr_format_length, format_character, format_string): Ditto.
34388 (should_warn_p, maybe_warn): Same.
34390 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
34392 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
34395 * asan.c (get_translation_unit_decl): Remove function.
34396 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
34398 2017-01-30 Martin Liska <mliska@suse.cz>
34400 PR gcov-profile/79259
34401 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
34402 -fprofile-generate.
34404 2017-01-30 Martin Liska <mliska@suse.cz>
34407 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
34408 Initialize variables with NULL value.
34410 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
34413 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
34415 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
34417 2017-01-30 Richard Biener <rguenther@suse.de>
34419 PR tree-optimization/79276
34420 * tree-vrp.c (process_assert_insertions): Properly adjust common
34421 when removing a duplicate.
34423 2017-01-30 Richard Biener <rguenther@suse.de>
34425 PR tree-optimization/79256
34426 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
34427 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
34429 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
34431 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
34434 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
34435 ("*r<noxa>sbg_<mode>_sll_bitmask")
34436 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
34437 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
34438 Use contiguous_bitmask_nowrap_operand.
34440 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34443 * config/rs6000/altivec.h (vec_xl): Revise #define.
34444 (vec_xst): Likewise.
34446 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
34448 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
34450 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
34452 PR rtl-optimization/79194
34453 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
34454 traps before call to bypass_conditional_jumps.
34456 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
34458 PR tree-optimization/71374
34459 * lra-constraints.c (check_conflict_input_operands): New.
34460 (match_reload): Use it.
34462 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
34465 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
34466 account to calculate conflict_set.
34468 2017-01-27 Bin Cheng <bin.cheng@arm.com>
34470 PR rtl-optimization/78559
34471 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
34472 other_insn in combine.
34474 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
34476 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
34477 uint16_type_node for BT_UINT16.
34479 2017-01-27 David Malcolm <dmalcolm@redhat.com>
34481 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
34482 "RTL Tests" to menu.
34483 (GIMPLE Tests): New node.
34484 (RTL Tests): New node.
34486 2017-01-27 Richard Biener <rguenther@suse.de>
34488 PR tree-optimization/79245
34489 * tree-loop-distribution.c (distribute_loop): Apply cost
34490 modeling also to detected patterns.
34492 2017-01-27 Richard Biener <rguenther@suse.de>
34494 PR tree-optimization/71433
34495 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
34496 (compare_assert_loc): New function.
34497 (process_assert_insertions): Sort and optimize assert locations
34498 to remove duplicates and push down identical assertions on
34499 edges to their destination block.
34501 2017-01-27 Richard Biener <rguenther@suse.de>
34503 PR tree-optimization/79244
34504 * tree-vrp.c (remove_range_assertions): Forcefully propagate
34505 out SSA names even if abnormal.
34507 2017-01-27 Jakub Jelinek <jakub@redhat.com>
34509 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
34510 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
34511 instead of MPFR_RNDN.
34513 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
34516 * arm.c (arm_option_override): Don't call build_target_option_node
34517 until after doing all option overrides.
34518 (arm_valid_target_attribute_tree): Likewise.
34520 2017-01-27 Martin Liska <mliska@suse.cz>
34522 * doc/invoke.texi (-fprofile-arcs): Document profiling support
34523 for {cd}tors and C++ {cd}tors.
34525 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
34527 * config/s390/s390.md ("*setmem_long_and")
34528 ("*setmem_long_and_31z"): Use zero_extend instead of and.
34530 2017-01-26 Martin Sebor <msebor@redhat.com>
34532 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
34535 2017-01-26 Martin Sebor <msebor@redhat.com>
34537 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
34538 HAVE_DFmode before using XFmode or DFmode.
34539 (parse_directive): Avoid using the z length modifier to avoid
34540 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
34542 PR middle-end/78703
34543 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
34544 to accept adjustment as an array.
34545 (get_int_range): New function.
34546 (struct directive): Make width and prec arrays.
34547 (directive::set_width, directive::set_precision): Call get_int_range.
34548 (format_integer, format_floating): Handle width and precision ranges.
34549 (format_string, parse_directive): Same.
34551 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34554 * dwarf2out.c (generate_skeleton_bottom_up): For children with
34555 comdat_type_p set, just clone them, but keep the children in the
34559 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
34560 which have direct callers with -fvar-tracking-assignments enabled
34562 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
34563 inside of type units.
34565 2017-01-26 Martin Sebor <msebor@redhat.com>
34567 PR middle-end/78703
34568 * gimple-ssa-sprintf.c (struct result_range): Add likely and
34570 (struct format_result): Replace number_chars, number_chars_min,
34571 and number_chars_max with a single member of struct result_range.
34573 (format_result::operator+=): Adjust.
34574 (struct fmtresult): Remove bounded. Handle likely and unlikely
34576 (fmtresult::adjust_for_width_or_precision): New function.
34577 (fmtresult:type_max_digits): New function.
34578 (bytes_remaining): Handle likely and unlikely counters.
34579 (min_bytes_remaining): Remove.
34580 (format_percent): Simplify.
34581 (format_integer, format_floating): Set likely and unlikely counters.
34582 (get_string_length, format_character, format_string): Same.
34583 (format_plain, should_warn_p): New function.
34584 (maybe_warn): Call should_warn_p. Update diagnostic messages
34585 and handle those for all directives, including plain strings.
34586 (format_directive): Handle likely and unlikely counters.
34587 Remove unnecessary quoting from diagnostics. Add an informational
34589 (add_bytes): Remove.
34590 (pass_sprintf_length::compute_format_length): Simplify.
34591 (try_substitute_return_value): Handle likely and unlikely counters.
34593 2017-01-26 Carl Love <cel@us.ibm.com>
34595 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
34596 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
34598 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
34601 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
34602 endianess for subregs into account.
34603 * lra-constraints.c (lra_constraints): Do risky transformations
34604 always on the first iteration.
34605 * lra-lives.c (check_pseudos_live_through_calls): Add arg
34606 last_call_used_reg_set.
34607 (process_bb_lives): Define and use last_call_used_reg_set.
34608 * lra.c (lra): Always continue after lra_constraints on the first
34611 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
34613 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
34615 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
34617 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34619 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
34620 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
34621 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
34622 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
34623 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
34624 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
34625 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
34626 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
34627 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
34629 2017-01-26 Marek Polacek <polacek@redhat.com>
34632 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
34633 for the third operand.
34635 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34637 PR middle-end/79236
34638 * omp-low.c (struct omp_context): Add simt_stmt field.
34639 (scan_omp_for): Return omp_context *.
34640 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
34641 context to the _simt_ SIMD stmt.
34642 (lower_omp_for): For combined SIMD with sibling _simt_
34643 SIMD, make sure to use the same decls in _looptemp_
34644 clauses as in the sibling.
34646 2017-01-26 David Sherwood <david.sherwood@arm.com>
34648 PR middle-end/79212
34649 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
34652 2017-01-26 Jakub Jelinek <jakub@redhat.com>
34655 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
34656 emit fld b; fld a; if possible.
34658 * brig-builtins.def: Update copyright years.
34659 * config/arm/arm_acle_builtins.def: Update copyright years.
34661 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
34664 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
34665 constraint instead of o for the stxsd instruction.
34667 2017-01-25 Carl Love <cel@us.ibm.com>
34669 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
34670 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
34672 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
34674 * doc/invoke.texi (C++ Dialect Options): Fix typo.
34676 2017-01-25 Richard Biener <rguenther@suse.de>
34678 PR tree-optimization/69264
34679 * target.def (vector_alignment_reachable): Improve documentation.
34680 * doc/tm.texi: Regenerate.
34681 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
34683 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
34684 earlier changes with respect to TYPE_USER_ALIGN.
34685 (vector_alignment_reachable_p): Likewise. Improve dumping.
34687 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34690 * config/arm/arm.md (xordi3): Force constant operand into a register
34693 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34695 * doc/invoke.texi (-fstore-merging): Correct default optimization
34696 levels at which it is enabled.
34697 (-O): Move -fstore-merging from list to...
34700 2017-01-25 Richard Biener <rguenther@suse.de>
34703 * omp-expand.c: Include debug.h.
34704 (expand_omp_taskreg): Make sure to generate early debug before
34705 outlining anything from a function.
34706 (expand_omp_target): Likewise.
34707 (grid_expand_target_grid_body): Likewise.
34709 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
34712 * asan.c (get_translation_unit_decl): New function.
34713 (asan_add_global): Extract modules file name from globals
34714 TRANSLATION_UNIT_DECL name.
34716 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
34719 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
34720 for long calls with APCS frame and VFP.
34722 2017-01-24 David Malcolm <dmalcolm@redhat.com>
34724 * cfg.c (original_copy_tables_initialized_p): New function.
34725 * cfg.h (original_copy_tables_initialized_p): New decl.
34726 * cfgrtl.c (relink_block_chain): Guard the call to
34727 free_original_copy_tables with a call to
34728 original_copy_tables_initialized_p.
34729 * cgraph.h (symtab_node::native_rtl_p): New decl.
34730 * cgraphunit.c (symtab_node::native_rtl_p): New function.
34731 (symtab_node::needed_p): Don't assert for early assembly output
34732 for __RTL functions.
34733 (cgraph_node::finalize_function): Set "force_output" for __RTL
34735 (cgraph_node::analyze): Bail out early for __RTL functions.
34736 (analyze_functions): Update assertion to support __RTL functions.
34737 (cgraph_node::expand): Bail out early for __RTL functions.
34738 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
34740 * function.h (struct function): Update comment for field
34742 * gimple-expr.c: Include "tree-pass.h".
34743 (gimple_has_body_p): Return false for __RTL functions.
34744 * Makefile.in (OBJS): Add run-rtl-passes.o.
34745 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
34747 (gcc::pass_manager::get_clean_slate): New accessor.
34748 * passes.c: Include "insn-addr.h".
34749 (should_skip_pass_p): Add logging. Update logic for running
34750 "expand" to be compatible with both __GIMPLE and __RTL. Guard
34751 property-provider override so it is only done for gimple passes.
34753 (skip_pass): New function.
34754 (execute_one_pass): Call skip_pass when skipping passes.
34755 * read-md.c (md_reader::read_char): Support filtering
34756 the input to a subset of line numbers.
34757 (md_reader::md_reader): Initialize fields
34758 m_first_line and m_last_line.
34759 (md_reader::read_file_fragment): New function.
34760 * read-md.h (md_reader::read_file_fragment): New decl.
34761 (md_reader::m_first_line): New field.
34762 (md_reader::m_last_line): New field.
34763 * read-rtl-function.c (function_reader::create_function): Only
34764 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
34765 curr_properties. Set DECL_INITIAL to a dummy block.
34766 (read_rtl_function_body_from_file_range): New function.
34767 * read-rtl-function.h (read_rtl_function_body_from_file_range):
34769 * run-rtl-passes.c: New file.
34770 * run-rtl-passes.h: New file.
34772 2017-01-24 Jeff Law <law@redhat.com>
34774 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
34777 2017-01-24 Bin Cheng <bin.cheng@arm.com>
34779 PR tree-optimization/79159
34780 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
34781 (record_nonwrapping_iv): Improve boundary using above function if no
34782 value range information.
34784 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
34785 Martin Jambor <mjambor@suse.cz>
34787 * brig-builtins.def: New file.
34788 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
34789 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
34790 (DEF_HSAIL_SAT_BUILTIN): Likewise.
34791 (DEF_HSAIL_INTR_BUILTIN): Likewise.
