1 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
3 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
4 (arc_ccfsm_advance): Fix checking for delay slots.
5 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
7 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9 * config/arc/arc.md (movqi_insn): Add stores to save constant long
11 (movhi_insn): Update store instruction constraint which are saving
12 6-bit short immediates.
13 (movsi_insn): Consider also short scaled load operations.
14 (zero_extendhisi2_i): Use Usd constraint instead of T.
15 (extendhisi2_i): Add q constraint.
16 (arc_clzsi2): Add type and length attributes.
17 (arc_ctzsi2): Likewise.
18 * config/arc/constraints.md (Usc): Update constraint, the
19 assembler can parse two relocations for a single instruction.
21 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
23 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
24 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
26 2017-08-31 Olivier Hainque <hainque@adacore.com>
28 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
29 match as powerpc-wrs-vxworks*.
31 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
33 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
34 register constraint for by-element operand.
35 (aarch64_mls_elt_merge<mode>): Likewise.
37 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
39 * config/arc/arc.c (arc_can_follow_jump): Check for short
42 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
44 * config.gcc: Use g.opt for arc.
45 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
46 functionality moved to ...
47 (legitimate_scaled_address_p): New function, ...here.
48 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
49 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
50 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
52 (arc_override_options): Handle G option.
53 (arc_output_pic_addr_const): Correct function definition.
54 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
55 (arc_decl_anon_ns_mem_p): Delete.
56 (arc_in_small_data_p): Overhaul this function to take into
57 consideration the value given via G option.
58 (arc_rewrite_small_data_1): Renamed and corrected old
59 arc_rewrite_small_data function.
60 (arc_rewrite_small_data): New function.
61 (small_data_pattern): Don't use pic_offset_table_rtx.
62 * config/arc/arc.h (CC1_SPEC): Recognize G option.
63 * config/arc/simdext.md (movmisalignv2hi): Use
64 prepare_move_operands function.
66 (movmisalign*): Likewise.
67 * doc/invoke.texi (ARC options): Document -G option.
69 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
71 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
73 * config/arc/arc.c (arc_print_operand): Output scalled address for
74 sdata whenever is possible.
75 (arc_in_small_data_p): Allow sdata for 64bit datum when double
76 load/stores are available.
77 (compact_sda_memory_operand): Check for the alignment required by
78 code density instructions.
79 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
81 * config/arc/constraints.md (Usd): Update constraint.
82 (Us0): New constraint.
83 (Usc): Update constraint.
85 2017-08-31 Richard Biener <rguenther@suse.de>
88 * dwarf2out.c (dwarf2out_early_global_decl): Process each
91 2017-08-31 Tamar Christina <tamar.christina@arm.com>
93 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
94 Resize type_signature.
96 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
97 Alan Hayward <alan.hayward@arm.com>
98 David Sherwood <david.sherwood@arm.com>
100 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
101 subregs whose inner modes can be stored in GPRs.
102 (aarch64_classify_index): Likewise.
104 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
105 Alan Hayward <alan.hayward@arm.com>
106 David Sherwood <david.sherwood@arm.com>
108 * config/aarch64/iterators.md (V_cmp_result): Rename to...
109 (V_INT_EQUIV): ...this.
110 (v_cmp_result): Rename to...
111 (v_int_equiv): ...this.
112 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
113 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
114 (copysign<mode>3): Likewise.
115 (aarch64_simd_bsl<mode>_internal): Likewise.
116 (aarch64_simd_bsl<mode>): Likewise.
117 (vec_cmp<mode><mode>): Likewise.
118 (vcond<mode><mode>): Likewise.
119 (vcond<v_cmp_mixed><mode>): Likewise.
120 (vcondu<mode><v_cmp_mixed>): Likewise.
121 (aarch64_cm<optab><mode>): Likewise.
122 (aarch64_cmtst<mode>): Likewise.
123 (aarch64_fac<optab><mode>): Likewise.
124 (vec_perm_const<mode>): Likewise.
125 (vcond_mask_<mode><v_cmp_result>): Rename to...
126 (vcond_mask_<mode><v_int_equiv>): ...this.
127 (vec_cmp<mode><v_cmp_result>): Rename to...
128 (vec_cmp<mode><v_int_equiv>): ...this.
130 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
131 Alan Hayward <alan.hayward@arm.com>
132 David Sherwood <david.sherwood@arm.com>
134 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
136 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
137 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
138 (UNSPEC_LD4_DREG): New unspecs.
139 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
140 (aarch64_ld2<mode>_dreg_be): Replace with...
141 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
143 (aarch64_ld3<mode>_dreg_le)
144 (aarch64_ld3<mode>_dreg_be): Replace with...
145 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
147 (aarch64_ld4<mode>_dreg_le)
148 (aarch64_ld4<mode>_dreg_be): Replace with...
149 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
152 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
154 PR tree-optimization/81987
155 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
156 insert an initializer in a location not dominated by the stride
159 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
161 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
162 on the entire header of the finally block in the fallthru case.
164 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
166 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
168 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
170 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
171 rs6000_emit_move_from_cr and call renamed function.
172 (rs6000_emit_prologue): Call renamed functions.
173 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
174 movesi_from_cr, remove volatile CRs.
176 2017-08-30 Jon Beniston <jon@beniston.com>
177 Richard Biener <rguenther@suse.de>
179 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
180 of VECTOR_MODE_P check.
181 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
182 element vector types.
184 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
186 * df.h (df_read_modify_subreg_p): Remove in favor of...
187 * rtl.h (read_modify_subreg_p): ...this new function. Take a
188 const_rtx instead of an rtx.
189 * cprop.c (local_cprop_find_used_regs): Update accordingly.
190 * df-problems.c (df_word_lr_mark_ref): Likewise.
191 * ira-lives.c (mark_pseudo_reg_live): Likewise.
192 (mark_pseudo_reg_dead): Likewise.
193 (mark_ref_dead): Likewise.
194 * reginfo.c (init_subregs_of_mode): Likewise.
195 * sched-deps.c (sched_analyze_1): Likewise.
196 * df-scan.c (df_def_record_1): Likewise.
197 (df_uses_record): Likewise.
198 (df_read_modify_subreg_p): Remove in favor of...
199 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
200 const_rtx instead of an rtx.
202 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
203 Alan Hayward <alan.hayward@arm.com>
204 David Sherwood <david.sherwood@arm.com>
206 * rtl.h (partial_subreg_p): New function.
207 * caller-save.c (save_call_clobbered_regs): Use it.
208 * calls.c (expand_call): Likewise.
209 * combine.c (combinable_i3pat): Likewise.
210 (simplify_set): Likewise.
211 (make_extraction): Likewise.
212 (make_compound_operation_int): Likewise.
213 (gen_lowpart_or_truncate): Likewise.
214 (force_to_mode): Likewise.
215 (make_field_assignment): Likewise.
216 (reg_truncated_to_mode): Likewise.
217 (record_truncated_value): Likewise.
218 (move_deaths): Likewise.
219 * cse.c (record_jump_cond): Likewise.
220 (cse_insn): Likewise.
221 * cselib.c (cselib_lookup_1): Likewise.
222 * expmed.c (extract_bit_field_using_extv): Likewise.
223 * function.c (assign_parm_setup_reg): Likewise.
224 * ifcvt.c (noce_convert_multiple_sets): Likewise.
225 * ira-build.c (create_insn_allocnos): Likewise.
226 * lra-coalesce.c (merge_pseudos): Likewise.
227 * lra-constraints.c (match_reload): Likewise.
228 (simplify_operand_subreg): Likewise.
229 (curr_insn_transform): Likewise.
230 * lra-lives.c (process_bb_lives): Likewise.
231 * lra.c (new_insn_reg): Likewise.
232 (lra_substitute_pseudo): Likewise.
233 * regcprop.c (mode_change_ok): Likewise.
234 (maybe_mode_change): Likewise.
235 (copyprop_hardreg_forward_1): Likewise.
236 * reload.c (push_reload): Likewise.
237 (find_reloads): Likewise.
238 (find_reloads_subreg_address): Likewise.
239 * reload1.c (alter_reg): Likewise.
240 (eliminate_regs_1): Likewise.
241 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
243 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
245 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
246 back to if statements, including unpack.
248 2017-08-30 Martin Liska <mliska@suse.cz>
251 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
253 (func_checker::compare_asm_inputs_outputs): ... this function.
254 (func_checker::compare_gimple_asm): Use the function to compare
256 * ipa-icf-gimple.h: Rename the function.
258 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
259 Alan Hayward <alan.hayward@arm.com>
260 David Sherwood <david.sherwood@arm.com>
262 * coretypes.h (complex_mode): New type.
263 * gdbhooks.py (build_pretty_printer): Handle it.
264 * machmode.h (complex_mode): New class.
265 (complex_mode::includes_p): New function.
266 (is_complex_int_mode): Likewise.
267 (is_complex_float_mode): Likewise.
268 * genmodes.c (get_mode_class): Handle complex mode classes.
269 * function.c (expand_function_end): Use is_complex_int_mode.
271 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
272 Alan Hayward <alan.hayward@arm.com>
273 David Sherwood <david.sherwood@arm.com>
275 * coretypes.h (scalar_mode_pod): New typedef.
276 * gdbhooks.py (build_pretty_printer): Handle it.
277 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
278 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
279 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
280 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
283 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
284 Alan Hayward <alan.hayward@arm.com>
285 David Sherwood <david.sherwood@arm.com>
287 * machmode.h (mode_for_vector): Take a scalar_mode instead
289 * stor-layout.c (mode_for_vector): Likewise.
290 * explow.c (promote_mode): Use as_a <scalar_mode>.
291 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
293 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
294 Alan Hayward <alan.hayward@arm.com>
295 David Sherwood <david.sherwood@arm.com>
297 * target.def (preferred_simd_mode): Take a scalar_mode
298 instead of a machine_mode.
299 * targhooks.h (default_preferred_simd_mode): Likewise.
300 * targhooks.c (default_preferred_simd_mode): Likewise.
301 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
302 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
303 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
304 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
305 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
306 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
307 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
308 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
310 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
311 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
312 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
313 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
314 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
315 * doc/tm.texi: Regenerate.
316 * optabs-query.c (can_vec_mask_load_store_p): Return false for
319 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
320 Alan Hayward <alan.hayward@arm.com>
321 David Sherwood <david.sherwood@arm.com>
323 * target.def (scalar_mode_supported_p): Take a scalar_mode
324 instead of a machine_mode.
325 * targhooks.h (default_scalar_mode_supported_p): Likewise.
326 * targhooks.c (default_scalar_mode_supported_p): Likewise.
327 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
328 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
329 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
330 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
331 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
332 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
333 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
334 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
335 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
336 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
337 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
338 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
340 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
341 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
342 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
343 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
344 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
346 * doc/tm.texi: Regenerate.
348 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
349 Alan Hayward <alan.hayward@arm.com>
350 David Sherwood <david.sherwood@arm.com>
352 * coretypes.h (opt_scalar_mode): New typedef.
353 * gdbhooks.py (build_pretty_printers): Handle it.
354 * machmode.h (mode_iterator::get_2xwider): Add overload for
356 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
358 * expr.c (convert_mode_scalar): Likewise.
359 * omp-low.c (omp_clause_aligned_alignment): Likewise.
360 * optabs.c (expand_float): Likewise.
361 (expand_fix): Likewise.
362 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
364 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
365 Alan Hayward <alan.hayward@arm.com>
366 David Sherwood <david.sherwood@arm.com>
368 * optabs.c (expand_float): Explicitly check for scalars before
369 using a branching expansion.
370 (expand_fix): Likewise.
372 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
373 Alan Hayward <alan.hayward@arm.com>
374 David Sherwood <david.sherwood@arm.com>
376 * expr.c (convert_mode): Split scalar handling out into...
377 (convert_mode_scalar): ...this new function. Treat the modes
380 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
381 Alan Hayward <alan.hayward@arm.com>
382 David Sherwood <david.sherwood@arm.com>
384 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
386 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
388 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
389 Alan Hayward <alan.hayward@arm.com>
390 David Sherwood <david.sherwood@arm.com>
392 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
393 rather than a machine_mode.
394 (fixed_from_string): Likewise.
395 (fixed_convert): Likewise.
396 (fixed_convert_from_int): Likewise.
397 (fixed_convert_from_real): Likewise.
398 (real_convert_from_fixed): Likewise.
399 * fixed-value.c (fixed_from_double_int): Likewise.