34792 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
34793 * builtin-types.def (BT_INT8): New.
34794 (BT_INT16): Likewise.
34795 (BT_UINT8): Likewise.
34796 (BT_UINT16): Likewise.
34797 (BT_FN_ULONG): Likewise.
34798 (BT_FN_UINT_INT): Likewise.
34799 (BT_FN_UINT_ULONG): Likewise.
34800 (BT_FN_UINT_LONG): Likewise.
34801 (BT_FN_UINT_PTR): Likewise.
34802 (BT_FN_ULONG_PTR): Likewise.
34803 (BT_FN_INT8_FLOAT): Likewise.
34804 (BT_FN_INT16_FLOAT): Likewise.
34805 (BT_FN_UINT32_FLOAT): Likewise.
34806 (BT_FN_UINT16_FLOAT): Likewise.
34807 (BT_FN_UINT8_FLOAT): Likewise.
34808 (BT_FN_UINT64_FLOAT): Likewise.
34809 (BT_FN_UINT16_UINT32): Likewise.
34810 (BT_FN_UINT32_UINT16): Likewise.
34811 (BT_FN_UINT16_UINT16_UINT16): Likewise.
34812 (BT_FN_INT_PTR_INT): Likewise.
34813 (BT_FN_UINT_PTR_UINT): Likewise.
34814 (BT_FN_LONG_PTR_LONG): Likewise.
34815 (BT_FN_ULONG_PTR_ULONG): Likewise.
34816 (BT_FN_VOID_UINT64_UINT64): Likewise.
34817 (BT_FN_UINT8_UINT8_UINT8): Likewise.
34818 (BT_FN_INT8_INT8_INT8): Likewise.
34819 (BT_FN_INT16_INT16_INT16): Likewise.
34820 (BT_FN_INT_INT_INT): Likewise.
34821 (BT_FN_UINT_FLOAT_UINT): Likewise.
34822 (BT_FN_FLOAT_UINT_UINT): Likewise.
34823 (BT_FN_ULONG_UINT_UINT): Likewise.
34824 (BT_FN_ULONG_UINT_PTR): Likewise.
34825 (BT_FN_ULONG_ULONG_ULONG): Likewise.
34826 (BT_FN_UINT_UINT_UINT): Likewise.
34827 (BT_FN_VOID_UINT_PTR): Likewise.
34828 (BT_FN_UINT_UINT_PTR: Likewise.
34829 (BT_FN_UINT32_UINT64_PTR): Likewise.
34830 (BT_FN_INT_INT_UINT_UINT): Likewise.
34831 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
34832 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
34833 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
34834 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
34835 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
34836 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
34837 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
34838 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
34839 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
34840 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
34841 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
34842 * doc/frontends.texi: List BRIG FE.
34843 * doc/install.texi (Testing): Add BRIG tesring requirements.
34844 * doc/invoke.texi (Overall Options): Mention BRIG.
34845 * doc/standards.texi (Standards): Doucment BRIG HSA version.
34847 2017-01-24 Richard Biener <rguenther@suse.de>
34849 PR translation/79208
34850 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
34852 2017-01-24 Martin Jambor <mjambor@suse.cz>
34855 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
34856 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
34857 and known_contexts.
34859 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
34861 PR middle-end/79123
34862 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
34863 casts from signed to unsigned really don't have a range.
34865 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
34867 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
34868 GMP_RNDx for compatiblity.
34870 2017-01-24 Martin Liska <mliska@suse.cz>
34873 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
34874 that would prevent us to call alloca with -1 as argument.
34876 2017-01-24 Jakub Jelinek <jakub@redhat.com>
34878 * dwarf2out.c (output_compilation_unit_header, output_file_names):
34879 Avoid -Wformat-security warning.
34881 2017-01-23 Andrew Pinski <apinski@cavium.com>
34883 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
34886 2017-01-23 Martin Sebor <msebor@redhat.com>
34888 PR middle-end/78703
34889 * gimple-ssa-sprintf.c (warn_level): New global.
34890 (format_integer): Use it here and throughout the rest of the file.
34891 Use the same switch to compute sign as base.
34892 (maybe_warn): New function.
34893 (format_directive): Factor out warnings into maybe_warn.
34894 Add debugging output. Use warn_level.
34895 (add_bytes): Use warn_level.
34896 (pass_sprintf_length::compute_format_length): Add debugging output.
34897 (try_substitute_return_value): Same.
34898 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
34900 PR middle-end/78703
34901 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
34902 (struct fmtresult, format_integer, format_floating): Adjust.
34903 (fmtresult::fmtresult): Set max correctly in two argument ctor.
34904 (get_string_length, format_string,format_directive): Same.
34905 (pass_sprintf_length::compute_format_length): Same.
34906 (try_substitute_return_value): Simplify slightly.
34908 PR middle-end/78703
34909 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
34910 (fmtresult::operator+=): Outlined.
34911 (struct fmtresult): Add ctors.
34912 (struct conversion_spec): Rename...
34913 (struct directive): ...to this. Add and remove data members.
34914 (directive::set_width, directive::set_precision): New functions.
34915 (format_percent): Use fmtresult ctor.
34916 (get_width_and_precision): Remove.
34917 (format_integer): Make naming changes. Avoid computing width and
34919 (format_floating): Same. Adjust indentation.
34920 (format_character, format_none): New functions.
34921 (format_string): Moved character handling to format_character.
34922 (format_directive): Remove arguments, change return type.
34923 (parse_directive): New function.
34924 (pass_sprintf_length::compute_format_length): Move directive
34925 parsing to parse_directive.
34927 2017-01-23 Jakub Jelinek <jakub@redhat.com>
34929 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
34930 (assign_assembler_name_if_needed): ... this.
34931 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
34932 (assign_assembler_name_if_needed): ... this.
34933 (free_lang_data_in_cgraph): Adjust callers.
34934 * cgraphunit.c (cgraph_node::analyze): Likewise.
34935 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
34938 2017-01-23 Richard Biener <rguenther@suse.de>
34940 PR tree-optimization/79088
34941 PR tree-optimization/79188
34942 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
34943 resetting loop bounds after last path deletion. Reset loop
34944 bounds of the target loop, make code match the comments.
34945 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
34946 Make sure loops need no fixups.
34948 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
34950 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
34951 exponent support with double type for first argument.
34952 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
34953 type returned by __builtin_vec_extract_sig,
34954 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
34955 functions from "vector int" to "vector unsigned int" or from
34956 "vector long long int" to "vector unsigned long long int".
34957 Changed type returned by __builtin_vec_extract_exp,
34958 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
34959 functions from "vector int" to "vector unsigned int" or from
34960 "vector long long int" to "vector unsigned long long int".
34961 Changed return type of __builtin_vec_test_data_class,
34962 __builtin_vec_test_data_class_sp, and
34963 __builtin_vec_test_data_class_dp from "vector int" to
34964 "vector bool int" or from "vector long long int" to "vector bool
34965 long long int" and changed second argument type from "unsigned
34966 int" to "int". Added new overloaded function forms "vector float
34967 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
34968 "vector float __builtin_vec_insert_exp_sp (vector float, vector
34969 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
34970 double, vector unsigned long long int)" and "vector double
34971 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
34972 long int)". Changed return type of
34973 __builtin_scalar_test_data_class and
34974 __builtin_scalar_test_data_class_sp and
34975 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
34976 int" and changed second argument from "unsigned int" to "int".
34977 Changed type returned by __builtin_scalar_test_neg,
34978 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
34979 from "int" to "bool int". Added new overloaded function form
34980 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
34981 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
34982 exponent double-precision with floating point first argument.
34983 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
34984 documentation of scalar_test_data_class, scalar_test_neg,
34985 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
34986 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
34987 vec_test_data_class built-in functions to reflect refinements in
34988 their type signatures.
34990 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
34992 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
34994 (aarch64_elf_asm_destructor): Likewise.
34996 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
34998 PR rtl-optimization/78634
34999 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
35000 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
35001 * ifcvt.c (noce_try_cmove): Add missing cost check.
35003 PR rtl-optimization/71724
35004 * combine.c (if_then_else_cond): Look for situations where it is
35005 beneficial to undo the work of one of the recursive calls.
35007 2017-01-23 Bin Cheng <bin.cheng@arm.com>
35009 PR tree-optimization/70754
35010 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
35011 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
35012 combined stmt before it if not NULL.
35013 (combine_chains): Process refs reversely and compute dominance point
35016 2017-01-23 Martin Liska <mliska@suse.cz>
35018 PR tree-optimization/79196
35019 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
35020 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
35022 (strlen_optimize_stmt): Call the renamed function.
35024 2017-01-23 Michael Matz <matz@suse.de>
35026 PR tree-optimization/78384
35027 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
35029 2017-01-23 Richard Biener <rguenther@suse.de>
35031 PR tree-optimization/79186
35032 * tree-vrp.c (register_new_assert_for): Make sure we've seen
35033 both incoming edges before moving an assert.
35035 2017-01-23 Martin Jambor <mjambor@suse.cz>
35037 * ipa-prop.c (load_from_param_1): Removed.
35038 (load_from_unmodified_param): Bits from load_from_param_1 put back
35040 (load_from_param): Removed.
35041 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
35042 with stmt. Reverted back to use of load_from_unmodified_param.
35044 2017-01-23 Martin Jambor <mjambor@suse.cz>
35047 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
35048 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
35049 field a pointer to garbage collected vector, mark lattices and
35050 ipcp_orig_node with GTY((skip)).
35051 (ipa_get_param_count): Adjust to descriptors being a pointer.
35052 (ipa_get_param): Likewise.
35053 (ipa_get_type): Likewise.
35054 (ipa_get_param_move_cost): Likewise.
35055 (ipa_set_param_used): Likewise.
35056 (ipa_get_controlled_uses): Likewise.
35057 (ipa_set_controlled_uses): Likewise.
35058 (ipa_is_param_used): Likewise.
35059 (ipa_node_params_t): Move into garbage collector. New methods insert
35061 (ipa_node_params_sum): Annotate wth GTY(()).
35062 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
35064 (ipa_load_from_parm_agg): Adjust declaration.
35065 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
35066 * ipa-profile.c (ipa_profile): Likewise.
35067 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
35068 (ipa_populate_param_decls): Make descriptors parameter garbage
35070 (ipa_dump_param): Adjust to descriptors being a pointer.
35071 (ipa_alloc_node_params): Likewise.
35072 (ipa_initialize_node_params): Likewise.
35073 (load_from_param_1): Make descriptors parameter garbage collected.
35074 (load_from_unmodified_param): Likewise.
35075 (load_from_param): Likewise.
35076 (ipa_load_from_parm_agg): Likewise.
35077 (ipa_node_params::~ipa_node_params): Removed.
35078 (ipa_free_all_node_params): Remove call to delete operator.
35079 (ipa_node_params_t::insert): New.
35080 (ipa_node_params_t::remove): Likewise.
35081 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
35082 copy known_csts and known_contexts vectors.
35083 (ipa_read_node_info): Adjust to descriptors being a pointer.
35084 (ipcp_modif_dom_walker): Make m_descriptors field garbage
35086 (ipcp_transform_function): Make descriptors variable garbage
35089 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
35091 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
35092 * config/i386/avx512dqintrin.h: Ditto.
35093 * config/i386/avx512fintrin.h: Ditto.
35094 * config/i386/i386.c: Handle new builtins.
35095 * config/i386/i386-builtin.def: Add new builtins.
35096 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
35097 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
35099 2017-01-23 Jakub Jelinek <jakub@redhat.com>
35100 Martin Liska <mliska@suse.cz>
35102 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
35103 * asan.c (asan_expand_poison_ifn): Support stores and use
35104 appropriate ASAN report function.
35105 * internal-fn.c (expand_ASAN_POISON_USE): New function.
35106 * internal-fn.def (ASAN_POISON_USE): Declare.
35107 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
35108 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
35109 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
35110 ASAN_POISON calls w/o LHS.
35111 * tree-ssa.c (execute_update_addresses_taken): Create clobber
35112 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
35113 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
35114 * gimplify.c (asan_poison_variables): Add attribute
35115 use_after_scope_memory to variables that really needs to live
35117 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
35118 having the attribute.
35120 2017-01-23 Martin Liska <mliska@suse.cz>
35122 * asan.c (create_asan_shadow_var): New function.
35123 (asan_expand_poison_ifn): Likewise.
35124 * asan.h (asan_expand_poison_ifn): New declaration.
35125 * internal-fn.c (expand_ASAN_POISON): Likewise.
35126 * internal-fn.def (ASAN_POISON): New builtin.
35127 * sanopt.c (pass_sanopt::execute): Expand
35128 asan_expand_poison_ifn.
35129 * tree-inline.c (copy_decl_for_dup_finish): Make function
35131 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
35132 * tree-ssa.c (is_asan_mark_p): New function.
35133 (execute_update_addresses_taken): Rewrite local variables
35134 (identified just by use-after-scope as addressable) into SSA.
35136 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
35138 * doc/install.texi (Specific): opensource.apple.com uses https
35139 now. Remove trailing slash.
35141 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
35143 * README.Portability: Remove note on an Irix compatibility issue.
35145 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
35147 * gcov.c (INCLUDE_ALGORITHM): Define.
35148 (INCLUDE_VECTOR): Define.
35149 No longer include <vector> and <algorithm> directly.
35151 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
35153 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
35155 * doc/invoke.texi (Code Gen Options): Ditto.
35157 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
35160 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
35162 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
35164 rtl-optimization/79125
35165 * cprop.c (local_cprop_pass): Handle cases where we make an
35166 unconditional trap.
35168 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
35172 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
35173 read from, for big endian.
35175 2017-01-20 Jiong Wang <jiong.wang@arm.com>
35177 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
35178 register pauth builtins for LP64 only.
35180 2017-01-20 Marek Polacek <polacek@redhat.com>
35183 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
35186 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
35188 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
35190 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
35191 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
35192 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
35194 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35197 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
35198 in big-endian mode when they are not a single duplicated value.
35200 2017-01-20 Richard Biener <rguenther@suse.de>
35202 * BASE-VER: Bump to 7.0.1.
35204 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
35206 * omp-low.c (omplow_simd_context): New struct. Use it...
35207 (lower_rec_simd_input_clauses): ...here and...
35208 (lower_rec_input_clauses): ...here to hold common data. Adjust all
35209 references to idx, lane, max_vf, is_simt.
35211 2017-01-20 Graham Markall <graham.markall@embecosm.com>
35213 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
35216 2017-01-20 Martin Jambor <mjambor@suse.cz>
35218 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
35219 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
35221 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
35222 (GTFILES): Rename hsa.c to hsa-common.c.
35223 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
35224 * hsa-dump.c: Likewise.
35225 * hsa-gen.c: Likewise.
35226 * hsa-regalloc.c: Likewise.
35227 * ipa-hsa.c: Likewise.
35228 * omp-expand.c: Likewise.
35229 * omp-low.c: Likewise.
35230 * toplev.c: Likewise.
35232 2017-01-20 Marek Polacek <polacek@redhat.com>
35235 * doc/invoke.texi: Document -Wduplicated-branches.
35236 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
35237 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
35238 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
35239 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
35240 return 0 only when not OEP_LEXICOGRAPHIC.
35241 (fold_build_cleanup_point_expr): Use the expression
35242 location when building CLEANUP_POINT_EXPR.
35243 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
35244 * tree.c (add_expr): Handle error_mark_node.
35246 2017-01-20 Martin Liska <mliska@suse.cz>
35249 * tree-profile.c (init_ic_make_global_vars): Do not call
35251 (gimple_init_gcov_profiler): Likewise.
35253 2017-01-20 Martin Liska <mliska@suse.cz>
35256 * cgraph.h (maybe_create_reference): Remove argument and
35258 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
35260 * ipa-cp.c (create_specialized_node): Likewise.
35261 * symtab.c (symtab_node::maybe_create_reference): Handle
35262 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
35264 2017-01-20 Martin Liska <mliska@suse.cz>
35266 * read-rtl-function.c (function_reader::create_function): Use
35267 build_decl instread of build_decl_stat.
35269 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
35271 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
35272 * config/i386/avx512dqintrin.h: Ditto.
35273 * config/i386/avx512fintrin.h: Ditto.
35274 * config/i386/i386-builtin-types.def: Add new types.
35275 * config/i386/i386.c: Handle new types.
35276 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
35277 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
35278 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
35279 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
35280 (__builtin_ia32_kshiftridi): New.
35281 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
35283 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
35287 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
35288 define to rs6000_init_stack_protect_guard.
35289 (rs6000_init_stack_protect_guard): New function.
35291 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
35292 Yunqiang Su <yunqiang.su@imgtec.com>
35294 * config.gcc (supported_defaults): Add madd4.
35295 (with_madd4): Add validation.
35296 (all_defaults): Add madd4.
35297 * config/mips/mips.opt (mmadd4): New option.
35298 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
35300 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
35302 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
35303 (ISA_HAS_FUSED_MADD4): Likewise.
35304 * doc/invoke.texi (-mmadd4): Document the new option.
35305 * doc/install.texi (--with-madd4): Document the new option.
35307 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35309 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
35310 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
35311 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
35312 (aarch64_init_pauth_hint_builtins): New.
35313 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
35314 (aarch64_expand_builtin): Expand new builtins.
35316 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35318 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
35319 * combine-stack-adj.c (no_unhandled_cfa): Handle
35320 REG_CFA_TOGGLE_RA_MANGLE.
35321 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
35322 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
35323 info for return address signing.
35324 (aarch64_expand_epilogue): Likewise.
35326 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35328 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
35329 * config/aarch64/aarch64-protos.h
35330 (aarch64_return_address_signing_enabled): New declaration.
35331 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
35333 (aarch64_expand_prologue): Sign return address before it's pushed onto
35335 (aarch64_expand_epilogue): Authenticate return address fetched from
35337 (aarch64_override_options): Sanity check for ILP32 and ISA level.
35338 (aarch64_attributes): New function attributes for "sign-return-address".
35339 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
35340 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
35341 ("*do_return"): Generate combined instructions according to key index.
35342 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
35343 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
35345 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
35346 * config/aarch64/aarch64.opt (msign-return-address=): New.
35347 * doc/extend.texi (AArch64 Function Attributes): Documents
35348 "sign-return-address=".
35349 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
35351 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
35353 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
35354 overall option summary.
35356 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35358 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
35359 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
35360 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
35361 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
35363 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
35365 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
35366 -mpower9-minmax by default for -mcpu=power9.
35367 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
35368 128-bit floating point.
35370 2017-01-20 Alan Modra <amodra@gmail.com>
35372 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
35373 optimizing for size.
35375 2017-01-20 Alan Modra <amodra@gmail.com>
35378 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
35379 for strcmp and strncmp from corresponding builtin decl.
35381 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
35383 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
35384 instead of i386/rtems-64.h.
35385 * config/i386/rtems-64.h: Remove.
35387 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
35391 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
35393 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
35395 2017-01-19 Tamar Christina <tamar.christina@arm.com>
35397 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
35398 Change int to HOST_WIDE_INT.
35399 * config/aarch64/aarch64-protos.h
35400 (aarch64_simd_gen_const_vector_dup): Likewise.
35401 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
35403 2017-01-19 David Malcolm <dmalcolm@redhat.com>
35405 * langhooks-def.h (lhd_type_for_size): New decl.
35406 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
35407 * langhooks.c (lhd_type_for_size): New function, taken from
35410 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
35412 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
35413 define_bypass for CR latency.
35414 (power9-cracked-alu): Update bypass latency and remove power9-branch.
35415 (power9-alu2): Add define_bypass for CR latency.
35417 (power9-mul): Update insn latency.
35418 (power9-mul-compare): Update insn latency, bypass latency and remove
35421 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35423 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
35425 * config/aarch64/aarch64.md
35426 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
35427 aarch64_nopcrelative_literal_loads.
35428 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
35430 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
35432 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
35433 TARGET_LOONGSON_3A.
35434 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
35436 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
35439 * config.gcc (supported_defaults): Add lxc1-sxc1.
35440 (with_lxc1_sxc1): Add validation.
35441 (all_defaults): Add lxc1-sxc1.
35442 * config/mips/mips.opt (mlxc1-sxc1): New option.
35443 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
35445 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
35446 __mips_no_lxc1_sxc1.
35447 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
35448 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
35449 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
35451 2017-01-19 Richard Biener <rguenther@suse.de>
35453 PR tree-optimization/72488
35454 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
35455 sure to restore SSA info.
35456 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
35458 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
35460 PR rtl-optimization/79121
35461 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
35462 of the inner type when shifting an extended value.
35464 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
35467 * symtab.c (symtab_node::equal_address_to): Fix comparing of
35468 interposable aliases.
35470 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
35473 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
35474 Use the evmergelohi instruction.
35475 (mov_si<mode>_e500_subreg4_2_le): Likewise.
35476 (mov_sitf_e500_subreg8_2_be): Likewise.
35477 (mov_sitf_e500_subreg12_2_le): Likewise.
35478 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
35479 (mov_si<mode>_e500_subreg4_2_be): Likewise.
35480 (mov_sitf_e500_subreg8_2_le): Likewise.
35481 (mov_sitf_e500_subreg12_2_be): Likewise.
35483 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35485 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
35486 attribute from vecsimple to vecperm.
35487 (altivec_vbpermq2): Likewise.
35489 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35492 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
35494 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35495 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
35496 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
35497 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
35498 case where N arg is SIZE_MAX.
35499 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
35500 (cmpstrsi): Add pattern.
35502 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
35504 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
35505 __builtin_vec_revb builtins.
35506 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
35507 built-in functions to support generation of the ISA 3.0 XXBR<x>
35508 vector byte reverse instructions.
35509 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
35510 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
35511 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
35512 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
35513 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
35514 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
35515 (P9V_BUILTIN_VEC_REVB): Likewise.
35516 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
35517 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
35518 (p9_xxbrq_v16qi): Likewise.
35519 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
35520 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
35521 (p9_xxbrh_v8hi): Likewise.
35522 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
35523 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
35524 vec_revb built-in functions.
35526 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
35528 PR rtl-optimization/78952
35529 * config/i386/i386.md (any_extract): New code iterator.
35530 (*insvqi_2): Use any_extract for source operand.
35531 (*insvqi_3): Use any_shiftrt for source operand.
35533 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
35535 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
35537 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
35539 2017-01-18 Matthias Klose <doko@ubuntu.com>
35541 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
35543 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35545 * config/rs6000/altivec.h (vec_bperm): Change #define.
35546 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
35547 (altivec_vbpermq2): New define_insn.
35548 (altivec_vbpermd): Likewise.
35549 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
35550 function interface.
35551 (VBPERMD): Likewise.
35552 (VBPERM): New polymorphic function interface.