400 (fixed_from_string): Likewise.
401 (fixed_convert): Likewise.
402 (fixed_convert_from_int): Likewise.
403 (fixed_convert_from_real): Likewise.
404 (real_convert_from_fixed): Likewise.
405 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
407 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
408 Alan Hayward <alan.hayward@arm.com>
409 David Sherwood <david.sherwood@arm.com>
411 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
412 of separate mode class checks. Do not allow vector modes here.
413 (immed_wide_int_const): Use as_a <scalar_mode>.
414 * explow.c (trunc_int_for_mode): Likewise.
415 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
416 (wi::shwi): Likewise.
417 (wi::min_value): Likewise.
418 (wi::max_value): Likewise.
419 * dwarf2out.c (loc_descriptor): Likewise.
420 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
423 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
424 Alan Hayward <alan.hayward@arm.com>
425 David Sherwood <david.sherwood@arm.com>
427 * tree.h (SCALAR_TYPE_MODE): New macro.
428 * expr.c (expand_expr_addr_expr_1): Use it.
429 (expand_expr_real_2): Likewise.
430 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
431 (fold_convert_const_fixed_from_int): Likewise.
432 (fold_convert_const_fixed_from_real): Likewise.
433 (native_encode_fixed): Likewise
434 (native_encode_complex): Likewise
435 (native_encode_vector): Likewise.
436 (native_interpret_fixed): Likewise.
437 (native_interpret_real): Likewise.
438 (native_interpret_complex): Likewise.
439 (native_interpret_vector): Likewise.
440 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
441 (simd_clone_adjust_argument_types): Likewise.
442 (simd_clone_init_simd_arrays): Likewise.
443 (simd_clone_adjust): Likewise.
444 * stor-layout.c (layout_type): Likewise.
445 * tree.c (build_minus_one_cst): Likewise.
446 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
447 * tree-inline.c (estimate_move_cost): Likewise.
448 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
449 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
450 (vectorizable_reduction): Likewise.
451 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
452 (vect_recog_mixed_size_cond_pattern): Likewise.
453 (check_bool_pattern): Likewise.
454 (adjust_bool_pattern): Likewise.
455 (search_type_for_mask_1): Likewise.
456 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
457 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
458 (vectorizable_load): Likewise.
459 (vectorizable_store): Likewise.
460 * ubsan.c (ubsan_encode_value): Likewise.
461 * varasm.c (output_constant): Likewise.
463 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
464 Alan Hayward <alan.hayward@arm.com>
465 David Sherwood <david.sherwood@arm.com>
467 * coretypes.h (scalar_mode): New class.
468 * machmode.h (scalar_mode): Likewise.
469 (scalar_mode::includes_p): New function.
470 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
471 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
472 * genmodes.c (get_mode_class): Handle remaining scalar modes.
473 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
474 * expmed.c (store_bit_field_1): Likewise.
475 (extract_bit_field_1): Likewise.
476 * expr.c (write_complex_part): Likewise.
477 (read_complex_part): Likewise.
478 (emit_move_complex_push): Likewise.
479 (expand_expr_real_2): Likewise.
480 * function.c (assign_parm_setup_reg): Likewise.
481 (assign_parms_unsplit_complex): Likewise.
482 * optabs.c (expand_binop): Likewise.
483 * rtlanal.c (subreg_get_info): Likewise.
484 * simplify-rtx.c (simplify_immed_subreg): Likewise.
485 * varasm.c (output_constant_pool_2): Likewise.
487 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
488 Alan Hayward <alan.hayward@arm.com>
489 David Sherwood <david.sherwood@arm.com>
491 * expmed.c (extract_high_half): Use scalar_int_mode and remove
493 (expmed_mult_highpart_optab): Likewise.
494 (expmed_mult_highpart): Likewise.
496 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
497 Alan Hayward <alan.hayward@arm.com>
498 David Sherwood <david.sherwood@arm.com>
500 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
501 instead of a machine_mode.
502 (builtin_memset_read_str): Likewise.
503 * builtins.c (c_readstr): Likewise.
504 (builtin_memcpy_read_str): Likewise.
505 (builtin_strncpy_read_str): Likewise.
506 (builtin_memset_read_str): Likewise.
507 (builtin_memset_gen_str): Likewise.
508 (expand_builtin_signbit): Use scalar_int_mode for local variables.
509 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
510 instead of a machine_mode.
511 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
513 (make_extraction): Likewise.
514 (try_widen_shift_mode): Take and return scalar_int_modes instead
516 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
517 a scalar_int_mode instead of a machine_mode.
518 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
519 (avr_addr_space_pointer_mode): Likewise.
520 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
521 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
522 (msp430_unwind_word_mode): Likewise.
523 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
524 (spu_addr_space_pointer_mode): Likewise.
525 (spu_addr_space_address_mode): Likewise.
526 (spu_libgcc_cmp_return_mode): Likewise.
527 (spu_libgcc_shift_count_mode): Likewise.
528 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
529 (rl78_addr_space_pointer_mode): Likewise.
530 (fl78_unwind_word_mode): Likewise.
531 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
533 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
534 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
535 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
536 (mips_valid_pointer_mode): Likewise.
537 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
538 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
539 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
541 (ft32_addr_space_address_mode): Likewise.
542 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
543 scalar_int_mode instead of a machine_mode.
544 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
546 (m32c_addr_space_address_mode): Likewise.
547 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
548 (rs6000_eh_return_filter_mode): Likewise.
549 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
550 (rs6000_eh_return_filter_mode): Likewise.
551 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
552 (s390_libgcc_shift_count_mode): Likewise.
553 (s390_unwind_word_mode): Likewise.
554 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
556 * target.def (mode_rep_extended): Likewise.
557 (valid_pointer_mode): Likewise.
558 (addr_space.valid_pointer_mode): Likewise.
559 (eh_return_filter_mode): Return a scalar_int_mode rather than
561 (libgcc_cmp_return_mode): Likewise.
562 (libgcc_shift_count_mode): Likewise.
563 (unwind_word_mode): Likewise.
564 (addr_space.pointer_mode): Likewise.
565 (addr_space.address_mode): Likewise.
566 * doc/tm.texi: Regenerate.
567 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
569 (do_jump): Use scalar_int_mode for local variables.
570 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
571 rather than a machine_mode.
572 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
573 (scompare_loc_descriptor_wide): Likewise.
574 (scompare_loc_descriptor_narrow): Likewise.
575 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
577 * except.c (sjlj_emit_dispatch_table): Likewise.
578 (expand_builtin_eh_copy_values): Likewise.
579 * explow.c (convert_memory_address_addr_space_1): Likewise.
580 Take a scalar_int_mode rather than a machine_mode.
581 (convert_memory_address_addr_space): Take a scalar_int_mode rather
583 (memory_address_addr_space): Use scalar_int_mode for local variables.
584 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
585 rather than a machine_mode.
586 * expmed.c (mask_rtx): Likewise.
587 (init_expmed_one_conv): Likewise.
588 (expand_mult_highpart_adjust): Likewise.
589 (extract_high_half): Likewise.
590 (expmed_mult_highpart_optab): Likewise.
591 (expmed_mult_highpart): Likewise.
592 (expand_smod_pow2): Likewise.
593 (expand_sdiv_pow2): Likewise.
594 (emit_store_flag_int): Likewise.
595 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
597 (extract_low_bits): Likewise.
598 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
600 * expr.c (pieces_addr::adjust): Likewise.
601 (can_store_by_pieces): Likewise.
602 (store_by_pieces): Likewise.
603 (clear_by_pieces_1): Likewise.
604 (expand_expr_addr_expr_1): Likewise.
605 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
606 (expand_expr_real_1): Likewise.
607 (try_casesi): Likewise.
608 * final.c (shorten_branches): Likewise.
609 * fold-const.c (fold_convert_const_int_from_fixed): Change the
610 type of "mode" to machine_mode.
611 * internal-fn.c (expand_arith_overflow_result_store): Take a
612 scalar_int_mode rather than a machine_mode.
613 (expand_mul_overflow): Use scalar_int_mode for local variables.
614 * loop-doloop.c (doloop_modify): Likewise.
615 (doloop_optimize): Likewise.
616 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
618 (expand_doubleword_shift_condmove): Likewise.
619 (expand_doubleword_shift): Likewise.
620 (expand_doubleword_clz): Likewise.
621 (expand_doubleword_popcount): Likewise.
622 (expand_doubleword_parity): Likewise.
623 (expand_absneg_bit): Use scalar_int_mode for local variables.
624 (prepare_float_lib_cmp): Likewise.
625 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
626 rather than a machine_mode.
627 (convert_memory_address_addr_space): Likewise.
628 (get_mode_bounds): Likewise.
629 (get_address_mode): Return a scalar_int_mode rather than a
631 * rtlanal.c (get_address_mode): Likewise.
632 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
634 * targhooks.c (default_mode_rep_extended): Likewise.
635 (default_valid_pointer_mode): Likewise.
636 (default_addr_space_valid_pointer_mode): Likewise.
637 (default_eh_return_filter_mode): Return a scalar_int_mode rather
639 (default_libgcc_cmp_return_mode): Likewise.
640 (default_libgcc_shift_count_mode): Likewise.
641 (default_unwind_word_mode): Likewise.
642 (default_addr_space_pointer_mode): Likewise.
643 (default_addr_space_address_mode): Likewise.
644 * targhooks.h (default_eh_return_filter_mode): Likewise.
645 (default_libgcc_cmp_return_mode): Likewise.
646 (default_libgcc_shift_count_mode): Likewise.
647 (default_unwind_word_mode): Likewise.
648 (default_addr_space_pointer_mode): Likewise.
649 (default_addr_space_address_mode): Likewise.
650 (default_mode_rep_extended): Take a scalar_int_mode rather than
652 (default_valid_pointer_mode): Likewise.
653 (default_addr_space_valid_pointer_mode): Likewise.
654 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
656 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
657 rather than a machine_mode.
658 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
660 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
661 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
664 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
665 Alan Hayward <alan.hayward@arm.com>
666 David Sherwood <david.sherwood@arm.com>
668 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
669 the mode argument to scalar_int_mode.
670 (do_jump_by_parts_zero_rtx): Likewise.
671 (do_jump_by_parts_equality_rtx): Likewise.
672 (do_jump_by_parts_greater): Take a mode argument.
673 (do_jump_by_parts_equality): Likewise.
674 (do_jump_1): Update calls accordingly.
676 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
677 Alan Hayward <alan.hayward@arm.com>
678 David Sherwood <david.sherwood@arm.com>
680 * is-a.h (safe_dyn_cast): New function.
681 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
682 (jump_table_for_label): Likewise.
683 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
684 instead of an rtx_insn *.
685 (shorten_branches): Use dyn_cast instead of LABEL_P and
686 JUMP_TABLE_DATA_P. Use jump_table_for_label and
687 rtx_jump_table_data::get_data_mode.
688 (final_scan_insn): Likewise.
690 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
691 Alan Hayward <alan.hayward@arm.com>
692 David Sherwood <david.sherwood@arm.com>
694 * combine.c (try_combine): Use is_a <scalar_int_mode> when
695 trying to combine a full-register integer set with a subreg
698 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
699 Alan Hayward <alan.hayward@arm.com>
700 David Sherwood <david.sherwood@arm.com>
702 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
703 that is always either address_mode or pointer_mode.
705 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
706 Alan Hayward <alan.hayward@arm.com>
707 David Sherwood <david.sherwood@arm.com>
709 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
710 when the two are known to be equal.
712 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
713 Alan Hayward <alan.hayward@arm.com>
714 David Sherwood <david.sherwood@arm.com>
716 * simplify-rtx.c (simplify_const_unary_operation): Use
717 is_a <scalar_int_mode> instead of checking for a nonzero
718 precision. Forcibly convert op_mode to a scalar_int_mode
719 in that case. More clearly differentiate the operand and
720 result modes and use the former when deciding what the value
721 of a count-bits operation should be. Use is_int_mode instead
722 of checking for a MODE_INT. Remove redundant check for whether
723 this mode has a zero precision.
725 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
726 Alan Hayward <alan.hayward@arm.com>
727 David Sherwood <david.sherwood@arm.com>
729 * optabs.c (widen_leading): Change the type of the mode argument
730 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
731 (widen_bswap): Likewise.
732 (expand_parity): Likewise.
733 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
734 (expand_ffs): Likewise.
735 (epand_unop): Check for scalar integer modes before calling the
738 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
739 Alan Hayward <alan.hayward@arm.com>
740 David Sherwood <david.sherwood@arm.com>
742 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
744 (expand_expr_real_1): Update call accordingly.
746 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
747 Alan Hayward <alan.hayward@arm.com>
748 David Sherwood <david.sherwood@arm.com>
750 * expmed.c (store_bit_field_using_insv): Add op0_mode and
751 value_mode arguments. Use scalar_int_mode internally.
752 (store_bit_field_1): Rename the new integer mode from imode
753 to op0_mode and use it instead of GET_MODE (op0). Update calls
754 to store_split_bit_field, store_bit_field_using_insv and
755 store_fixed_bit_field.
756 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
757 Use scalar_int_mode internally. Use a bit count rather than a mode
758 when calculating the largest bit size for get_best_mode.
759 Update calls to store_split_bit_field and store_fixed_bit_field_1.
760 (store_fixed_bit_field_1): Add mode and value_mode arguments.
761 Remove assertion that OP0 has a scalar integer mode.
762 (store_split_bit_field): Add op0_mode and value_mode arguments.
763 Update calls to extract_fixed_bit_field.
764 (extract_bit_field_using_extv): Add an op0_mode argument.
765 Use scalar_int_mode internally.
766 (extract_bit_field_1): Rename the new integer mode from imode to
767 op0_mode and use it instead of GET_MODE (op0). Update calls to
768 extract_split_bit_field, extract_bit_field_using_extv and
769 extract_fixed_bit_field.
770 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
771 to extract_split_bit_field and extract_fixed_bit_field_1.
772 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
773 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
775 (extract_split_bit_field): Add an op0_mode argument. Update call
776 to extract_fixed_bit_field.
778 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
779 Alan Hayward <alan.hayward@arm.com>
780 David Sherwood <david.sherwood@arm.com>
782 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
783 * explow.c (hard_function_value): Likewise.
784 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
785 convert_to_mode call outside the loop.
786 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
787 for the mode iterator. Require the mode specified by max_pieces
789 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
791 (copy_blkmode_to_reg): Likewise.
792 (set_storage_via_setmem): Likewise.
793 * optabs.c (prepare_cmp_insn): Likewise.
794 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
795 * stor-layout.c (finish_bitfield_representative): Likewise.
797 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
798 Alan Hayward <alan.hayward@arm.com>
799 David Sherwood <david.sherwood@arm.com>
801 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
802 * expr.c (convert_move): Use them.
803 (convert_modes): Likewise.
804 (store_expr_with_bounds): Likewise.
806 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
807 Alan Hayward <alan.hayward@arm.com>
808 David Sherwood <david.sherwood@arm.com>
810 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
811 parameter for the mode of "x". Remove the "known_x", "known_mode"
812 and "known_ret" arguments. Change the type of the mode argument
814 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
815 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
816 (reg_num_sign_bit_copies_for_combine): Likewise.
817 * rtlanal.c (nonzero_bits1): Likewise.
818 (num_sign_bit_copies1): Likewise.
819 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
820 (reg_num_sign_bit_copies_general): Likewise.
821 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
822 (reg_nonzero_bits_general): Likewise.
824 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
825 Alan Hayward <alan.hayward@arm.com>
826 David Sherwood <david.sherwood@arm.com>
828 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
829 than in subroutines. Return 1 for non-integer modes.
830 (cached_num_sign_bit_copies): Change the type of the mode parameter
832 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
833 classes. Handle CONST_INT_P first and then check whether X also
834 has a scalar integer mode. Check the same thing for inner registers
835 of a SUBREG and for values that are being extended or truncated.
837 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
838 Alan Hayward <alan.hayward@arm.com>
839 David Sherwood <david.sherwood@arm.com>
841 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
842 in subroutines. Return the mode mask for non-integer modes.
843 (cached_nonzero_bits): Change the type of the mode parameter
845 (nonzero_bits1): Likewise. Remove early exit for other mode
846 classes. Handle CONST_INT_P first and then check whether X
847 also has a scalar integer mode.
849 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
850 Alan Hayward <alan.hayward@arm.com>
851 David Sherwood <david.sherwood@arm.com>
853 * expr.c (widest_int_mode_for_size): Make the comment match the code.
854 Return a scalar_int_mode and assert that the size is greater than
856 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
857 (op_by_pieces_d::op_by_pieces_d): Likewise.
858 (op_by_pieces_d::run): Likewise.
859 (can_store_by_pieces): Likewise.
861 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
862 Alan Hayward <alan.hayward@arm.com>
863 David Sherwood <david.sherwood@arm.com>
865 * combine.c (extract_left_shift): Add a mode argument and update
867 (make_compound_operation_int): Change the type of the mode parameter
868 to scalar_int_mode and update the call to extract_left_shift.
870 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
871 Alan Hayward <alan.hayward@arm.com>
872 David Sherwood <david.sherwood@arm.com>
874 * combine.c (simplify_and_const_int): Change the type of the mode
875 parameter to scalar_int_mode.
876 (simplify_and_const_int_1): Likewise. Update recursive call.
878 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
879 Alan Hayward <alan.hayward@arm.com>
880 David Sherwood <david.sherwood@arm.com>
882 * combine.c (simplify_compare_const): Check that the mode is a
883 scalar_int_mode (rather than VOIDmode) before testing its
885 (simplify_comparison): Move COMPARISON_P handling out of the
886 loop and restrict the latter part of the loop to scalar_int_modes.
887 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
888 and when considering SUBREG_REGs. Use is_int_mode instead of
889 checking GET_MODE_CLASS against MODE_INT.
891 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
892 Alan Hayward <alan.hayward@arm.com>
893 David Sherwood <david.sherwood@arm.com>
895 * combine.c (try_widen_shift_mode): Move check for equal modes to...
896 (simplify_shift_const_1): ...here. Use scalar_int_mode for
897 shift_unit_mode and for modes involved in scalar shifts.
899 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
900 Alan Hayward <alan.hayward@arm.com>
901 David Sherwood <david.sherwood@arm.com>
903 * combine.c (force_int_to_mode): New function, split out from...
904 (force_to_mode): ...here. Keep xmode up-to-date and use it
905 instead of GET_MODE (x).
907 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
908 Alan Hayward <alan.hayward@arm.com>
909 David Sherwood <david.sherwood@arm.com>
911 * optabs-query.h (extraction_insn::struct_mode): Change type to
912 opt_scalar_int_mode and update comment.
913 (extraction_insn::field_mode): Change type to scalar_int_mode.
914 (extraction_insn::pos_mode): Likewise.
915 * combine.c (make_extraction): Update accordingly.
916 * optabs-query.c (get_traditional_extraction_insn): Likewise.
917 (get_optab_extraction_insn): Likewise.
918 * recog.c (simplify_while_replacing): Likewise.
919 * expmed.c (narrow_bit_field_mem): Change the type of the mode
920 parameter to opt_scalar_int_mode.
922 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
923 Alan Hayward <alan.hayward@arm.com>
924 David Sherwood <david.sherwood@arm.com>
926 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
927 to a scalar_int_mode instead of a machine_mode.
928 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
929 (get_best_mode): Return a boolean and use a pointer argument to store
930 the selected mode. Replace the limit mode parameter with a bit limit.
931 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
932 for the values returned by bit_field_mode_iterator::next_mode.
933 (store_bit_field): Update call to get_best_mode.
934 (store_fixed_bit_field): Likewise.
935 (extract_fixed_bit_field): Likewise.
936 * expr.c (optimize_bitfield_assignment_op): Likewise.
937 * fold-const.c (optimize_bit_field_compare): Likewise.
938 (fold_truth_andor_1): Likewise.
939 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
940 Update for new type of m_mode.
941 (get_best_mode): As above.
943 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
944 Alan Hayward <alan.hayward@arm.com>
945 David Sherwood <david.sherwood@arm.com>
947 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
948 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
949 (store_bit_field): Check is_a <scalar_int_mode> before calling
950 strict_volatile_bitfield_p.
951 (extract_bit_field): Likewise.
953 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
954 Alan Hayward <alan.hayward@arm.com>
955 David Sherwood <david.sherwood@arm.com>
957 * target.def (cstore_mode): Return a scalar_int_mode.
958 * doc/tm.texi: Regenerate.
959 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
960 * targhooks.h (default_cstore_mode): Likewise.
961 * targhooks.c (default_cstore_mode): Likewise, using a forced
963 * expmed.c (emit_cstore): Expect the target of the cstore to be
966 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
967 Alan Hayward <alan.hayward@arm.com>
968 David Sherwood <david.sherwood@arm.com>
970 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
972 (niter_desc): Likewise mode.
973 (iv_analyze): Add a mode parameter.
975 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
976 and change its type to scalar_int_mode.
977 * loop-iv.c: Update commentary at head of file.
978 (iv_constant): Pass the mode paraeter before the rtx it describes
979 and change its type to scalar_int_mode. Remove VOIDmode handling.
980 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
981 (iv_extend): Likewise.
982 (shorten_into_mode): Likewise.
983 (iv_add): Use scalar_int_mode.
985 (iv_shift): Likewise.
986 (canonicalize_iv_subregs): Likewise.
987 (get_biv_step_1): Pass the outer_mode parameter before the rtx
988 it describes and change its mode to scalar_int_mode. Also change
989 the type of the returned inner_mode to scalar_int_mode.
990 (get_biv_step): Likewise, turning outer_mode from a pointer
991 into a direct parameter. Update call to get_biv_step_1.
992 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
993 iv_constant and get_biv_step.
994 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
995 and change its type to scalar_int_mode. Don't initialise iv->mode
996 to VOIDmode and remove later checks for its still being VOIDmode.
997 Update calls to iv_analyze_op and iv_analyze_expr. Check
998 is_a <scalar_int_mode> when changing the mode under consideration.
999 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
1000 Update call to iv_analyze_expr.
1001 (iv_analyze_op): Add a mode parameter. Reject subregs whose
1002 inner register is not also a scalar_int_mode. Update call to
1004 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
1005 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
1006 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
1007 separate mode class checks. Update calls to iv_analyze. Remove
1008 fix-up of VOIDmodes after iv_analyze_biv.
1009 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
1010 don't have a scalar_int_mode. Update call to biv_p.
1012 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1013 Alan Hayward <alan.hayward@arm.com>
1014 David Sherwood <david.sherwood@arm.com>
1016 * cfgexpand.c (convert_debug_memory_address): Use
1017 as_a <scalar_int_mode>.
1018 * combine.c (expand_compound_operation): Likewise.
1019 (make_extraction): Likewise.
1020 (change_zero_ext): Likewise.
1021 (simplify_comparison): Likewise.
1022 * cse.c (cse_insn): Likewise.
1023 * dwarf2out.c (minmax_loc_descriptor): Likewise.
1024 (mem_loc_descriptor): Likewise.
1025 (loc_descriptor): Likewise.
1026 * expmed.c (init_expmed_one_mode): Likewise.
1027 (synth_mult): Likewise.
1028 (emit_store_flag_1): Likewise.
1029 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
1030 of a comparison with size.
1031 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
1032 (reduce_to_bit_field_precision): Likewise.
1033 * function.c (expand_function_end): Likewise.
1034 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
1035 * loop-doloop.c (doloop_modify): Likewise.
1036 * optabs.c (expand_binop): Likewise.
1037 (expand_unop): Likewise.
1038 (expand_copysign_absneg): Likewise.
1039 (prepare_cmp_insn): Likewise.
1040 (maybe_legitimize_operand): Likewise.
1041 * recog.c (const_scalar_int_operand): Likewise.
1042 * rtlanal.c (get_address_mode): Likewise.
1043 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1044 (simplify_cond_clz_ctz): Likewise.
1045 * tree-nested.c (get_nl_goto_field): Likewise.
1046 * tree.c (build_vector_type_for_mode): Likewise.
1047 * var-tracking.c (use_narrower_mode): Likewise.
1049 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1050 Alan Hayward <alan.hayward@arm.com>
1051 David Sherwood <david.sherwood@arm.com>
1053 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
1054 * builtins.c (expand_builtin_signbit): Use it.
1055 * cfgexpand.c (expand_debug_expr): Likewise.
1056 * dojump.c (do_jump): Likewise.
1057 (do_compare_and_jump): Likewise.
1058 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
1059 * expmed.c (make_tree): Likewise.
1060 * expr.c (expand_expr_real_2): Likewise.