35553 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
35554 Add entries for P9V_BUILTIN_VEC_VBPERM.
35555 * doc/extend.texi: Add interfaces for vec_bperm.
35557 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35559 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
35560 first letter of error messages.
35561 (s390_resolve_overloaded_builtin): Likewise.
35562 * config/s390/s390.c (s390_expand_builtin): Likewise.
35563 (s390_invalid_arg_for_unprototyped_fn): Likewise.
35564 (s390_valid_target_attribute_inner_p): Likewise.
35565 * config/s390/s390.md ("tabort"): Likewise.
35567 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
35569 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
35570 (ISA_AVOID_DIV_HILO): New macro.
35571 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
35572 (ISA_HAS_DDIV): Likewise.
35574 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35576 * doc/invoke.texi (fabi-version): Correct number of occurrences.
35578 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35580 * doc/invoke.texi (fabi-version): Spelling fix.
35582 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35585 * doc/invoke.texi (fabi-version): Mention mangling fix for
35588 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
35591 * doc/invoke.texi (fabi-version): Document discriminator mangling.
35593 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
35596 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
35597 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
35599 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
35601 (stack_protect_test): Ditto.
35602 * config/rs6000/rs6000.opt (mstack-protector-guard=,
35603 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
35605 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
35606 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
35607 -mstack-protector-guard-offset=.
35608 (RS/6000 and PowerPC Options): Ditto.
35610 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
35612 * config/i386/i386.h (MASK_CLASS_P): New define.
35613 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
35614 there are no registers from different register sets also when
35615 mask registers are used. Update function comment.
35616 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
35617 to (*k/*r) and (*k/*km) alternatives.
35619 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
35621 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
35622 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
35623 (EH_RETURN_HANDLER_RTX): New define.
35624 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
35625 Force frame pointer in EH return functions.
35626 (aarch64_expand_epilogue): Add barrier for eh_return.
35627 (aarch64_final_eh_return_addr): Remove.
35628 (aarch64_eh_return_handler_rtx): New function.
35629 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
35631 (aarch64_eh_return_handler_rtx): New prototype.
35633 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35635 * config/rs6000/altivec.h (vec_rlmi): New #define.
35636 (vec_vrlnm): Likewise.
35637 (vec_rlnm): Likewise.
35638 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
35639 (UNSPEC_VRLNM): Likewise.
35640 (VIlong): New mode iterator.
35641 (altivec_vrl<VI_char>mi): New define_insn.
35642 (altivec_vrl<VI_char>nm): Likewise.
35643 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
35645 (VRLDNM): Likewise.
35646 (RLNM): New polymorphic function entry.
35647 (VRLWMI): New monomorphic function entry.
35648 (VRLDMI): Likewise.
35649 (RLMI): New polymorphic function entry.
35650 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
35651 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
35652 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
35655 2017-01-17 Jakub Jelinek <jakub@redhat.com>
35658 * dwarf2out.c (field_byte_offset): Restore the
35659 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
35660 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
35661 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
35664 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
35667 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
35669 2017-01-17 Jakub Jelinek <jakub@redhat.com>
35672 * dwarf2out.c (add_data_member_location_attribute): For constant
35673 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
35674 instead of DW_AT_data_member_location, DW_AT_bit_offset and
35675 DW_AT_byte_size attributes.
35677 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
35679 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
35680 after forcing to constant memory when the code model is medium.
35682 2017-01-17 Julia Koval <julia.koval@intel.com>
35685 * config/i386/avx512fintrin.h
35686 (_mm512_i32gather_ps): Change __addr type to void const*.
35687 (_mm512_mask_i32gather_ps): Ditto.
35688 (_mm512_i32gather_pd): Ditto.
35689 (_mm512_mask_i32gather_pd): Ditto.
35690 (_mm512_i64gather_ps): Ditto.
35691 (_mm512_mask_i64gather_ps): Ditto.
35692 (_mm512_i64gather_pd): Ditto.
35693 (_mm512_mask_i64gather_pd): Ditto.
35694 (_mm512_i32gather_epi32): Ditto.
35695 (_mm512_mask_i32gather_epi32): Ditto.
35696 (_mm512_i32gather_epi64): Ditto.
35697 (_mm512_mask_i32gather_epi64): Ditto.
35698 (_mm512_i64gather_epi32): Ditto.
35699 (_mm512_mask_i64gather_epi32): Ditto.
35700 (_mm512_i64gather_epi64): Ditto.
35701 (_mm512_mask_i64gather_epi64): Ditto.
35702 (_mm512_i32scatter_ps): Change __addr type to void*.
35703 (_mm512_mask_i32scatter_ps): Ditto.
35704 (_mm512_i32scatter_pd): Ditto.
35705 (_mm512_mask_i32scatter_pd): Ditto.
35706 (_mm512_i64scatter_ps): Ditto.
35707 (_mm512_mask_i64scatter_ps): Ditto.
35708 (_mm512_i64scatter_pd): Ditto.
35709 (_mm512_mask_i64scatter_pd): Ditto.
35710 (_mm512_i32scatter_epi32): Ditto.
35711 (_mm512_mask_i32scatter_epi32): Ditto.
35712 (_mm512_i32scatter_epi64): Ditto.
35713 (_mm512_mask_i32scatter_epi64): Ditto.
35714 (_mm512_i64scatter_epi32): Ditto.
35715 (_mm512_mask_i64scatter_epi32): Ditto.
35716 (_mm512_i64scatter_epi64): Ditto.
35717 (_mm512_mask_i64scatter_epi64): Ditto.
35718 * config/i386/avx512pfintrin.h
35719 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
35720 (_mm512_mask_prefetch_i32gather_ps): Ditto.
35721 (_mm512_mask_prefetch_i64gather_pd): Ditto.
35722 (_mm512_mask_prefetch_i64gather_ps): Ditto.
35723 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
35724 (_mm512_prefetch_i32scatter_ps): Ditto.
35725 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
35726 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
35727 (_mm512_prefetch_i64scatter_pd): Ditto.
35728 (_mm512_prefetch_i64scatter_ps): Ditto.
35729 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
35730 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
35731 * config/i386/avx512vlintrin.h
35732 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
35733 (_mm_mmask_i32gather_ps): Ditto.
35734 (_mm256_mmask_i32gather_pd): Ditto.
35735 (_mm_mmask_i32gather_pd): Ditto.
35736 (_mm256_mmask_i64gather_ps): Ditto.
35737 (_mm_mmask_i64gather_ps): Ditto.
35738 (_mm256_mmask_i64gather_pd): Ditto.
35739 (_mm_mmask_i64gather_pd): Ditto.
35740 (_mm256_mmask_i32gather_epi32): Ditto.
35741 (_mm_mmask_i32gather_epi32): Ditto.
35742 (_mm256_mmask_i32gather_epi64): Ditto.
35743 (_mm_mmask_i32gather_epi64): Ditto.
35744 (_mm256_mmask_i64gather_epi32): Ditto.
35745 (_mm_mmask_i64gather_epi32): Ditto.
35746 (_mm256_mmask_i64gather_epi64): Ditto.
35747 (_mm_mmask_i64gather_epi64): Ditto.
35748 (_mm256_i32scatter_ps): Change __addr type to void*.
35749 (_mm256_mask_i32scatter_ps): Ditto.
35750 (_mm_i32scatter_ps): Ditto.
35751 (_mm_mask_i32scatter_ps): Ditto.
35752 (_mm256_i32scatter_pd): Ditto.
35753 (_mm256_mask_i32scatter_pd): Ditto.
35754 (_mm_i32scatter_pd): Ditto.
35755 (_mm_mask_i32scatter_pd): Ditto.
35756 (_mm256_i64scatter_ps): Ditto.
35757 (_mm256_mask_i64scatter_ps): Ditto.
35758 (_mm_i64scatter_ps): Ditto.
35759 (_mm_mask_i64scatter_ps): Ditto.
35760 (_mm256_i64scatter_pd): Ditto.
35761 (_mm256_mask_i64scatter_pd): Ditto.
35762 (_mm_i64scatter_pd): Ditto.
35763 (_mm_mask_i64scatter_pd): Ditto.
35764 (_mm256_i32scatter_epi32): Ditto.
35765 (_mm256_mask_i32scatter_epi32): Ditto.
35766 (_mm_i32scatter_epi32): Ditto.
35767 (_mm_mask_i32scatter_epi32): Ditto.
35768 (_mm256_i32scatter_epi64): Ditto.
35769 (_mm256_mask_i32scatter_epi64): Ditto.
35770 (_mm_i32scatter_epi64): Ditto.
35771 (_mm_mask_i32scatter_epi64): Ditto.
35772 (_mm256_i64scatter_epi32): Ditto.
35773 (_mm256_mask_i64scatter_epi32): Ditto.
35774 (_mm_i64scatter_epi32): Ditto.
35775 (_mm_mask_i64scatter_epi32): Ditto.
35776 (_mm256_i64scatter_epi64): Ditto.
35777 (_mm256_mask_i64scatter_epi64): Ditto.
35778 (_mm_i64scatter_epi64): Ditto.
35779 (_mm_mask_i64scatter_epi64): Ditto.
35780 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
35781 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
35782 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
35783 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
35784 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
35785 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
35786 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
35787 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
35788 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
35789 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
35790 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
35791 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
35792 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
35793 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
35794 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
35795 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
35796 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
35797 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
35798 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
35799 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
35800 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
35801 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
35802 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
35803 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
35804 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
35805 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
35806 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
35807 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
35808 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
35809 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
35810 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
35811 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
35812 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
35813 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
35814 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
35815 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
35816 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
35817 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
35818 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
35819 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
35820 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
35821 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
35822 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
35823 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
35824 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
35825 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
35826 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
35827 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
35828 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
35829 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
35830 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
35831 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
35832 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
35833 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
35834 definitions accordingly.
35836 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
35837 Kuan-Lin Chen <kuanlinchentw@gmail.com>
35840 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
35843 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
35846 * ira-conflicts.c (ira_build_conflicts): Update total conflict
35847 hard regs for inner regno.
35849 2017-01-17 Martin Liska <mliska@suse.cz>
35852 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
35853 assumption and add comment.
35855 2017-01-17 Nathan Sidwell <nathan@acm.org>
35857 * ipa-visibility.c (localize_node): New function, broken out of ...
35858 (function_and_variable_visibility): ... here. Call it.
35860 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
35862 PR middle-end/77445
35863 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
35864 correctly set frequency of oudgoing edge.
35865 (duplicate_thread_path): Fix profile updating.
35867 2017-01-17 Jakub Jelinek <jakub@redhat.com>
35870 * configure.ac: Add GCC_BASE_VER.
35871 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
35872 version from BASE-VER file.
35873 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
35874 (gcc.o): Depend on $(BASEVER).
35875 * common.opt (dumpfullversion): New option.
35876 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
35877 * doc/invoke.texi: Document -dumpfullversion.
35878 * doc/install.texi: Document --with-gcc-major-version-only.
35879 * configure: Regenerated.
35881 2017-01-17 Richard Biener <rguenther@suse.de>
35883 PR tree-optimization/71433
35884 * tree-vrp.c (register_new_assert_for): Merge same asserts
35885 on all incoming edges.
35886 (process_assert_insertions_for): Handle insertions at the
35889 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
35891 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
35892 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
35894 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
35897 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
35900 2017-01-17 Alan Modra <amodra@gmail.com>
35903 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
35904 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
35905 symbolic stack limit when pic.