1061 (expand_expr_real_1): Likewise.
1062 (try_casesi): Likewise.
1063 * fold-const-call.c (fold_const_call_ss): Likewise.
1064 * fold-const.c (unextend): Likewise.
1065 (extract_muldiv_1): Likewise.
1066 (fold_single_bit_test): Likewise.
1067 (native_encode_int): Likewise.
1068 (native_encode_string): Likewise.
1069 (native_interpret_int): Likewise.
1070 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
1071 * internal-fn.c (expand_addsub_overflow): Likewise.
1072 (expand_neg_overflow): Likewise.
1073 (expand_mul_overflow): Likewise.
1074 (expand_arith_overflow): Likewise.
1075 * match.pd: Likewise.
1076 * stor-layout.c (layout_type): Likewise.
1077 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
1078 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
1079 * tree-ssanames.c (get_range_info): Likewise.
1080 * tree-switch-conversion.c (array_value_type) Likewise.
1081 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
1082 (vect_recog_divmod_pattern): Likewise.
1083 (vect_recog_mixed_size_cond_pattern): Likewise.
1084 * tree-vrp.c (extract_range_basic): Likewise.
1085 (simplify_float_conversion_using_ranges): Likewise.
1086 * tree.c (int_fits_type_p): Likewise.
1087 * ubsan.c (instrument_bool_enum_load): Likewise.
1088 * varasm.c (mergeable_string_section): Likewise.
1089 (narrowing_initializer_constant_valid_p): Likewise.
1090 (output_constant): Likewise.
1092 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1093 Alan Hayward <alan.hayward@arm.com>
1094 David Sherwood <david.sherwood@arm.com>
1096 * machmode.h (NARROWEST_INT_MODE): New macro.
1097 * expr.c (alignment_for_piecewise_move): Use it instead of
1098 GET_CLASS_NARROWEST_MODE (MODE_INT).
1099 (push_block): Likewise.
1100 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
1102 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1104 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1105 Alan Hayward <alan.hayward@arm.com>
1106 David Sherwood <david.sherwood@arm.com>
1108 * postreload.c (move2add_valid_value_p): Change the type of the
1109 mode parameter to scalar_int_mode.
1110 (move2add_use_add2_insn): Add a mode parameter and use it instead
1112 (move2add_use_add3_insn): Likewise.
1113 (reload_cse_move2add): Update accordingly.
1115 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1116 Alan Hayward <alan.hayward@arm.com>
1117 David Sherwood <david.sherwood@arm.com>
1119 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
1121 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
1122 * optabs.c (expand_unop): Likewise.
1124 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1125 Alan Hayward <alan.hayward@arm.com>
1126 David Sherwood <david.sherwood@arm.com>
1128 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
1129 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
1130 (popcount_loc_descriptor): Likewise.
1131 (bswap_loc_descriptor): Likewise.
1132 (rotate_loc_descriptor): Likewise.
1133 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
1134 calling the functions above.
1136 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1137 Alan Hayward <alan.hayward@arm.com>
1138 David Sherwood <david.sherwood@arm.com>
1140 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
1142 (try_combine): Likewise.
1143 (simplify_if_then_else): Likewise.
1144 * cse.c (cse_insn): Likewise.
1145 * dwarf2out.c (mem_loc_descriptor): Likewise.
1146 * emit-rtl.c (gen_lowpart_common): Likewise.
1147 * simplify-rtx.c (simplify_truncation): Likewise.
1148 (simplify_binary_operation_1): Likewise.
1149 (simplify_const_relational_operation): Likewise.
1150 (simplify_ternary_operation): Likewise.
1151 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
1153 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1154 Alan Hayward <alan.hayward@arm.com>
1155 David Sherwood <david.sherwood@arm.com>
1157 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
1158 * reload.c (push_reload): Likewise.
1159 (find_reloads): Likewise.
1161 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1162 Alan Hayward <alan.hayward@arm.com>
1163 David Sherwood <david.sherwood@arm.com>
1165 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
1166 (make_compound_operation_int): Likewise.
1167 (change_zero_ext): Likewise.
1168 * expr.c (convert_move): Likewise.
1169 (convert_modes): Likewise.
1170 * fwprop.c (forward_propagate_subreg): Likewise.
1171 * loop-iv.c (get_biv_step_1): Likewise.
1172 * optabs.c (widen_operand): Likewise.
1173 * postreload.c (move2add_valid_value_p): Likewise.
1174 * recog.c (simplify_while_replacing): Likewise.
1175 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1176 (simplify_binary_operation_1): Likewise. Remove redundant
1177 mode equality check.
1179 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1180 Alan Hayward <alan.hayward@arm.com>
1181 David Sherwood <david.sherwood@arm.com>
1183 * combine.c (combine_simplify_rtx): Add checks for
1184 is_a <scalar_int_mode>.
1185 (simplify_if_then_else): Likewise.
1186 (make_field_assignment): Likewise.
1187 (simplify_comparison): Likewise.
1188 * ifcvt.c (noce_try_bitop): Likewise.
1189 * loop-invariant.c (canonicalize_address_mult): Likewise.
1190 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1192 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1193 Alan Hayward <alan.hayward@arm.com>
1194 David Sherwood <david.sherwood@arm.com>
1196 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
1197 is_a <scalar_int_mode> instead of != BLKmode.
1199 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1200 Alan Hayward <alan.hayward@arm.com>
1201 David Sherwood <david.sherwood@arm.com>
1203 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
1204 instead of != VOIDmode.
1205 * combine.c (if_then_else_cond): Likewise.
1206 (change_zero_ext): Likewise.
1207 * dwarf2out.c (mem_loc_descriptor): Likewise.
1208 (loc_descriptor): Likewise.
1209 * rtlanal.c (canonicalize_condition): Likewise.
1210 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
1212 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1213 Alan Hayward <alan.hayward@arm.com>
1214 David Sherwood <david.sherwood@arm.com>
1216 * simplify-rtx.c (simplify_binary_operation_1): Use
1217 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
1219 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1220 Alan Hayward <alan.hayward@arm.com>
1221 David Sherwood <david.sherwood@arm.com>
1223 * wide-int.h (int_traits<unsigned char>) New class.
1224 (int_traits<unsigned short>) Likewise.
1225 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
1226 Use GET_MODE_UNIT_PRECISION and remove redundant test for
1228 * combine.c (set_nonzero_bits_and_sign_copies): Use
1229 is_a <scalar_int_mode>.
1230 (find_split_point): Likewise.
1231 (combine_simplify_rtx): Likewise.
1232 (simplify_logical): Likewise.
1233 (expand_compound_operation): Likewise.
1234 (expand_field_assignment): Likewise.
1235 (make_compound_operation): Likewise.
1236 (extended_count): Likewise.
1237 (change_zero_ext): Likewise.
1238 (simplify_comparison): Likewise.
1239 * dwarf2out.c (scompare_loc_descriptor): Likewise.
1240 (ucompare_loc_descriptor): Likewise.
1241 (minmax_loc_descriptor): Likewise.
1242 (mem_loc_descriptor): Likewise.
1243 (loc_descriptor): Likewise.
1244 * expmed.c (init_expmed_one_mode): Likewise.
1245 * lra-constraints.c (lra_constraint_offset): Likewise.
1246 * optabs.c (prepare_libcall_arg): Likewise.
1247 * postreload.c (move2add_note_store): Likewise.
1248 * reload.c (operands_match_p): Likewise.
1249 * rtl.h (load_extend_op): Likewise.
1250 * rtlhooks.c (gen_lowpart_general): Likewise.
1251 * simplify-rtx.c (simplify_truncation): Likewise.
1252 (simplify_unary_operation_1): Likewise.
1253 (simplify_binary_operation_1): Likewise.
1254 (simplify_const_binary_operation): Likewise.
1255 (simplify_const_relational_operation): Likewise.
1256 (simplify_subreg): Likewise.
1257 * stor-layout.c (bitwise_mode_for_mode): Likewise.
1258 * var-tracking.c (adjust_mems): Likewise.
1259 (prepare_call_arguments): Likewise.
1261 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1262 Alan Hayward <alan.hayward@arm.com>
1263 David Sherwood <david.sherwood@arm.com>
1265 * machmode.h (is_int_mode): New fuction.
1266 * combine.c (find_split_point): Use it.
1267 (combine_simplify_rtx): Likewise.
1268 (simplify_if_then_else): Likewise.
1269 (simplify_set): Likewise.
1270 (simplify_shift_const_1): Likewise.
1271 (simplify_comparison): Likewise.
1272 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
1273 * cse.c (notreg_cost): Likewise.
1274 (cse_insn): Likewise.
1275 * cselib.c (cselib_lookup_1): Likewise.
1276 * dojump.c (do_jump_1): Likewise.
1277 (do_compare_rtx_and_jump): Likewise.
1278 * dse.c (get_call_args): Likewise.
1279 * dwarf2out.c (rtl_for_decl_init): Likewise.
1280 (native_encode_initializer): Likewise.
1281 * expmed.c (emit_store_flag_1): Likewise.
1282 (emit_store_flag): Likewise.
1283 * expr.c (convert_modes): Likewise.
1284 (store_field): Likewise.
1285 (expand_expr_real_1): Likewise.
1286 * fold-const.c (fold_read_from_constant_string): Likewise.
1287 * gimple-ssa-sprintf.c (get_format_string): Likewise.
1288 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
1289 * optabs.c (expand_binop): Likewise.
1290 (expand_unop): Likewise.
1291 (expand_abs_nojump): Likewise.
1292 (expand_one_cmpl_abs_nojump): Likewise.
1293 * simplify-rtx.c (mode_signbit_p): Likewise.
1294 (val_signbit_p): Likewise.
1295 (val_signbit_known_set_p): Likewise.
1296 (val_signbit_known_clear_p): Likewise.
1297 (simplify_relational_operation_1): Likewise.
1298 * tree.c (vector_type_mode): Likewise.
1300 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1301 Alan Hayward <alan.hayward@arm.com>
1302 David Sherwood <david.sherwood@arm.com>
1304 * machmode.h (smallest_mode_for_size): Fix formatting.
1305 (smallest_int_mode_for_size): New function.
1306 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
1307 instead of smallest_mode_for_size.
1308 * combine.c (make_extraction): Likewise.
1309 * config/arc/arc.c (arc_expand_movmem): Likewise.
1310 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
1311 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
1312 * config/s390/s390.c (s390_expand_insv): Likewise.
1313 * config/sparc/sparc.c (assign_int_registers): Likewise.
1314 * config/spu/spu.c (spu_function_value): Likewise.
1315 (spu_function_arg): Likewise.
1316 * coverage.c (get_gcov_type): Likewise.
1317 (get_gcov_unsigned_t): Likewise.
1318 * dse.c (find_shift_sequence): Likewise.
1319 * expmed.c (store_bit_field_1): Likewise.
1320 * expr.c (convert_move): Likewise.
1321 (store_field): Likewise.
1322 * internal-fn.c (expand_arith_overflow): Likewise.
1323 * optabs-query.c (get_best_extraction_insn): Likewise.
1324 * optabs.c (expand_twoval_binop_libfunc): Likewise.
1325 * stor-layout.c (layout_type): Likewise.
1326 (initialize_sizetypes): Likewise.
1327 * targhooks.c (default_get_mask_mode): Likewise.
1328 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
1330 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1331 Alan Hayward <alan.hayward@arm.com>
1332 David Sherwood <david.sherwood@arm.com>
1334 * machmode.h (opt_mode::else_blk): New function.
1335 (int_mode_for_mode): Declare.
1336 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
1337 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
1339 * cfgexpand.c (expand_debug_expr): Likewise.
1340 * combine.c (gen_lowpart_or_truncate): Likewise.
1341 (gen_lowpart_for_combine): Likewise.
1342 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
1343 * config/avr/avr.c (avr_to_int_mode): Likewise.
1344 (avr_out_plus_1): Likewise.
1345 (avr_out_plus): Likewise.
1346 (avr_out_round): Likewise.
1347 * config/i386/i386.c (ix86_split_to_parts): Likewise.
1348 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
1349 (s390_expand_vcond): Likewise.
1350 * config/spu/spu.c (spu_split_immediate): Likewise.
1351 (spu_expand_mov): Likewise.
1352 * dse.c (get_stored_val): Likewise.
1353 * expmed.c (store_bit_field_1): Likewise.
1354 (convert_extracted_bit_field): Use int_mode_for_mode instead of
1356 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
1357 (extract_low_bits): Likewise.
1358 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
1359 handling rather than repeating the check.
1360 (emit_group_store): Likewise.
1361 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
1362 * optabs.c (expand_absneg_bit): Likewise.
1363 (expand_copysign_absneg): Likewise.
1364 (expand_copysign_bit): Likewise.
1365 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
1366 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
1367 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
1368 * var-tracking.c (prepare_call_arguments): Likewise.
1369 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
1370 int_mode_for_mode instead of mode_for_size.
1371 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
1373 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1374 Alan Hayward <alan.hayward@arm.com>
1375 David Sherwood <david.sherwood@arm.com>
1377 * machmode.h (int_mode_for_size): New function.
1378 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
1379 instead of mode_for_size.
1380 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
1382 * combine.c (expand_field_assignment): Use int_mode_for_size
1383 instead of mode_for_size.
1384 (make_extraction): Likewise.
1385 (simplify_shift_const_1): Likewise.
1386 (simplify_comparison): Likewise.
1387 * dojump.c (do_jump): Likewise.
1388 * dwarf2out.c (mem_loc_descriptor): Likewise.
1389 * emit-rtl.c (init_derived_machine_modes): Likewise.
1390 * expmed.c (flip_storage_order): Likewise.
1391 (convert_extracted_bit_field): Likewise.
1392 * expr.c (copy_blkmode_from_reg): Likewise.
1393 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
1394 * internal-fn.c (expand_mul_overflow): Likewise.
1395 * lower-subreg.c (simple_move): Likewise.
1396 * optabs-libfuncs.c (init_optabs): Likewise.
1397 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1398 * tree.c (vector_type_mode): Likewise.
1399 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
1400 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
1401 * tree-vect-generic.c (expand_vector_parallel): Likewise.
1402 * tree-vect-stmts.c (vectorizable_load): Likewise.
1403 (vectorizable_store): Likewise.
1405 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1406 Alan Hayward <alan.hayward@arm.com>
1407 David Sherwood <david.sherwood@arm.com>
1409 * coretypes.h (pod_mode): New type.
1410 (scalar_int_mode_pod): New typedef.
1411 * machmode.h (pod_mode): New class.
1412 (int_n_data_t::m): Change type to scalar_int_mode_pod.
1413 * genmodes.c (emit_mode_int_n): Update accordingly.
1414 * lower-subreg.h (target_lower_subreg): Change type to
1415 scalar_int_mode_pod.
1416 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
1417 scalar_int_mode_pod.
1419 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1420 Alan Hayward <alan.hayward@arm.com>
1421 David Sherwood <david.sherwood@arm.com>
1423 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
1424 machine_mode to scalar_int_mode.
1425 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
1426 (rs6000_option_override_internal): Remove cast to int.
1427 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
1428 machine_mode to scalar_int_mode.
1429 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
1430 (rs6000_option_override_internal): Remove cast to int.
1431 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
1432 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
1434 * config/s390/s390.c (s390_expand_builtin): Likewise.
1435 * coretypes.h (scalar_int_mode): New type.
1436 (opt_scalar_int_mode): New typedef.
1437 * machmode.h (scalar_int_mode): New class.
1438 (scalar_int_mode::includes_p): New function.
1439 (byte_mode): Change type to scalar_int_mode.
1440 (word_mode): Likewise.
1441 (ptr_mode): Likewise.
1442 * emit-rtl.c (byte_mode): Likewise.
1443 (word_mode): Likewise.
1444 (ptr_mode): Likewise.
1445 (init_derived_machine_modes): Update accordingly.
1446 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
1447 and MODE_PARTIAL_INT.
1448 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
1449 opt_scalar_int_mode.
1451 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1452 Alan Hayward <alan.hayward@arm.com>
1453 David Sherwood <david.sherwood@arm.com>
1455 * target.def (libgcc_floating_mode_supported_p): Take a
1457 * doc/tm.texi: Regenerate.
1458 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
1460 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
1461 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
1464 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1465 Alan Hayward <alan.hayward@arm.com>
1466 David Sherwood <david.sherwood@arm.com>
1468 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
1469 * doc/tm.texi: Regenerate.
1470 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
1471 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
1472 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
1473 * targhooks.h (default_floatn_mode): Likewise.
1474 * targhooks.c (default_floatn_mode): Likewise.
1475 * tree.c (build_common_tree_nodes): Update accordingly.
1477 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1478 Alan Hayward <alan.hayward@arm.com>
1479 David Sherwood <david.sherwood@arm.com>
1481 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
1482 (mode_iterator::iterate_p): Likewise.
1483 (mode_iterator::get_wider): Likewise.
1484 * expr.c (init_expr_target): Use opt_scalar_float_mode.
1486 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1487 Alan Hayward <alan.hayward@arm.com>
1488 David Sherwood <david.sherwood@arm.com>
1490 * coretypes.h (opt_scalar_float_mode): New typedef.
1491 * machmode.h (float_mode_for_size): New function.
1492 * emit-rtl.c (double_mode): Delete.
1493 (init_emit_once): Use float_mode_for_size.
1494 * stor-layout.c (layout_type): Likewise.
1495 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
1497 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1498 Alan Hayward <alan.hayward@arm.com>
1499 David Sherwood <david.sherwood@arm.com>
1501 * output.h (assemble_real): Take a scalar_float_mode.
1502 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
1503 * config/arm/arm.md (consttable_4): Likewise.
1504 (consttable_8): Likewise.
1505 (consttable_16): Likewise.
1506 * config/mips/mips.md (consttable_float): Likewise.
1507 * config/s390/s390.c (s390_output_pool_entry): Likewise.
1508 * varasm.c (assemble_real): Take a scalar_float_mode.
1509 (output_constant_pool_2): Update accordingly.
1510 (output_constant): Likewise.
1512 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1513 Alan Hayward <alan.hayward@arm.com>
1514 David Sherwood <david.sherwood@arm.com>
1516 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
1517 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
1518 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
1519 (native_encode_real): Likewise.
1520 (native_interpret_real): Likewise.
1521 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
1522 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1524 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1525 Alan Hayward <alan.hayward@arm.com>
1526 David Sherwood <david.sherwood@arm.com>
1528 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
1529 <scalar_float_mode>. Simplify.
1530 (gen_extend_conv_libfunc): Likewise.
1532 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1533 Alan Hayward <alan.hayward@arm.com>
1534 David Sherwood <david.sherwood@arm.com>
1536 * coretypes.h (scalar_float_mode): New type.
1537 * machmode.h (mode_traits::from_int): Use machine_mode if
1538 USE_ENUM_MODES is defined.
1539 (is_a): New function.
1541 (dyn_cast): Likewise.
1542 (scalar_float_mode): New class.
1543 (scalar_float_mode::includes_p): New function.
1544 (is_float_mode): Likewise.
1545 * gdbhooks.py (MachineModePrinter): New class.
1546 (build_pretty_printer): Use it for scalar_float_mode.
1547 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
1548 (format_helper::format_helper): Turn into a template.
1549 * genmodes.c (get_mode_class): New function.
1550 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
1551 or machine_mode if none.
1552 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
1553 as_a <scalar_float_mode>.
1554 * dwarf2out.c (mem_loc_descriptor): Likewise.
1555 (insert_float): Likewise.
1556 (add_const_value_attribute): Likewise.
1557 * simplify-rtx.c (simplify_immed_subreg): Likewise.
1558 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
1559 (expand_unop): Update accordingly.
1560 (expand_abs_nojump): Likewise.
1561 (expand_copysign_absneg): Take a scalar_float_mode.
1562 (expand_copysign_bit): Likewise.
1563 (expand_copysign): Update accordingly.
1565 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1566 Alan Hayward <alan.hayward@arm.com>
1567 David Sherwood <david.sherwood@arm.com>
1569 * coretypes.h (opt_mode): New class.
1570 * machmode.h (opt_mode): Likewise.
1571 (opt_mode::else_void): New function.
1572 (opt_mode::require): Likewise.
1573 (opt_mode::exists): Likewise.
1574 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
1575 (GET_MODE_2XWIDER_MODE): Likewise.
1576 (mode_iterator::get_wider): Update accordingly.
1577 (mode_iterator::get_2xwider): Likewise.
1578 (mode_iterator::get_known_wider): Likewise, turning into a template.
1579 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
1580 forcing a wider mode to exist.
1581 * config/cr16/cr16.h (LONG_REG_P): Likewise.
1582 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
1583 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
1584 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
1585 * lower-subreg.c (init_lower_subreg): Likewise.
1586 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
1587 on the final iteration.
1588 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
1589 a wider mode exists before asking for a move pattern.
1590 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
1591 forcing a wider mode to exist.
1592 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
1593 returning false if no such mode exists.
1594 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
1595 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
1596 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
1597 Avoid checking for a MODE_INT if we already know the mode is not a
1599 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
1600 forcing a wider mode to exist.
1601 (expmed_mult_highpart_optab): Likewise.
1602 (expmed_mult_highpart): Likewise.
1603 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
1605 * lto-streamer-in.c (lto_input_mode_table): Likewise.
1606 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
1607 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
1608 * internal-fn.c (expand_mul_overflow): Update use of
1609 GET_MODE_2XWIDER_MODE.
1610 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1611 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
1612 GET_MODE_WIDER_MODE.
1613 (convert_plusminus_to_widen): Likewise.
1614 * tree-switch-conversion.c (array_value_type): Likewise.
1615 * var-tracking.c (emit_note_insn_var_location): Likewise.
1616 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1617 Return false inside rather than outside the loop if no wider mode
1619 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
1620 and GET_MODE_2XWIDER_MODE
1621 (can_compare_p): Use else_void.
1622 * gdbhooks.py (OptMachineModePrinter): New class.
1623 (build_pretty_printer): Use it for opt_mode.
1625 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1626 Alan Hayward <alan.hayward@arm.com>
1627 David Sherwood <david.sherwood@arm.com>
1629 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
1630 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
1632 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1633 Alan Hayward <alan.hayward@arm.com>
1634 David Sherwood <david.sherwood@arm.com>
1636 * machmode.h (mode_traits): New structure.
1637 (get_narrowest_mode): New function.
1638 (mode_iterator::start): Likewise.
1639 (mode_iterator::iterate_p): Likewise.
1640 (mode_iterator::get_wider): Likewise.
1641 (mode_iterator::get_known_wider): Likewise.
1642 (mode_iterator::get_2xwider): Likewise.
1643 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
1644 (FOR_EACH_MODE): Likewise.
1645 (FOR_EACH_MODE_FROM): Likewise.
1646 (FOR_EACH_MODE_UNTIL): Likewise.
1647 (FOR_EACH_WIDER_MODE): Likewise.
1648 (FOR_EACH_2XWIDER_MODE): Likewise.
1649 * builtins.c (expand_builtin_strlen): Use new mode iterators.
1650 * combine.c (simplify_comparison): Likewise
1651 * config/i386/i386.c (type_natural_mode): Likewise.
1652 * cse.c (cse_insn): Likewise.
1653 * dse.c (find_shift_sequence): Likewise.
1654 * emit-rtl.c (init_derived_machine_modes): Likewise.
1655 (init_emit_once): Likewise.
1656 * explow.c (hard_function_value): Likewise.
1657 * expmed.c (extract_fixed_bit_field_1): Likewise.
1658 (extract_bit_field_1): Likewise.
1659 (expand_divmod): Likewise.
1660 (emit_store_flag_1): Likewise.
1661 * expr.c (init_expr_target): Likewise.
1662 (convert_move): Likewise.
1663 (alignment_for_piecewise_move): Likewise.
1664 (widest_int_mode_for_size): Likewise.
1665 (emit_block_move_via_movmem): Likewise.
1666 (copy_blkmode_to_reg): Likewise.
1667 (set_storage_via_setmem): Likewise.
1668 (compress_float_constant): Likewise.
1669 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1670 * optabs-query.c (get_best_extraction_insn): Likewise.
1671 * optabs.c (expand_binop): Likewise.
1672 (expand_twoval_unop): Likewise.
1673 (expand_twoval_binop): Likewise.
1674 (widen_leading): Likewise.
1675 (widen_bswap): Likewise.
1676 (expand_parity): Likewise.
1677 (expand_unop): Likewise.
1678 (prepare_cmp_insn): Likewise.
1679 (prepare_float_lib_cmp): Likewise.
1680 (expand_float): Likewise.
1681 (expand_fix): Likewise.