35907 2017-01-16 Martin Sebor <msebor@redhat.com>
35909 PR tree-optimization/78608
35910 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
35912 2017-01-16 Jeff Law <law@redhat.com>
35915 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
35916 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
35917 for several include directories that may be relative to sysroot.
35918 * config/i386/x-mingw32 (gplus_includedir): Define.
35919 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
35920 (native_system_includedir): Likewise.
35921 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
35922 override if TARGET_SYSTEM_ROOT is defined.
35923 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
35925 PR tree-optimization/79090
35926 PR tree-optimization/33562
35927 PR tree-optimization/61912
35928 PR tree-optimization/77485
35929 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
35930 and computed trims into the dump file.
35932 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
35934 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
35936 2017-01-16 Jakub Jelinek <jakub@redhat.com>
35939 * gimplify.c (gimplify_init_constructor): If want_value and
35940 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
35944 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
35945 sequence. Formatting fixes.
35946 (doloop_optimize): Formatting fixes.
35949 * gcc.c (debug_level_greater_than_spec_func): New function.
35950 (static_spec_functions): Add debug-level-gt spec function.
35951 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
35953 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
35954 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
35955 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
35956 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
35957 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
35958 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
35960 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
35962 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
35963 QImode fixups to general and mask registers only.
35965 2017-01-16 Carl Love <cel@us.ibm.com>
35967 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
35968 for built-in functions
35969 vector signed char vec_nabs (vector signed char)
35970 vector signed short vec_nabs (vector signed short)
35971 vector signed int vec_nabs (vector signed int)
35972 vector signed long long vec_nabs (vector signed long long)
35973 vector float vec_nabs (vector float)
35974 vector double vec_nabs (vector double)
35975 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
35977 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
35978 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
35979 * doc/extend.texi: Update the documentation file for the new built-in
35982 2017-01-16 Martin Sebor <msebor@redhat.com>
35984 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
35987 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35989 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
35990 UNSPEC_VSX__XXSPLTD to require special splat handling.
35992 2017-01-16 David Malcolm <dmalcolm@redhat.com>
35995 * system.h: Poison strndup.
35997 2017-01-16 Alan Modra <amodra@gmail.com>
36000 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
36003 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
36005 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
36007 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
36009 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
36010 call recog here. Assert that INSN_CODE (insn) is non-negative.
36012 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
36015 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
36017 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
36018 in the currently scheduled RTL fragment.
36020 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
36022 PR rtl-optimization/78751
36023 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
36026 2017-01-14 Jeff Law <law@redhat.com>
36028 PR tree-optimization/79090
36029 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
36030 variable length stores.
36031 (compute_trims): Delete dead assignment to *trim_tail.
36032 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
36035 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
36037 PR rtl-optimization/78626
36038 PR rtl-optimization/78727
36039 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
36040 of a block, and split such blocks after everything else is finished.
36042 2017-01-14 Alan Modra <amodra@gmail.com>
36045 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
36046 target legitimate_combined_insn.
36047 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
36048 (rs6000_legitimate_combined_insn): New function.
36049 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
36051 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
36052 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
36053 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
36055 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
36057 * doc/frontends.texi (G++ and GCC): Remove references to Java.
36059 2017-01-13 Jeff Law <law@redhat.com>
36061 PR tree-optimization/33562
36062 PR tree-optimization/61912
36063 PR tree-optimization/77485
36064 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
36066 (delete_dead_assignment): Likewise.
36067 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
36068 statement to delete_dead_call and delete_dead_assignment.
36070 2017-01-13 David Malcolm <dmalcolm@redhat.com>
36073 * substring-locations.c (format_warning_va): Strengthen case 1 so
36074 that both endpoints of the substring must be within the format
36075 range for just the substring to be printed.
36077 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
36079 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
36080 * config/i386/i386.c (ix86_target_string): Add missing options
36081 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
36082 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
36083 flags_other and ix86_target_other to flags2_other. Display unknown
36085 (ix86_valid_target_attribute_inner_p): Add missing options and
36086 reorder options by implied ISAs, as in ix86_target_string.
36088 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
36090 * hash-table.h (hash_table::too_empty_p): New function.
36091 (hash_table::expand): Use it.
36092 (hash_table::traverse): Likewise.
36093 (hash_table::empty_slot): Use sizeof (value_type) instead of
36094 sizeof (PTR) to convert bytes to elements. Shrink the table
36095 if the current size is excessive for the current number of
36098 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
36100 * ira-costs.c (record_reg_classes): Break from the inner loop
36101 early once alt_fail is known to be true. Update outer loop
36102 handling accordingly.
36104 2017-01-13 Jeff Law <law@redhat.com>
36106 * tree-ssa-dse.c (decrement_count): New function.
36107 (increment_start_addr, maybe_trim_memstar_call): Likewise.
36108 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
36109 when we know the partially dead statement is a mem* function.
36111 PR tree-optimization/61912
36112 PR tree-optimization/77485
36113 * tree-ssa-dse.c: Include expr.h.
36114 (maybe_trim_constructor_store): New function.
36115 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
36117 PR tree-optimization/33562
36118 PR tree-optimization/61912
36119 PR tree-optimization/77485
36120 * doc/invoke.texi: Document new dse-max-object-size param.
36121 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
36122 * tree-ssa-dse.c: Include params.h.
36123 (dse_store_status): New enum.
36124 (initialize_ao_ref_for_dse): New, partially extracted from
36126 (valid_ao_ref_for_dse, normalize_ref): New.
36127 (setup_live_bytes_from_ref, compute_trims): Likewise.
36128 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
36129 (maybe_trim_partially_dead_store): Likewise.
36130 (maybe_trim_complex_store): Likewise.
36131 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
36132 Track what bytes live from the original store. Return tri-state
36133 for dead, partially dead or live.
36134 (dse_dom_walker): Add constructor, destructor and new private members.
36135 (delete_dead_call, delete_dead_assignment): New extracted from
36137 (dse_optimize_stmt): Make a member of dse_dom_walker.
36138 Use initialize_ao_ref_for_dse.
36140 PR tree-optimization/33562
36141 PR tree-optimization/61912
36142 PR tree-optimization/77485
36143 * sbitmap.h (bitmap_count_bits): Prototype.
36144 (bitmap_clear_range, bitmap_set_range): Likewise.
36145 * sbitmap.c (bitmap_clear_range): New function.
36146 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
36148 2017-01-13 Martin Liska <mliska@suse.cz>
36151 * function.c (set_cfun): Add new argument force.
36152 * function.h (set_cfun): Likewise.
36153 * ipa-inline-transform.c (inline_call): Use the function when
36154 strict alising from is dropped for function we inline to.
36156 2017-01-13 Richard Biener <rguenther@suse.de>
36158 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
36159 for dumping GIMPLE INTEGER_CSTs.
36161 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
36163 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
36164 to 201112L since C++17.
36166 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
36169 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
36170 if -fsanitize=kernel-address is present.
36172 2017-01-13 Richard Biener <rguenther@suse.de>
36174 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
36175 as _Literal ( type ) number in case usual suffixes do not
36176 preserve all information.
36178 2017-01-13 Richard Biener <rguenther@suse.de>
36180 PR tree-optimization/77283
36181 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
36182 and ssa-iterators.h.
36183 (is_feasible_trace): Implement a cost model based on joiner
36186 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
36189 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
36190 char or short to __float128/_Float128 directly.
36192 2017-01-12 Martin Sebor <msebor@redhat.com>
36194 to -Wformat-overflow.
36195 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
36196 (min_bytes_remaining): Same.
36197 (get_string_length): Same.
36198 (format_string): Same.
36199 (format_directive): Same.
36201 (pass_sprintf_length::handle_gimple_call): Same.
36203 2017-01-12 Jakub Jelinek <jakub@redhat.com>
36205 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
36206 info.nowrite calls with no lhs that can't throw. Return bool
36207 whether gsi_remove has been called or not.
36208 (pass_sprintf_length::handle_gimple_call): Return bool whether
36209 try_substitute_return_value called gsi_remove. Formatting fix.
36210 (pass_sprintf_length::execute): Don't use gsi_remove if
36211 handle_gimple_call returned true.
36214 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
36215 be removed due to side-effects, don't remove following barrier nor
36216 turn the successor edge into fallthru edge.
36218 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36221 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
36222 element-reversing loads and stores as not swappable.
36224 2017-01-12 Nathan Sidwell <nathan@acm.org>
36225 Nicolai Stange <nicstange@gmail.com>
36227 * combine.c (try_combine): Don't ignore result of overlap checking
36228 loop. Combine overlap & asm check into single loop.
36230 2017-01-12 Richard Biener <rguenther@suse.de>
36232 * tree-pretty-print.c (dump_generic_node): Provide -gimple
36233 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
36235 2017-01-12 Richard Biener <rguenther@suse.de>
36237 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
36238 and TS_TARGET_OPTION directly derive from TS_BASE.
36239 * tree-core.h (tree_optimization_option): Derive from tree_base.
36240 (tree_target_option): Likewise.
36242 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
36244 * config/i386/i386.c (memory_address_length): Increase len
36245 only when rip_relative_addr_p returns false.
36247 2017-01-11 Julia Koval <julia.koval@intel.com>
36249 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
36250 (OPTION_MASK_ISA_SGX_SET): New.
36251 (ix86_handle_option): Handle OPT_msgx.
36252 * config.gcc: Added sgxintrin.h.
36253 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
36254 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
36255 * config/i386/i386.c (ix86_target_string): Add -msgx.
36257 (ix86_option_override_internal): Handle new options.
36258 (ix86_valid_target_attribute_inner_p): Add sgx.
36259 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
36260 * config/i386/i386.opt: Add msgx.
36261 * config/i386/sgxintrin.h: New file.
36262 * config/i386/x86intrin.h: Add sgxintrin.h.
36264 2017-01-11 Jakub Jelinek <jakub@redhat.com>
36267 * fold-const.c (maybe_nonzero_address): Return 1 for function
36269 (tree_single_nonzero_warnv_p): Don't handle function local objects
36273 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
36276 2017-01-11 David Malcolm <dmalcolm@redhat.com>
36279 * opts.c: Include "spellcheck.h"
36280 (struct string_fragment): New struct.
36281 (struct edit_distance_traits<const string_fragment &>): New
36283 (get_closest_sanitizer_option): New function.
36284 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
36286 2017-01-11 Jakub Jelinek <jakub@redhat.com>
36288 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
36290 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
36291 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
36292 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
36293 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
36294 for initial die_offset if dwarf_split_debug_info.
36295 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
36296 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
36298 (output_skeleton_debug_sections): Formatting fix. Use
36299 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
36300 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
36302 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
36304 * config/arm/cortex-a53.md: Add bypasses for
36305 cortex_a53_r2f_cvt.
36306 (cortex_a53_r2f): Only use for transfers.
36307 (cortex_a53_f2r): Likewise.
36308 (cortex_a53_r2f_cvt): Add reservation for conversions.
36309 (cortex_a53_f2r_cvt): Likewise.
36311 2017-01-11 Tamar Christina <tamar.christina@arm.com>
36313 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
36314 to all inlined functions, change static to extern.
36316 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
36319 * config/arm/arm.c (legitimize_pic_address): Handle reference to
36321 (arm_assemble_integer): Likewise.
36323 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
36325 * config.gcc: Use new awk script to check CPU, FPU and architecture
36326 parameters for --with-... options.
36327 * config/arm/parsecpu.awk: New file
36328 * config/arm/arm-cpus.in: New file.