1682 (expand_sfix_optab): Likewise.
1683 * postreload.c (move2add_use_add2_insn): Likewise.
1684 * reg-stack.c (reg_to_stack): Likewise.
1685 * reginfo.c (choose_hard_reg_mode): Likewise.
1686 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
1687 * stor-layout.c (mode_for_size): Likewise.
1688 (smallest_mode_for_size): Likewise.
1689 (mode_for_vector): Likewise.
1690 (finish_bitfield_representative): Likewise.
1691 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
1692 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
1693 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1694 * var-tracking.c (prepare_call_arguments): Likewise.
1696 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1697 Alan Hayward <alan.hayward@arm.com>
1698 David Sherwood <david.sherwood@arm.com>
1700 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
1701 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
1702 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
1703 * machmode.h (mode_size): Move earlier in file.
1704 (mode_precision): Likewise.
1705 (mode_inner): Likewise.
1706 (mode_nunits): Likewise.
1707 (mode_unit_size): Likewise.
1708 (unit_unit_precision): Likewise.
1709 (mode_wider): Likewise.
1710 (mode_2xwider): Likewise.
1711 (machine_mode): New class.
1712 (mode_to_bytes): New function.
1713 (mode_to_bits): Likewise.
1714 (mode_to_precision): Likewise.
1715 (mode_to_inner): Likewise.
1716 (mode_to_unit_size): Likewise.
1717 (mode_to_unit_precision): Likewise.
1718 (mode_to_nunits): Likewise.
1719 (GET_MODE_SIZE): Use mode_to_bytes.
1720 (GET_MODE_BITSIZE): Use mode_to_bits.
1721 (GET_MODE_PRECISION): Use mode_to_precision.
1722 (GET_MODE_INNER): Use mode_to_inner.
1723 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
1724 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
1725 (GET_MODE_NUNITS): Use mode_to_nunits.
1726 * system.h (ALWAYS_INLINE): New macro.
1727 * config/powerpcspe/powerpcspe-c.c
1728 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
1729 int for arg1_mode and arg2_mode.
1731 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1732 Alan Hayward <alan.hayward@arm.com>
1733 David Sherwood <david.sherwood@arm.com>
1735 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
1736 Prefix mode names with E_ in case statements.
1737 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1738 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
1739 (aarch64_split_simd_move): Likewise.
1740 (aarch64_gen_storewb_pair): Likewise.
1741 (aarch64_gen_loadwb_pair): Likewise.
1742 (aarch64_gen_store_pair): Likewise.
1743 (aarch64_gen_load_pair): Likewise.
1744 (aarch64_get_condition_code_1): Likewise.
1745 (aarch64_constant_pool_reload_icode): Likewise.
1746 (get_rsqrte_type): Likewise.
1747 (get_rsqrts_type): Likewise.
1748 (get_recpe_type): Likewise.
1749 (get_recps_type): Likewise.
1750 (aarch64_gimplify_va_arg_expr): Likewise.
1751 (aarch64_simd_container_mode): Likewise.
1752 (aarch64_emit_load_exclusive): Likewise.
1753 (aarch64_emit_store_exclusive): Likewise.
1754 (aarch64_expand_compare_and_swap): Likewise.
1755 (aarch64_gen_atomic_cas): Likewise.
1756 (aarch64_emit_bic): Likewise.
1757 (aarch64_emit_atomic_swap): Likewise.
1758 (aarch64_emit_atomic_load_op): Likewise.
1759 (aarch64_evpc_trn): Likewise.
1760 (aarch64_evpc_uzp): Likewise.
1761 (aarch64_evpc_zip): Likewise.
1762 (aarch64_evpc_ext): Likewise.
1763 (aarch64_evpc_rev): Likewise.
1764 (aarch64_evpc_dup): Likewise.
1765 (aarch64_gen_ccmp_first): Likewise.
1766 (aarch64_gen_ccmp_next): Likewise.
1767 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
1768 (alpha_emit_xfloating_libcall): Likewise.
1769 (emit_insxl): Likewise.
1770 (alpha_arg_type): Likewise.
1771 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
1772 (arc_preferred_simd_mode): Likewise.
1773 (arc_secondary_reload): Likewise.
1774 (get_arc_condition_code): Likewise.
1775 (arc_print_operand): Likewise.
1776 (arc_legitimate_constant_p): Likewise.
1777 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1778 * config/arc/arc.md (casesi_load): Likewise.
1779 (casesi_compact_jump): Likewise.
1780 * config/arc/predicates.md (proper_comparison_operator): Likewise.
1781 (cc_use_register): Likewise.
1782 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1783 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
1784 (arm_init_iwmmxt_builtins): Likewise.
1785 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
1786 (neon_expand_vector_init): Likewise.
1787 (arm_attr_length_move_neon): Likewise.
1788 (maybe_get_arm_condition_code): Likewise.
1789 (arm_emit_vector_const): Likewise.
1790 (arm_preferred_simd_mode): Likewise.
1791 (arm_output_iwmmxt_tinsr): Likewise.
1792 (thumb1_output_casesi): Likewise.
1793 (thumb2_output_casesi): Likewise.
1794 (arm_emit_load_exclusive): Likewise.
1795 (arm_emit_store_exclusive): Likewise.
1796 (arm_expand_compare_and_swap): Likewise.
1797 (arm_evpc_neon_vuzp): Likewise.
1798 (arm_evpc_neon_vzip): Likewise.
1799 (arm_evpc_neon_vrev): Likewise.
1800 (arm_evpc_neon_vtrn): Likewise.
1801 (arm_evpc_neon_vext): Likewise.
1802 (arm_validize_comparison): Likewise.
1803 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
1804 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
1805 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
1806 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
1807 (c6x_preferred_simd_mode): Likewise.
1808 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
1809 (epiphany_rtx_costs): Likewise.
1810 * config/epiphany/predicates.md (proper_comparison_operator):
1812 * config/frv/frv.c (condexec_memory_operand): Likewise.
1813 (frv_emit_move): Likewise.
1814 (output_move_single): Likewise.
1815 (output_condmove_single): Likewise.
1816 (frv_hard_regno_mode_ok): Likewise.
1817 (frv_matching_accg_mode): Likewise.
1818 * config/h8300/h8300.c (split_adds_subs): Likewise.
1819 (h8300_rtx_costs): Likewise.
1820 (h8300_print_operand): Likewise.
1821 (compute_mov_length): Likewise.
1822 (output_logical_op): Likewise.
1823 (compute_logical_op_length): Likewise.
1824 (compute_logical_op_cc): Likewise.
1825 (h8300_shift_needs_scratch_p): Likewise.
1826 (output_a_shift): Likewise.
1827 (compute_a_shift_length): Likewise.
1828 (compute_a_shift_cc): Likewise.
1829 (expand_a_rotate): Likewise.
1830 (output_a_rotate): Likewise.
1831 * config/i386/i386.c (classify_argument): Likewise.
1832 (function_arg_advance_32): Likewise.
1833 (function_arg_32): Likewise.
1834 (function_arg_64): Likewise.
1835 (function_value_64): Likewise.
1836 (ix86_gimplify_va_arg): Likewise.
1837 (ix86_legitimate_constant_p): Likewise.
1838 (put_condition_code): Likewise.
1839 (split_double_mode): Likewise.
1840 (ix86_avx256_split_vector_move_misalign): Likewise.
1841 (ix86_expand_vector_logical_operator): Likewise.
1842 (ix86_split_idivmod): Likewise.
1843 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
1844 (ix86_build_const_vector): Likewise.
1845 (ix86_build_signbit_mask): Likewise.
1846 (ix86_match_ccmode): Likewise.
1847 (ix86_cc_modes_compatible): Likewise.
1848 (ix86_expand_branch): Likewise.
1849 (ix86_expand_sse_cmp): Likewise.
1850 (ix86_expand_sse_movcc): Likewise.
1851 (ix86_expand_int_sse_cmp): Likewise.
1852 (ix86_expand_vec_perm_vpermi2): Likewise.
1853 (ix86_expand_vec_perm): Likewise.
1854 (ix86_expand_sse_unpack): Likewise.
1855 (ix86_expand_int_addcc): Likewise.
1856 (ix86_split_to_parts): Likewise.
1857 (ix86_vectorize_builtin_gather): Likewise.
1858 (ix86_vectorize_builtin_scatter): Likewise.
1859 (avx_vpermilp_parallel): Likewise.
1860 (inline_memory_move_cost): Likewise.
1861 (ix86_tieable_integer_mode_p): Likewise.
1862 (x86_maybe_negate_const_int): Likewise.
1863 (ix86_expand_vector_init_duplicate): Likewise.
1864 (ix86_expand_vector_init_one_nonzero): Likewise.
1865 (ix86_expand_vector_init_one_var): Likewise.
1866 (ix86_expand_vector_init_concat): Likewise.
1867 (ix86_expand_vector_init_interleave): Likewise.
1868 (ix86_expand_vector_init_general): Likewise.
1869 (ix86_expand_vector_set): Likewise.
1870 (ix86_expand_vector_extract): Likewise.
1871 (emit_reduc_half): Likewise.
1872 (ix86_emit_i387_round): Likewise.
1873 (ix86_mangle_type): Likewise.
1874 (ix86_expand_round_sse4): Likewise.
1875 (expand_vec_perm_blend): Likewise.
1876 (canonicalize_vector_int_perm): Likewise.
1877 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
1878 (expand_vec_perm_1): Likewise.
1879 (expand_vec_perm_interleave3): Likewise.
1880 (expand_vec_perm_even_odd_pack): Likewise.
1881 (expand_vec_perm_even_odd_1): Likewise.
1882 (expand_vec_perm_broadcast_1): Likewise.
1883 (ix86_vectorize_vec_perm_const_ok): Likewise.
1884 (ix86_expand_vecop_qihi): Likewise.
1885 (ix86_expand_mul_widen_hilo): Likewise.
1886 (ix86_expand_sse2_abs): Likewise.
1887 (ix86_expand_pextr): Likewise.
1888 (ix86_expand_pinsr): Likewise.
1889 (ix86_preferred_simd_mode): Likewise.
1890 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
1891 * config/i386/sse.md (*andnot<mode>3): Likewise.
1892 (<mask_codefor><code><mode>3<mask_name>): Likewise.
1893 (*<code><mode>3): Likewise.
1894 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
1895 (ia64_expand_atomic_op): Likewise.
1896 (ia64_arg_type): Likewise.
1897 (ia64_mode_to_int): Likewise.
1898 (ia64_scalar_mode_supported_p): Likewise.
1899 (ia64_vector_mode_supported_p): Likewise.
1900 (expand_vec_perm_broadcast): Likewise.
1901 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
1902 (iq2000_function_arg_advance): Likewise.
1903 (iq2000_function_arg): Likewise.
1904 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
1905 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
1906 (m68k_libcall_value): Likewise.
1907 (m68k_function_value): Likewise.
1908 (sched_attr_op_type): Likewise.
1909 * config/mcore/mcore.c (mcore_output_move): Likewise.
1910 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
1912 (microblaze_function_arg): Likewise.
1913 * config/mips/mips.c (mips16_build_call_stub): Likewise.
1914 (mips_print_operand): Likewise.
1915 (mips_mode_ok_for_mov_fmt_p): Likewise.
1916 (mips_vector_mode_supported_p): Likewise.
1917 (mips_preferred_simd_mode): Likewise.
1918 (mips_expand_vpc_loongson_even_odd): Likewise.
1919 (mips_expand_vec_unpack): Likewise.
1920 (mips_expand_vi_broadcast): Likewise.
1921 (mips_expand_vector_init): Likewise.
1922 (mips_expand_vec_reduc): Likewise.
1923 (mips_expand_msa_cmp): Likewise.
1924 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
1925 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
1926 (cc_flags_for_mode): Likewise.
1927 * config/msp430/msp430.c (msp430_print_operand): Likewise.
1928 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
1929 (nds32_output_casesi_pc_relative): Likewise.
1930 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1931 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
1932 (nvptx_gen_unpack): Likewise.
1933 (nvptx_gen_pack): Likewise.
1934 (nvptx_gen_shuffle): Likewise.
1935 (nvptx_gen_wcast): Likewise.
1936 (nvptx_preferred_simd_mode): Likewise.
1937 * config/pa/pa.c (pa_secondary_reload): Likewise.
1938 * config/pa/predicates.md (base14_operand): Likewise.
1939 * config/powerpcspe/powerpcspe-c.c
1940 (altivec_resolve_overloaded_builtin): Likewise.
1941 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
1943 (rs6000_preferred_simd_mode): Likewise.
1944 (output_vec_const_move): Likewise.
1945 (rs6000_expand_vector_extract): Likewise.
1946 (rs6000_split_vec_extract_var): Likewise.
1947 (reg_offset_addressing_ok_p): Likewise.
1948 (rs6000_legitimate_offset_address_p): Likewise.
1949 (rs6000_legitimize_address): Likewise.
1950 (rs6000_emit_set_const): Likewise.
1951 (rs6000_const_vec): Likewise.
1952 (rs6000_emit_move): Likewise.
1953 (spe_build_register_parallel): Likewise.
1954 (rs6000_darwin64_record_arg_recurse): Likewise.
1955 (swap_selector_for_mode): Likewise.
1956 (spe_init_builtins): Likewise.
1957 (paired_init_builtins): Likewise.
1958 (altivec_init_builtins): Likewise.
1959 (do_load_for_compare): Likewise.
1960 (rs6000_generate_compare): Likewise.
1961 (rs6000_expand_float128_convert): Likewise.
1962 (emit_load_locked): Likewise.
1963 (emit_store_conditional): Likewise.
1964 (rs6000_output_function_epilogue): Likewise.
1965 (rs6000_handle_altivec_attribute): Likewise.
1966 (rs6000_function_value): Likewise.
1967 (emit_fusion_gpr_load): Likewise.
1968 (emit_fusion_p9_load): Likewise.
1969 (emit_fusion_p9_store): Likewise.
1970 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
1971 (fusion_gpr_mem_load): Likewise.
1972 (fusion_addis_mem_combo_load): Likewise.
1973 (fusion_addis_mem_combo_store): Likewise.
1974 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
1975 (fusion_gpr_mem_load): Likewise.
1976 (fusion_addis_mem_combo_load): Likewise.
1977 (fusion_addis_mem_combo_store): Likewise.
1978 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1980 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
1981 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
1982 (rs6000_preferred_simd_mode): Likewise.
1983 (output_vec_const_move): Likewise.
1984 (rs6000_expand_vector_extract): Likewise.
1985 (rs6000_split_vec_extract_var): Likewise.
1986 (reg_offset_addressing_ok_p): Likewise.
1987 (rs6000_legitimate_offset_address_p): Likewise.
1988 (rs6000_legitimize_address): Likewise.
1989 (rs6000_emit_set_const): Likewise.
1990 (rs6000_const_vec): Likewise.
1991 (rs6000_emit_move): Likewise.
1992 (rs6000_darwin64_record_arg_recurse): Likewise.
1993 (swap_selector_for_mode): Likewise.
1994 (paired_init_builtins): Likewise.
1995 (altivec_init_builtins): Likewise.
1996 (rs6000_expand_float128_convert): Likewise.
1997 (emit_load_locked): Likewise.
1998 (emit_store_conditional): Likewise.
1999 (rs6000_output_function_epilogue): Likewise.
2000 (rs6000_handle_altivec_attribute): Likewise.
2001 (rs6000_function_value): Likewise.
2002 (emit_fusion_gpr_load): Likewise.
2003 (emit_fusion_p9_load): Likewise.
2004 (emit_fusion_p9_store): Likewise.
2005 * config/rx/rx.c (rx_gen_move_template): Likewise.
2006 (flags_from_mode): Likewise.
2007 * config/s390/predicates.md (s390_alc_comparison): Likewise.
2008 (s390_slb_comparison): Likewise.
2009 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
2010 (s390_vector_mode_supported_p): Likewise.
2011 (s390_cc_modes_compatible): Likewise.
2012 (s390_match_ccmode_set): Likewise.
2013 (s390_canonicalize_comparison): Likewise.
2014 (s390_emit_compare_and_swap): Likewise.
2015 (s390_branch_condition_mask): Likewise.
2016 (s390_rtx_costs): Likewise.
2017 (s390_secondary_reload): Likewise.
2018 (__SECONDARY_RELOAD_CASE): Likewise.
2019 (s390_expand_cs): Likewise.
2020 (s390_preferred_simd_mode): Likewise.
2021 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
2022 * config/sh/sh.c (sh_print_operand): Likewise.
2023 (dump_table): Likewise.
2024 (sh_secondary_reload): Likewise.
2025 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2026 * config/sh/sh.md (casesi_worker_1): Likewise.
2027 (casesi_worker_2): Likewise.
2028 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
2029 (fcc_comparison_operator): Likewise.
2030 * config/sparc/sparc.c (sparc_expand_move): Likewise.
2031 (emit_soft_tfmode_cvt): Likewise.
2032 (sparc_preferred_simd_mode): Likewise.
2033 (output_cbranch): Likewise.
2034 (sparc_print_operand): Likewise.
2035 (sparc_expand_vec_perm_bmask): Likewise.
2036 (vector_init_bshuffle): Likewise.
2037 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
2038 (spu_vector_mode_supported_p): Likewise.
2039 (spu_expand_insv): Likewise.
2040 (spu_emit_branch_or_set): Likewise.
2041 (spu_handle_vector_attribute): Likewise.
2042 (spu_builtin_splats): Likewise.
2043 (spu_builtin_extract): Likewise.
2044 (spu_builtin_promote): Likewise.
2045 (spu_expand_sign_extend): Likewise.
2046 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
2047 (tilegx_simd_int): Likewise.
2048 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
2049 (tilepro_simd_int): Likewise.
2050 * config/v850/v850.c (const_double_split): Likewise.
2051 (v850_print_operand): Likewise.
2052 (ep_memory_offset): Likewise.
2053 * config/vax/vax.c (vax_rtx_costs): Likewise.
2054 (vax_output_int_move): Likewise.
2055 (vax_output_int_add): Likewise.
2056 (vax_output_int_subtract): Likewise.
2057 * config/visium/predicates.md (visium_branch_operator): Likewise.
2058 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
2059 (visium_print_operand_address): Likewise.
2060 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2061 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
2062 (xtensa_expand_conditional_branch): Likewise.
2063 (xtensa_copy_incoming_a7): Likewise.
2064 (xtensa_output_literal): Likewise.
2065 * dfp.c (decimal_real_maxval): Likewise.
2066 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
2068 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2069 Alan Hayward <alan.hayward@arm.com>
2070 David Sherwood <david.sherwood@arm.com>
2072 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
2073 (mode_nunits_inline): Likewise.
2074 (mode_inner_inline): Likewise.
2075 (mode_unit_size_inline): Likewise.
2076 (mode_unit_precision_inline): Likewise.
2077 (emit_insn_modes_h): Likewise. Also emit a #define of the
2079 (emit_mode_wider): Add an E_ prefix to mode names.
2080 (emit_mode_complex): Likewise.
2081 (emit_mode_inner): Likewise.
2082 (emit_mode_adjustments): Likewise.
2083 (emit_mode_int_n): Likewise.
2084 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
2085 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
2086 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
2087 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
2088 (CRC32_BUILTIN, ENTRY): Likewise.
2089 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
2090 (aarch64_pop_regs): Likewise.
2091 (aarch64_process_components): Likewise.
2092 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
2093 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
2094 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
2095 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
2096 * config/arm/arm.c (arm_init_libfuncs): Likewise.
2097 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
2099 * config/i386/i386-builtin.def (pcmpestr): Likewise.
2100 (pcmpistr): Likewise.
2101 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
2102 * config/mmix/mmix.c (mmix_output_condition): Likewise.
2103 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
2105 * config/rl78/rl78.c (mduc_regs): Likewise.
2106 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2107 (htm_expand_builtin): Likewise.
2108 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
2109 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
2110 * config/xtensa/xtensa.c (print_operand): Likewise.
2111 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
2112 (NUM_MODE_VECTOR_INT): Likewise.
2113 * genoutput.c (null_operand): Likewise.
2114 (output_operand_data): Likewise.
2115 * genrecog.c (print_parameter_value): Likewise.
2116 * lra.c (debug_operand_data): Likewise.
2118 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2120 * dwarf2out.c (scompare_loc_descriptor_wide)
2121 (scompare_loc_descriptor_narrow): New functions, split out from...
2122 (scompare_loc_descriptor): ...here.
2123 * expmed.c (emit_store_flag_int): New function, split out from...
2124 (emit_store_flag): ...here.
2126 2017-08-30 Richard Biener <rguenther@suse.de>
2128 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
2129 (dwarf2out_early_finish): Move setting of AT_pubnames from
2130 early debug output to early finish.
2132 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
2134 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
2135 and -mdata-region to the assembler.
2137 2017-08-30 Richard Biener <rguenther@suse.de>
2139 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
2141 (gen_subprogram_die): Add DW_AT_object_pointer only early.
2142 (dwarf2out_early_global_decl): Only generate a DIE for the
2143 abstract origin if it doesn't already exist or is a declaration DIE.
2144 (resolve_addr): Do not add the linkage name twice when
2145 generating a stub DIE for the DW_TAG_GNU_call_site target.
2147 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2149 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2150 Use machine_mode rather than int for arg1_mode.
2152 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
2155 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
2156 that the second argument of the built-in functions to unpack
2157 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
2158 switch statement instead a lot of if statements.
2159 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
2160 Allow 64-bit values to be in Altivec registers as well as
2161 traditional floating point registers.
2162 (pack<mode>, FMOVE128_VSX iterator): Likewise.
2164 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
2166 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
2167 MAX_REGS_PER_ADDRESS == 1.
2169 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
2171 * config/i386/i386.opt (flag_fentry): Do not init to -1.
2172 * config/i386/i386.c (ix86_option_override_internal): Simplify
2173 setting of opts->x_flag_entry.
2175 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2176 Jakub Jelinek <jakub@redhat.com>
2177 Richard Biener <rguenther@suse.de>
2179 PR tree-optimization/81503
2180 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
2181 folded constant fits in the target type; reorder tests for clarity.
2183 2017-08-29 Martin Liska <mliska@suse.cz>
2185 * passes.def: Include pass_lower_switch.
2186 * stmt.c (dump_case_nodes): Remove and move to
2187 tree-switch-conversion.
2188 (case_values_threshold): Likewise.
2189 (expand_switch_as_decision_tree_p): Likewise.
2190 (emit_case_decision_tree): Likewise.
2191 (expand_case): Likewise.
2192 (balance_case_nodes): Likewise.
2193 (node_has_low_bound): Likewise.
2194 (node_has_high_bound): Likewise.
2195 (node_is_bounded): Likewise.
2196 (emit_case_nodes): Likewise.
2197 (struct simple_case_node): New struct.
2198 (add_case_node): Remove.
2199 (emit_case_dispatch_table): Use vector instead of case_list.
2200 (reset_out_edges_aux): Remove.
2201 (compute_cases_per_edge): Likewise.
2202 (expand_case): Build list of simple_case_node.
2203 (expand_sjlj_dispatch_table): Use it.
2204 * tree-switch-conversion.c (struct case_node): Moved from
2205 stmt.c and adjusted.
2206 (emit_case_nodes): Likewise.
2207 (node_has_low_bound): Likewise.
2208 (node_has_high_bound): Likewise.
2209 (node_is_bounded): Likewise.
2210 (case_values_threshold): Likewise.
2211 (reset_out_edges_aux): Likewise.
2212 (compute_cases_per_edge): Likewise.
2213 (add_case_node): Likewise.
2214 (dump_case_nodes): Likewise.
2215 (balance_case_nodes): Likewise.
2216 (expand_switch_as_decision_tree_p): Likewise.
2217 (emit_jump): Likewise.
2218 (emit_case_decision_tree): Likewise.
2219 (try_switch_expansion): Likewise.
2220 (do_jump_if_equal): Likewise.
2221 (emit_cmp_and_jump_insns): Likewise.
2222 (fix_phi_operands_for_edge): New function.
2223 (record_phi_operand_mapping): Likewise.
2224 (class pass_lower_switch): New pass.
2225 (pass_lower_switch::execute): New function.
2226 (make_pass_lower_switch): Likewise.
2227 (conditional_probability):
2228 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
2229 * tree-pass.h: Add make_pass_lower_switch.
2231 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
2234 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
2237 2017-08-29 Richard Biener <rguenther@suse.de>
2239 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
2240 we do not add a DW_AT_inline attribute twice.
2241 (gen_subprogram_die): Remove code setting DW_AT_inline on
2242 DECL_ABSTRACT_P nodes.
2244 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
2246 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
2247 calls to internal functions.