36329 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
36331 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
36333 * config/arm/t-arm: Update dependency rules.
36334 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
36335 of processing .def files.
36336 * config/arm/genopt.sh: Deleted.
36337 * config/arm/gentune.sh: Deleted.
36338 * config/arm/arm-cores.def: Deleted.
36339 * config/arm/arm-arches.def: Deleted.
36340 * config/arm/arm-fpus.def: Deleted.
36341 * config/arm/arm-tune.md: Regenerated.
36342 * config/arm/arm-tables.opt: Regenerated.
36343 * config/arm/arm-cpu.h: New generated file.
36344 * config/arm/arm-cpu-data.h: New generated file.
36345 * config/arm/arm-cpu-cdata.h: New generated file.
36347 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
36350 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
36352 (input_varpool_node): Unpack dynamically_initialized bit.
36354 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
36356 PR rtl-optimization/79032
36357 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
36358 the alignment of the adjusted memory reference against that of MODE,
36359 instead of the alignment of the original memory reference.
36361 2017-01-11 Martin Jambor <mjambor@suse.cz>
36363 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
36365 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
36366 decorated functions.
36368 2017-01-11 Richard Biener <rguenther@suse.de>
36370 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
36371 set range/nonnull info for PHI results. Do not set it on
36372 stmts marked for removal.
36374 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
36376 * expr.c (store_field): In the bitfield case, fetch the return value
36377 from the registers before applying a single big-endian adjustment.
36378 Always do a final load for a BLKmode value not larger than a word.
36380 2017-01-10 David Malcolm <dmalcolm@redhat.com>
36383 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
36384 that we correctly handle column numbers greater than
36385 LINE_MAP_MAX_COLUMN_NUMBER.
36387 2017-01-10 Martin Sebor <msebor@redhat.com>
36389 PR middle-end/78245
36390 * gimple-ssa-sprintf.c (get_destination_size): Call
36391 {init,fini}object_sizes.
36392 * tree-object-size.c (addr_object_size): Adjust.
36393 (pass_through_call): Adjust.
36394 (pass_object_sizes::execute): Adjust.
36395 * tree-object-size.h (fini_object_sizes): Declare.
36397 2017-01-10 Martin Sebor <msebor@redhat.com>
36399 PR tree-optimization/78775
36400 * builtins.c (get_size_range): Move...
36401 * calls.c: ...to here.
36402 (alloc_max_size): Accept zero argument.
36403 (operand_signed_p): Remove.
36404 (maybe_warn_alloc_args_overflow): Call get_size_range.
36405 * calls.h (get_size_range): Declare.
36407 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
36409 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
36410 from TI's devices.csv file as of September 2016.
36411 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
36413 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
36415 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
36416 * doc/invoke.texi: Likewise.
36417 * doc/md.texi: Likewise.
36418 * doc/objc.texi: Likewise.
36420 2017-01-10 Joshua Conner <joshconner@google.com>
36422 * config/arm/fuchsia-elf.h: New file.
36423 * config/fuchsia.h: New file.
36424 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
36425 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
36427 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
36429 2016-01-10 Richard Biener <rguenther@suse.de>
36431 PR tree-optimization/79034
36432 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
36433 Propagate out degenerate PHIs in the joiner.
36435 2017-01-10 Martin Liska <mliska@suse.cz>
36437 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
36438 (sort_congruence_classes_by_decl_uid): Likewise.
36439 (sort_congruence_class_groups_by_decl_uid): Likewise.
36440 (sem_item_optimizer::merge_classes): Sort class, groups in these
36441 classes and members in the groups by DECL_UID of declarations.
36442 This would make merge operations stable.
36444 2017-01-10 Martin Liska <mliska@suse.cz>
36446 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
36447 usage of m_classes_vec.
36448 (sem_item_optimizer::~sem_item_optimizer): Likewise.
36449 (sem_item_optimizer::get_group_by_hash): Likewise.
36450 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
36451 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
36452 (sem_item_optimizer::verify_classes): Likewise.
36453 (sem_item_optimizer::process_cong_reduction): Likewise.
36454 (sem_item_optimizer::dump_cong_classes): Likewise.
36455 (sem_item_optimizer::merge_classes): Likewise.
36456 * ipa-icf.h (congruence_class_hash): Rename from
36457 congruence_class_group_hash. Remove declaration of m_classes_vec.
36459 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
36461 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
36462 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
36463 * config.gcc: Add avx512vpopcntdqintrin.h.
36464 * config/i386/avx512vpopcntdqintrin.h: New.
36465 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
36466 * config/i386/i386-builtin-types.def: Add new types.
36467 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
36468 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
36469 __builtin_ia32_vpopcountq_v8di_mask): New.
36470 * config/i386/i386-c.c (ix86_target_macros_internal): Define
36471 __AVX512VPOPCNTDQ__.
36472 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
36473 (PTA_AVX512VPOPCNTDQ): Define.
36474 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
36475 TARGET_AVX512VPOPCNTDQ_P): Define.
36476 * config/i386/i386.opt: Add mavx512vpopcntdq.
36477 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
36478 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
36480 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36482 PR middle-end/77484
36483 * predict.def (PRED_CALL): Set to 67.
36485 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
36487 * expr.c (store_field): In the bitfield case, if the value comes from
36488 a function call and is of an aggregate type returned in registers, do
36489 not modify the field mode; extract the value in all cases if the mode
36490 is BLKmode and the size is not larger than a word.
36492 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
36495 * config/i386/cpuid.h: Fix undefined behavior.
36497 2017-01-04 Jeff Law <law@redhat.com>
36499 PR tree-optimization/79007
36500 PR tree-optimization/67955
36501 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
36502 conservative for pt.null when flag_non_call_exceptions is on.
36504 2017-01-09 Jakub Jelinek <jakub@redhat.com>
36506 PR translation/79019
36507 PR translation/79020
36508 * params.def (PARAM_INLINE_MIN_SPEEDUP,
36509 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
36510 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
36512 * config/avr/avr.opt (maccumulate-args): Likewise.
36513 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
36514 * common.opt (freport-bug): Likewise.
36515 * cif-code.def (CIF_FINAL_ERROR): Likewise.
36516 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
36517 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
36518 translatable string.
36519 * config/i386/i386.c (function_value_32): Likewise.
36520 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
36521 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
36523 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
36524 * common/config/msp430/msp430-common.c (msp430_handle_option):
36526 * symtab.c (symtab_node::verify_base): Likewise.
36527 * opts.c (set_debug_level): Likewise.
36528 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
36529 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
36530 missing whitespace to translatable strings.
36531 * config/avr/avr.md (bswapsi2): Fix typo in comment.
36532 * config/sh/superh.h: Likewise.
36533 * config/i386/xopintrin.h: Likewise.
36534 * config/i386/znver1.md: Likewise.
36535 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
36536 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
36537 * double-int.h (struct double_int): Likewise.
36538 * double-int.c (div_and_round_double): Likewise.
36539 * wide-int.cc: Likewise.
36540 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
36541 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
36542 * cfgcleanup.c (crossjumps_occured): Renamed to ...
36543 (crossjumps_occurred): ... this.
36544 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
36547 PR tree-optimization/78899
36548 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
36549 returning bool return struct loop *, NULL for failure and the new
36551 (versionable_outer_loop_p): Don't version outer loop if it has
36552 dont_vectorized bit set.
36553 (tree_if_conversion): When versioning outer loop, ensure
36554 tree_if_conversion is performed also on the inner loop of the
36555 non-vectorizable outer loop copy.
36556 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
36557 LOOP_VECTORIZED in inner loop of the scalar outer loop and
36558 prevent vectorization of it.
36559 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
36560 the outer loop vectorization of the non-scalar version is attempted
36561 before vectorization of the inner loop in scalar version. If
36562 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
36563 vectorization of its inner loop.
36564 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
36565 has 2 inner loops, rename also on edges from bb whose single pred
36566 is outer_loop->header. Fix typo in function comment.
36568 2017-01-09 Martin Sebor <msebor@redhat.com>
36571 * asan.c (asan_emit_stack_protection): Increase local buffer size
36572 to avoid snprintf truncation warning.
36574 2017-01-09 Andrew Pinski <apinski@cavium.com>
36576 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
36577 to reference thunderx2t99 for the tuning structure
36578 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
36580 (thunderx2t99_extra_costs): This.
36581 * config/aarch64/aarch64-tune.md: Regenerate.
36582 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
36583 (vulcan_addrcost_table): This.
36584 (vulcan_regmove_cost): Rename to ...
36585 (thunderx2t99_regmove_cost): This.
36586 (vulcan_vector_cost): Rename to ...
36587 (thunderx2t99_vector_cost): this.
36588 (vulcan_branch_cost): Rename to ...
36589 (thunderx2t99_branch_cost): This.
36590 (vulcan_tunings): Rename to ...
36591 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
36592 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
36594 2017-01-09 Martin Jambor <mjambor@suse.cz>
36598 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
36599 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
36600 (propagate_vr_accross_jump_function): Use the above function for all
36601 value range computations for pass-through jump functions and type
36602 converasion from explicit value range values.
36603 (ipcp_propagate_stage): Do not attempt to deduce types of formal
36604 parameters from TYPE_ARG_TYPES.
36605 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
36606 (ipa_write_node_info): Stream type of the actual argument.
36607 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
36609 2017-01-09 Martin Liska <mliska@suse.cz>
36612 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
36613 (lookup_compiler): Do not show error message with have_E.
36615 2017-01-09 Jakub Jelinek <jakub@redhat.com>
36617 PR tree-optimization/78938
36618 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
36619 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
36620 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
36621 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
36624 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36626 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
36629 2017-01-09 Richard Biener <rguenther@suse.de>
36631 PR tree-optimization/78997
36632 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
36633 name condition properly.
36635 2017-01-09 Richard Biener <rguenther@suse.de>
36638 * dwarf2out.c (is_cxx): New overload with context.
36639 (is_naming_typedef_decl): Use it.
36641 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
36643 * invoke.texi (Option Summary): Correct spacing in option lists
36644 and add line breaks to fix over-long lines.
36646 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
36648 PR middle-end/17660
36650 * extend.texi (Common Variable Attributes): Add xref to GCC
36651 Internals manual to explain mode attribute keywords.
36653 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
36656 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
36657 and Preprocessor Options.
36658 (Options for Linking): Document -pthread here....
36659 (RS/6000 and PowerPC Options): ...not here.
36660 (Solaris 2 Options): ...or here.
36661 * doc/cppopts.texi: Document -pthread.
36663 2017-01-08 Martin Sebor <msebor@redhat.com>
36665 PR middle-end/77708
36666 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
36667 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
36668 New member functions.
36669 (format_directive): Used them.
36671 (pass_sprintf_length::handle_gimple_call): Same.
36672 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
36673 to avoid truncation for any argument.
36674 (extract_affine_mul): Same.
36675 * tree.c (get_file_function_name): Same.
36677 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36679 PR middle-end/77484
36680 * predict.def (PRED_INDIR_CALL): Set to 86.
36682 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
36684 PR preprocessor/54124
36685 * doc/cppopts.texi: Reformat -d subtable to list the full name
36686 of the options. Add cross-reference to the docs for the general
36687 compiler -d options.
36688 * doc/invoke.texi (Developer Options): Add cross-reference to the
36689 preprocessor-specific -d option documentation.
36691 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
36693 PR preprocessor/13498
36694 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
36695 redudant material, and reflect new command-line options.
36696 (System Headers): Likewise.
36698 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
36700 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
36701 -isystem, and -idirafter. Copy-edit.
36702 * doc/cppopts.texi: Copy-edit. Remove contradiction about
36703 default for -ftrack-macro-expansion. Delete obsolete and
36704 badly-formatted implementation details about -fdebug-cpp output.
36705 * doc/cppwarnopts.texi: Copy-edit.
36707 2017-01-07 David Malcolm <dmalcolm@redhat.com>
36710 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
36711 that the transition from a max line width >= 1<<10 to narrower
36712 lines works correctly.
36714 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
36716 * doc/options.texi (PerFunction): New.
36717 * opt-functions.awk (switch_flags): Map both Optimization and
36718 PerFunction to CL_OPTIMIZATION.
36719 * opth-gen.awk: Test for PerFunction flag along with
36721 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
36722 it only when the latter is present. Skip those that don't in
36723 the hash function generator.
36724 * common.opt (fvar-tracking): Mark as PerFunction instead of
36726 (fvar-tracking-assignments): Likewise.
36727 (fvar-tracking-assignments-toggle): Likewise.
36728 (fvar-tracking-uninit): Likewise.
36730 2017-01-07 Jakub Jelinek <jakub@redhat.com>
36732 PR translation/79018
36733 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
36736 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
36739 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
36740 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
36741 TARGET_LONG_JUMP_TABLE_OFFSETS.
36742 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
36743 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
36744 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
36745 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
36746 * config/m68k/m68k.md (tablejump expander): Likewise.
36747 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
36748 TARGET_LONG_JUMP_TABLE_OFFSETS.
36749 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
36750 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
36752 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
36753 David Holsgrove <david.holsgrove@xilinx.com>
36755 * common/config/microblaze/microblaze-common.c
36756 (TARGET_EXCEPT_UNWIND_INFO): Remove.
36757 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
36759 * config/microblaze/microblaze.c (microblaze_must_save_register)
36760 (microblaze_expand_epilogue, microblaze_return_addr): Handle
36762 (microblaze_eh_return): New function.
36763 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
36764 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
36765 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
36766 * config/microblaze/microblaze.md (eh_return): New pattern.
36768 2017-01-06 Jakub Jelinek <jakub@redhat.com>
36770 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
36771 GCC_DIAGNOSTIC_STRINGIFY): Define.
36773 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
36775 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36777 * config/arm/arm.md (<mcrr>): New.
36779 * config/arm/arm.c (arm_arch5te): New.
36780 (arm_option_override): Set arm_arch5te.
36781 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
36783 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
36784 (arm_mcrr_qualifiers): ... this. New.
36785 (MRRC_QUALIFIERS): Define to...
36786 (arm_mrrc_qualifiers): ... this. New.
36787 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
36789 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
36790 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
36791 (MRRCI, mrrc, MRRC): New.
36792 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
36793 VUNSPEC_MRRC2): New.
36795 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36797 * config/arm/arm.md (<mcr>): New.
36799 * config/arm/arm.c (arm_coproc_builtin_available): Add
36800 support for mcr, mrc, mcr2 and mrc2.
36801 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
36802 (arm_mcr_qualifiers): ... this. New.
36803 (MRC_QUALIFIERS): Define to ...
36804 (arm_mrc_qualifiers): ... this. New.
36805 (MCR_QUALIFIERS): Define to ...
36806 (arm_mcr_qualifiers): ... this. New.
36807 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
36809 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
36810 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
36811 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
36812 VUNSPEC_MRC2): New.
36814 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36816 * config/arm/arm.md (*ldc): New.
36820 * config/arm/arm.c (arm_coproc_builtin_available): Add
36821 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
36822 (arm_coproc_ldc_stc_legitimate_address): New.
36823 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
36824 'qualifier_const_pointer'.
36825 (LDC_QUALIFIERS): Define to...
36826 (arm_ldc_qualifiers): ... this. New.
36827 (STC_QUALIFIERS): Define to...
36828 (arm_stc_qualifiers): ... this. New.
36829 * config/arm/arm-protos.h
36830 (arm_coproc_ldc_stc_legitimate_address): New.
36831 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
36832 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
36833 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
36834 stc2, stcl, stc2l): New.
36835 * config/arm/constraints.md (Uz): New.
36836 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
36837 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
36838 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
36839 VUNSPEC_STC2L): New.
36841 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36843 * config/arm/arm.md (<cdp>): New.
36844 * config/arm/arm.c (neon_const_bounds): Rename this ...
36845 (arm_const_bounds): ... this.
36846 (arm_coproc_builtin_available): New.
36847 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
36848 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
36849 (CDP_QUALIFIERS): Define to...
36850 (arm_cdp_qualifiers): ... this. New.
36852 (arm_expand_builtin_args): Add case for 6 arguments.
36853 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
36854 (arm_const_bounds): ... this.
36855 (arm_coproc_builtin_available): New.
36856 * config/arm/arm_acle.h (__arm_cdp): New.
36858 * config/arm/arm_acle_builtins.def (cdp): New.
36860 * config/arm/iterators.md (CDPI,CDP,cdp): New.
36861 * config/arm/neon.md: Rename all 'neon_const_bounds' to
36862 'arm_const_bounds'.
36863 * config/arm/types.md (coproc): New.
36864 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
36865 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
36866 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
36867 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
36869 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36871 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
36872 (UBINOP_QUALIFIERS): New.
36874 (acle_builtin_data): New. Change comment.
36875 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
36876 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
36877 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
36878 arm_acle_builtins.def.
36879 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
36880 (arm_init_acle_builtins): New.
36881 (CRC32_BUILTIN): Remove.
36882 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
36883 crc32cb, crc32ch and crc32cw.
36884 (arm_init_crc32_builtins): Remove.
36885 (arm_init_builtins): Use arm_init_acle_builtins rather
36886 than arm_init_crc32_builtins.
36887 (arm_expand_acle_builtin): New.
36888 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
36889 * config/arm/arm_acle_builtins.def: New.
36891 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
36893 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
36894 (arm_builtin_datum): ... this.
36895 (arm_init_neon_builtin): Rename to ...
36896 (arm_init_builtin): ... this. Add a new parameters PREFIX
36897 and USE_SIG_IN_NAME.
36898 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
36899 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
36900 'arm_builtin_datum'.
36901 (arm_init_vfp_builtins): Likewise.
36902 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
36903 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
36904 (arm_expand_neon_args): Rename to ...
36905 (arm_expand_builtin_args): ... this. Rename builtin_arg
36906 enum values and differentiate between ARG_BUILTIN_MEMORY
36907 and ARG_BUILTIN_NEON_MEMORY.
36908 (arm_expand_neon_builtin_1): Rename to ...
36909 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
36910 values, arm_expand_builtin_args and add bool parameter NEON.
36911 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
36912 (arm_expand_vfp_builtin): Likewise.
36913 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
36915 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
36917 PR middle-end/77484
36918 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
36919 * predict.c (tree_estimate_probability_bb): Reverse direction of
36920 polymorphic call predictor.
36922 2017-01-06 David Malcolm <dmalcolm@redhat.com>
36924 * passes.c (execute_one_pass): Split out pass-skipping logic into...
36925 (determine_pass_name_match): ...this new function and...
36926 (should_skip_pass_p): ...this new function.
36928 2017-01-06 Nathan Sidwell <nathan@acm.org>
36930 * ipa-visibility.c (function_and_variable_visibility): Reformat
36931 comments and long lines. Remove extrneous if.
36932 * symtab.c (symtab_node::make_decl_local): Fix code format.
36933 (symtab_node::set_section_for_node): Fix comment typo.
36935 2017-01-06 Martin Liska <mliska@suse.cz>
36938 * lra-constraints.c: Rename invariant to lra_invariant.
36939 * predict.c (set_even_probabilities): Initialize e to NULL.
36941 2017-01-05 Martin Sebor <msebor@redhat.com>
36943 PR tree-optimization/78910
36944 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
36945 (format_integer): Correct off-by-one error in the handling
36946 of precision with negative numbers in signed conversions..
36948 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
36950 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
36952 2017-01-05 Jakub Jelinek <jakub@redhat.com>
36954 PR tree-optimization/71016
36955 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
36956 factor_out_conditional_conversion. Formatting fix.
36957 (factor_out_conditional_conversion): Add cond_stmt argument.
36958 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
36959 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
36962 2017-01-05 David Malcolm <dmalcolm@redhat.com>
36964 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
36965 read-rtl-function.o, and selftest-rtl.o.
36966 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
36967 (selftest::aarch64_test_loading_full_dump): New function.
36968 (selftest::aarch64_run_selftests): New function.
36969 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
36970 selftest::aarch64_run_selftests.
36971 * config/i386/i386.c
36972 (selftest::ix86_test_loading_dump_fragment_1): New function.
36973 (selftest::ix86_test_loading_call_insn): New function.
36974 (selftest::ix86_test_loading_full_dump): New function.
36975 (selftest::ix86_test_loading_unspec): New function.
36976 (selftest::ix86_run_selftests): Call the new functions.
36977 * emit-rtl.c (maybe_set_max_label_num): New function.
36978 * emit-rtl.h (maybe_set_max_label_num): New decl.
36979 * function.c (instantiate_decls): Guard call to
36980 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
36981 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
36983 * gensupport.c (gen_reader::gen_reader): Pass "false"
36984 for new "compact" param of rtx_reader.
36985 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
36986 rather than an empty string for NULL strings.
36987 * read-md.c: Potentially include config.h rather than bconfig.h.
36988 Wrap include of errors.h with #ifdef GENERATOR_FILE.
36989 (have_error): New global, copied from errors.c.
36990 (md_reader::read_name): Rename to...
36991 (md_reader::read_name_1): ...this, adding "out_loc" param,
36992 and converting "missing name or number" to returning false, rather
36994 (md_reader::read_name): Reimplement in terms of read_name_1.
36995 (md_reader::read_name_or_nil): New function.
36996 (md_reader::read_string): Handle "(nil)" by returning NULL.
36997 (md_reader::md_reader): Add new param "compact".
36998 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
36999 (md_reader::read_file): New method.
37000 * read-md.h (md_reader::md_reader): Add new param "compact".
37001 (md_reader::read_file): New method.
37002 (md_reader::is_compact): New accessor.
37003 (md_reader::read_name): Convert return type from void to file_location.
37004 (md_reader::read_name_or_nil): New decl.
37005 (md_reader::read_name_1): New decl.
37006 (md_reader::m_compact): New field.
37007 (noop_reader::noop_reader): Pass "false" for new "compact" param
37009 (rtx_reader::rtx_reader): Add new "compact" param.
37010 (rtx_reader::read_rtx_operand): Make virtual and convert return
37011 type from void to rtx.
37012 (rtx_reader::read_until): New decl.
37013 (rtx_reader::handle_any_trailing_information): New virtual function.
37014 (rtx_reader::postprocess): New virtual function.
37015 (rtx_reader::finalize_string): New virtual function.
37016 (rtx_reader::m_in_call_function_usage): New field.
37017 (rtx_reader::m_reuse_rtx_by_id): New field.
37018 * read-rtl-function.c: New file.
37019 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
37020 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
37021 (selftest::verify_three_block_rtl_cfg): New decl.
37022 * read-rtl-function.h: New file.
37023 * read-rtl.c: Potentially include config.h rather than bconfig.h.
37024 For host, include function.h, memmodel.h, and emit-rtl.h.
37025 (one_time_initialization): New function.
37026 (struct compact_insn_name): New struct.
37027 (compact_insn_names): New array.
37028 (find_code): Handle insn codes in compact dumps.
37029 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
37030 (bind_subst_iter_and_attr): Likewise.
37031 (add_condition_to_string): Likewise.
37032 (add_condition_to_rtx): Likewise.
37033 (apply_attribute_uses): Likewise.
37034 (add_current_iterators): Likewise.
37035 (apply_iterators): Likewise.
37036 (initialize_iterators): Guard usage of apply_subst_iterator with
37037 #ifdef GENERATOR_FILE.
37038 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
37039 (md_reader::read_mapping): Likewise.
37040 (add_define_attr_for_define_subst): Likewise.
37041 (add_define_subst_attr): Likewise.
37042 (read_subst_mapping): Likewise.
37043 (check_code_iterator): Likewise.
37044 (rtx_reader::read_rtx): Likewise. Move one-time initialization
37046 (one_time_initialization): New function.
37047 (rtx_reader::read_until): New method.
37048 (read_flags): New function.
37049 (parse_reg_note_name): New function.
37050 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
37051 Handle reuse_rtx ids.
37052 Wrap iterator lookup within #ifdef GENERATOR_FILE.
37053 Add parsing support for RTL dumps, mirroring the special-cases in
37054 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
37055 values, and calling handle_any_trailing_information.
37056 (rtx_reader::read_rtx_operand): Convert return type from void
37057 to rtx, returning return_rtx. Handle case 'e'. Call
37058 finalize_string on XSTR and XTMPL fields.
37059 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
37060 "(nil)" values were omitted. Call the postprocess vfunc on the
37062 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
37063 class ctor. Initialize m_in_call_function_usage. Call
37064 one_time_initialization.
37065 * rtl-tests.c (selftest::test_uncond_jump): Call
37066 set_new_first_and_last_insn.
37067 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
37068 * selftest-rtl.c: New file.
37069 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
37070 (selftest::get_insn_by_uid): New decl.
37071 * selftest-run-tests.c (selftest::run_tests): Call
37072 read_rtl_function_c_tests.
37073 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
37074 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
37077 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
37079 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
37080 operands in a special way. Assert that pos+len <= mode precision.
37082 2017-01-05 Jakub Jelinek <jakub@redhat.com>
37084 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
37085 3 argument Alias with unlimited for the negative form.
37086 (fno-vect-cost-model): Removed.
37088 2017-01-05 Martin Liska <mliska@suse.cz>
37090 * hsa-gen.c (gen_hsa_divmod): New function.
37091 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
37093 2017-01-05 Martin Liska <mliska@suse.cz>
37096 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
37099 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37101 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
37102 small constant length operands.
37104 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37106 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
37107 between loop iterations.
37109 2017-01-05 Martin Liska <mliska@suse.cz>
37112 * gimplify.c (gimplify_decl_expr): Compare to
37113 asan_poisoned_variables instread of checking flags.
37114 (gimplify_target_expr): Likewise.
37115 (gimplify_expr): Likewise.
37116 (gimplify_function_tree): Conditionally initialize
37117 asan_poisoned_variables.
37119 2017-01-04 Jeff Law <law@redhat.com>
37121 PR tree-optimizatin/78812
37122 * rtl.h (contains_mem_rtx_p): Prototype.
37123 * ifcvt.c (containts_mem_rtx_p): Move from here to...
37124 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
37125 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
37126 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
37127 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
37129 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37131 * input.c (assert_char_at_range): Default-initialize actual_range.
37133 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37135 * df-scan.c (df_ref_create_structure): Make regno unsigned,
37136 to match the caller.
37138 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37140 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
37141 insns after final jump in test to emit dummy move.
37143 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37145 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
37146 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
37148 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37150 * multiple_target.c (create_dispatcher_calls): Init e_next.
37151 * tree-ssa-loop-split.c (split_loop): Init border.
37152 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
37155 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
37160 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
37161 (altivec_register_operand): Do not return true if the operand
37162 contains a SUBREG mixing SImode and SFmode.
37163 (vsx_register_operand): Likewise.
37164 (vsx_reg_sfsubreg_ok): New predicate.
37165 (vfloat_operand): Do not return true if the operand contains a
37166 SUBREG mixing SImode and SFmode.
37167 (vint_operand): Likewise.
37168 (vlogical_operand): Likewise.
37169 (gpc_reg_operand): Likewise.
37170 (int_reg_operand): Likewise.
37171 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
37172 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
37173 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
37175 (rs6000_emit_move_si_sf_subreg): New helper function.
37176 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
37177 fixup SUBREGs involving SImode and SFmode.
37178 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
37179 numbers for the new peephole2 optimization.
37180 (peephole2 for SFmode unions): New peephole2 to optimize cases in
37181 the GLIBC math library that do AND/IOR/XOR operations on single
37182 precision floating point.
37183 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
37184 target macros to say whether we need to avoid SUBREGs mixing
37186 (TARGET_ALLOW_SF_SUBREG): Likewise.
37187 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
37188 (UNSPEC_SI_FROM_SF): Likewise.
37189 (iorxor): Change spacing.
37190 (and_ior_xor): New iterator for AND, IOR, and XOR.
37191 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
37192 (movdi_from_sf_zero_ext): Likewise.
37193 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
37194 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
37195 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
37196 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
37197 (fms<mode>4): Likewise.
37198 (fnma<mode>4): Likewise.
37199 (fnms<mode>4): Likewise.
37200 (nfma<mode>4): Likewise.
37201 (nfms<mode>4): Likewise.
37203 2017-01-04 Marek Polacek <polacek@redhat.com>
37206 * doc/invoke.texi: Document -Wpointer-compare.
37208 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37210 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
37213 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
37214 descriptions for -gdwarf-5 and emit them as uleb128 instead of
37217 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
37220 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
37221 documentation of the powerpc_popcntb_ok attribute.
37222 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
37223 code to issue warning messages if a requested CPU configuration is
37224 not supported by the binary (assembler and loader) toolchain.
37225 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
37226 made to define a built-in function that has been disabled.
37227 (paired_init_builtins): Add assertion to prevent ICE if attempt is
37228 made to define a built-in function that has been disabled.
37229 (altivec_init_builtins): Add comment explaining why definition
37230 of the DST built-in functions is not preceded by an assertion
37231 check. Add assertions to prevent ICE if attempts are made to
37232 define an altivec predicate or an abs* built-in function that has
37234 (htm_init_builtins): Add comment explaining why definition of the
37235 htm built-in functions is not preceded by an assertion check.
37237 2017-01-04 Jeff Law <law@redhat.com>
37239 PR tree-optimizatin/67955
37240 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
37241 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
37242 the points-to solution does not include pt_null. Use DECL_PT_UID
37245 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
37247 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
37248 Use gen_int_mode instead of gen_lopwart for const_int operands.
37250 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37252 PR tree-optimization/71563
37253 * match.pd: Simplify X << Y into X if Y is known to be 0 or
37254 out of range value - has low bits known to be zero.
37256 2017-01-04 Alan Modra <amodra@gmail.com>
37258 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
37259 * configure: Regenerate.
37260 * config.in: Regenerate.
37262 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37265 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
37266 a substring of the message, but strcmp with the whole message. Ifdef
37267 ENABLE_NLS, translate the message first using dgettext.
37269 2017-01-03 Jeff Law <law@redhat.com>
37271 PR tree-optimizatin/78856
37272 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
37273 (mark_threaded_blocks): Remove code to truncate thread paths that
37274 cross multiple loop headers. Instead invalidate the cached loop
37275 iteration information and handle case of a thread path walking
37276 into an irreducible region.
37278 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
37281 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
37282 assertions. Add support for doing the signbit if the IEEE 128-bit
37283 floating point value is in a GPR.
37284 * config/rs6000/rs6000.md (Fsignbit): Delete.
37285 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
37286 Update the length attribute if the value is in a GPR.
37287 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
37288 the sign or zero extension instruction, since the value is always 0/1.
37289 (signbit<mode>2_dm2): Delete using <Fsignbit>.
37292 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
37293 extracting SImode to a GPR register so that we can generate a
37294 store, limit the vector to be in a traditional Altivec register
37295 for the vextuwrx instruction.
37297 2017-01-03 Ian Lance Taylor <iant@google.com>
37299 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
37301 2017-01-03 Martin Sebor <msebor@redhat.com>
37303 PR tree-optimization/78696
37304 * gimple-ssa-sprintf.c (format_floating): Correct handling of
37305 precision. Use MPFR for %f for greater fidelity. Correct handling
37307 (pass_sprintf_length::compute_format_length): Set width and precision
37308 specified by asrerisk to void_node for vararg functions.
37309 (try_substitute_return_value): Adjust dump output.
37311 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
37313 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
37315 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
37317 * doc/invoke.texi (SPARC options): Document -mlra as the default.
37318 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
37319 -mlra/-mno-lra was passed to the compiler.
37321 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
37323 PR rtl-optimization/65618
37324 * emit-rtl.c (try_split): Move initialization of "before" and
37325 "after" to just before the call to emit_insn_after_setloc.
37327 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
37329 * doc/md.texi (Standard Names): Remove reference to Java frontend.
37331 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
37333 * dwarf2out.c (gen_enumeration_type_die): When
37334 -gno-strict-dwarf, add a DW_AT_encoding attribute.
37336 2017-01-03 Jakub Jelinek <jakub@redhat.com>
37338 PR tree-optimization/78965
37339 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
37340 Change first argument from const call_info & to call_info &. For %n
37341 set info.nowrite to false.
37343 PR middle-end/78901
37344 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
37345 possibly throwing calls.
37347 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
37348 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
37349 and fns handling, rather than in a separate case SSA_NAME.
37351 2017-01-02 Jeff Law <law@redhat.com>
37353 * config/darwin-driver.c (darwin_driver_init): Const-correctness
37354 fixes for first_period and second_period variables.
37356 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
37359 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
37360 (*insvqi_1): New insn pattern.
37361 (*insvqi_1_mem_rex64): Ditto.
37362 (*insvqi_2): Ditto.
37363 (*insvqi_3): Rename from *insvqi.
37365 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
37367 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
37369 * doc/cfg.texi (Edges): Remove reference to Java.
37370 (Maintaining the CFG): Ditto.
37372 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37374 PR middle-end/77674
37375 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
37376 transparent aliases.
37378 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37380 PR middle-end/77484
37381 * predict.def (PRED_CALL): Update hitrate.
37382 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
37383 * predict.c (tree_estimate_probability_bb): Split CALL predictor
37384 into direct/indirect/polymorphic variants.
37386 2017-01-01 Jakub Jelinek <jakub@redhat.com>
37388 Update copyright years.
37390 * gcc.c (process_command): Update copyright notice dates.
37391 * gcov-dump.c (print_version): Ditto.
37392 * gcov.c (print_version): Ditto.
37393 * gcov-tool.c (print_version): Ditto.
37394 * gengtype.c (create_file): Ditto.
37395 * doc/cpp.texi: Bump @copying's copyright year.
37396 * doc/cppinternals.texi: Ditto.
37397 * doc/gcc.texi: Ditto.
37398 * doc/gccint.texi: Ditto.
37399 * doc/gcov.texi: Ditto.
37400 * doc/install.texi: Ditto.
37401 * doc/invoke.texi: Ditto.
37403 Copyright (C) 2017 Free Software Foundation, Inc.
37405 Copying and distribution of this file, with or without modification,
37406 are permitted in any medium without royalty provided the copyright
37407 notice and this notice are preserved.