2248 (gimplify_modify_expr): Likewise.
2249 * tree-call-cdce.c (use_internal_fn): Likewise.
2250 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
2251 (convert_to_divmod): Set the nothrow flag.
2252 * tree-if-conv.c (predicate_mem_writes): Likewise.
2253 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
2254 (vectorizable_call): Likewise.
2255 (vectorizable_store): Likewise.
2256 (vectorizable_load): Likewise.
2257 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
2258 (vect_recog_mask_conversion_pattern): Likewise.
2260 2017-08-29 Martin Liska <mliska@suse.cz>
2263 * gcc.c (driver_handle_option): Add new argument.
2264 * opts-common.c (handle_option): Pass
2265 target_option_override_hook.
2266 * opts-global.c (lang_handle_option): Add new option.
2267 (set_default_handlers): Add new argument.
2268 (decode_options): Likewise.
2269 * opts.c (target_handle_option): Likewise.
2270 (common_handle_option): Call target_option_override_hook.
2271 * opts.h (struct cl_option_handler_func): Add hook for
2272 target option override.
2273 (struct cl_option_handlers): Likewise.
2274 (set_default_handlers): Add new argument.
2275 (decode_options): Likewise.
2276 (common_handle_option): Likewise.
2277 (target_handle_option): Likewise.
2278 * toplev.c (toplev::main): Pass targetm.target_option.override
2281 2017-08-29 Richard Biener <rguenther@suse.de>
2282 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2284 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
2285 life to the active subtree.
2287 2017-08-28 Jeff Law <law@redhat.com>
2289 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
2290 derive_equivalences.
2291 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
2293 (edge_info::derive_equivalences): New private member function
2295 * tree-ssa-dom.c (class edge_info): Changed from a struct
2296 to a class. Add ctor/dtor, methods and data members.
2297 (edge_info::edge_info): Renamed from allocate_edge_info.
2298 Initialize additional members.
2299 (edge_info::~edge_info): New.
2300 (free_dom_edge_info): Delete the edge info.
2301 (record_edge_info): Use new class & associated member functions.
2302 Tighten forms for testing for edge equivalences.
2303 (record_temporary_equivalences): Iterate over the simple
2304 equivalences rather than assuming there's only one per edge.
2305 (cprop_into_successor_phis): Iterate over the simple
2306 equivalences rather than assuming there's only one per edge.
2307 (optimize_stmt): Use operand_equal_p rather than pointer
2308 equality for mini-DSE code.
2310 2017-08-28 Nathan Sidwell <nathan@acm.org>
2312 * gcc.c (execute): Fold SIGPIPE handling into switch
2313 statement. Adjust internal error message.
2315 2017-08-28 Richard Biener <rguenther@suse.de>
2318 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
2319 Do nothing for removed DIEs.
2321 2017-08-28 Richard Biener <rguenther@suse.de>
2323 PR tree-optimization/81977
2324 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
2327 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
2330 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
2331 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
2332 using targetm.gen_mem_thread_fence.
2334 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
2337 * config/i386/i386.md (*<btsc><mode>): Change operand 2
2338 predicate to register_operand. Reorder operands.
2339 (*btr<mode>): Ditto.
2340 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
2341 (*btr<mode>_mask): Ditto.
2343 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
2345 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
2346 * config/rs6000/xmmintrin.h: New file.
2347 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
2349 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2352 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
2353 parameter and_insn and return it.
2354 (recombine_lvx_pattern): Insert a copy to ensure availability of
2355 the base register of the copied masking operation at the point of
2356 the instruction replacement.
2357 (recombine_stvx_pattern): Likewise.
2359 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
2361 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
2362 undocumented switches.
2363 (-mpower9-dform-vector): Likewise.
2364 (-mpower9-dform): Likewise.
2365 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
2366 comments to delete references to -mpower9-dform* switches.
2367 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
2368 Delete reference to -mpower9-dform* switches, test for
2369 -mpower9-vector instead.
2370 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
2371 (OTHER_P9_VECTOR_MASKS): Likewise.
2372 (POWERPC_MASKS): Likewise.
2373 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
2374 tests against -mpower9-dform* to -mpower9-vector. Delete code
2375 that checked for -mpower9-dform* consistancy with other options.
2376 Add test for -mpower9-misc to enable other power9 switches.
2377 (rs6000_init_hard_regno_mode_ok): Likewise.
2378 (rs6000_option_override_internal): Likewise.
2379 (rs6000_emit_prologue): Likewise.
2380 (rs6000_emit_epilogue): Likewise.
2381 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
2382 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
2383 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
2385 (emit_fusion_p9_store): Likewise.
2386 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
2387 resetting these macros if the assembler does not support ISA 3.0
2389 (TARGET_P9_DFORM_VECTOR): Likewise.
2390 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
2391 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
2393 2017-08-25 Alan Modra <amodra@gmail.com>
2395 PR rtl-optimization/81747
2396 * cse.c (cse_extended_basic_block): Don't attempt to record
2397 equivalences for degenerate conditional branches that branch
2398 to their fall-through.
2400 2017-08-24 Martin Sebor <msebor@redhat.com>
2403 * gimple-fold.c (size_must_be_zero_p): New function.
2404 (gimple_fold_builtin_memory_op): Call it.
2406 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
2408 * config/rs6000/mm_malloc.h: New file.
2410 2017-08-24 Bin Cheng <bin.cheng@arm.com>
2412 PR tree-optimization/81913
2413 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
2414 analysis when either IVs in condition can wrap.
2416 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
2418 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
2419 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
2421 2017-08-24 Richard Biener <rguenther@suse.de>
2424 * targhooks.c (default_target_can_inline_p): Properly
2425 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
2426 is present and always compare.
2427 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
2428 infer -mfpmath=sse from TARGET_SSE_P.
2429 (ix86_can_inline_p): Properly use target_option_default_node when
2430 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
2432 2017-08-24 Richard Biener <rguenther@suse.de>
2435 * dwarf2out.c (output_die): Handle flag_generate_offload like
2437 (output_comp_unit): Likewise.
2438 (gen_array_type_die): Likewise.
2439 (dwarf2out_early_finish): Likewise.
2440 (note_variable_value_in_expr): Likewise.
2441 (dwarf2out_finish): Likewise. Adjust assert.
2442 * cgraphunit.c (symbol_table::compile): Move setting of
2443 flag_generate_offload earlier ...
2444 (symbol_table::finalize_compilation_unit): ... here, before
2445 early debug finalization.
2447 2017-08-24 Richard Biener <rguenther@suse.de>
2449 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
2450 and ipa-fnsummary.h.
2451 (ix86_can_inline_p): When ix86_fpmath flags do not match
2452 check whether the callee uses FP math at all.
2454 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
2457 * tree-ssanames.c (get_nonzero_bits): Use element_precision
2458 instead of TYPE_PRECISION.
2460 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
2461 Alan Hayward <alan.hayward@arm.com>
2462 David Sherwood <david.sherwood@arm.com>
2464 * combine.c (make_extraction): Use subreg_offset_from_lsb.
2466 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
2468 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
2470 (ix86_frame::stack_realign_allocate): New field.
2471 (struct machine_frame_state): Modify comments.
2472 (machine_frame_state::sp_realigned_fp_end): New field.
2473 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
2475 (sp_valid_at): Add assertion to assure no attempt to access invalid
2476 offset of a realigned stack.
2477 (fp_valid_at): Likewise.
2478 (choose_baseaddr): Modify comments.
2479 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
2480 ix86_expand_prologue.
2481 (ix86_expand_prologue): Modify stack realignment and allocation.
2482 (ix86_expand_epilogue): Modify comments.
2483 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
2484 avx2_runtime, avx512f, and avx512f_runtime.
2486 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
2488 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
2489 (mstackrealign): Do not init to -1.
2490 * config/i386/i386.c (ix86_option_override_internal):
2491 Check opts_set, not opts when setting default value of
2492 opts->x_ix86_force_align_arg_pointer.
2494 2017-08-23 Richard Biener <rguenther@suse.de>
2496 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
2497 lang_hooks.decl_printable_name.
2498 * print-rtl-function.c (print_rtx_function): Likewise.
2499 * tree-pretty-print.c (dump_function_header): Likewise.
2501 2017-08-23 Richard Biener <rguenther@suse.de>
2504 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
2505 -g0 at compile-time.
2507 2017-08-23 Tamar Christina <tamar.christina@arm.com>
2510 * doc/sourcebuild.texi (Other hardware attributes):
2513 2017-08-23 Tamar Christina <tamar.christina@arm.com>
2516 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
2517 Fix single-use check.
2519 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
2521 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
2523 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
2525 * doc/install.texi: Modify to add more details on running selected
2528 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
2530 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
2531 is combined with -mabi=ms.
2532 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
2535 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2537 PR tree-optimization/81488
2538 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
2539 and cached_basis fields.
2540 (MAX_SPREAD): New constant.
2541 (alloc_cand_and_find_basis): Initialize new fields.
2542 (clear_visited): New function.
2543 (create_phi_basis_1): Rename from create_phi_basis, set visited
2544 and cached_basis fields.
2545 (create_phi_basis): New wrapper function.
2546 (phi_add_costs_1): Rename from phi_add_costs, add spread
2547 parameter, set visited field, short-circuit when limits reached.
2548 (phi_add_costs): New wrapper function.
2549 (record_phi_increments_1): Rename from record_phi_increments, set
2551 (record_phi_increments): New wrapper function.
2552 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
2553 (phi_incr_cost): New wrapper function.
2554 (all_phi_incrs_profitable_1): Rename from
2555 all_phi_incrs_profitable, set visited field.
2556 (all_phi_incrs_profitable): New wrapper function.
2558 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
2559 Alan Hayward <alan.hayward@arm.com>
2560 David Sherwood <david.sherwood@arm.com>
2562 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
2563 that takes the outer and inner modes.
2564 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
2565 comparison as the canonical test for a paradoxical subreg.
2566 * combine.c (simplify_set): Use paradoxical_subreg_p.
2567 (make_extraction): Likewise.
2568 (force_to_mode): Likewise.
2569 (rtx_equal_for_field_assignment_p): Likewise.
2570 (gen_lowpart_for_combine): Likewise.
2571 (simplify_comparison): Likewise.
2572 * cse.c (equiv_constant): Likewise.
2573 * expmed.c (store_bit_field_1): Likewise.
2574 * final.c (alter_subreg): Likewise.
2575 * fwprop.c (propagate_rtx): Likewise.
2576 (forward_propagate_subreg): Likewise.
2577 * ira-conflicts.c (ira_build_conflicts): Likewise.
2578 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
2579 * lra-constraints.c (curr_insn_transform): Likewise.
2580 (split_reg): Likewise.
2581 * lra-eliminations.c (move_plus_up): Likewise.
2582 (lra_eliminate_regs_1): Likewise.
2583 * recog.c (general_operand): Likewise.
2584 * ree.c (combine_reaching_defs): Likewise.
2585 * reload.c (push_reload): Likewise.
2586 (find_reloads): Likewise.
2587 * reload1.c (elimination_effects): Likewise.
2588 (compute_reload_subreg_offset): Likewise.
2589 (choose_reload_regs): Likewise.
2590 * rtlanal.c (subreg_lsb_1): Likewise.
2591 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2592 (simplify_subreg): Likewise.
2593 * var-tracking.c (track_loc_p): Likewise.
2594 * emit-rtl.c (byte_lowpart_offset): Likewise.
2595 (paradoxical_subreg_p): Delete out-of-line definition.
2597 2017-08-22 Jeff Law <law@redhat.com>
2599 PR tree-optimization/81741
2600 PR tree-optimization/71947
2601 * tree-ssa-dom.c: Include tree-inline.h.
2602 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
2603 equivalences if one is more expensive to compute than the other.
2604 * tree-ssa-scopedtables.h (class const_or_copies): Make
2605 record_const_or_copy_raw method private.
2606 (class avail_exprs_stack): New method simplify_binary_operation.
2607 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
2608 avail_exprs_stack::simplify_binary_operation as needed.
2609 (avail_exprs_stack::simplify_binary_operation): New function.
2611 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
2613 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
2614 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
2615 (DOT_SYMBOLS): Likewise.
2616 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
2617 (RELOCATABLE_NEEDS_FIXUP): Likewise.
2618 (RS6000_ABI_NAME): Likewise.
2619 (TARGET_CMODEL): Likewise.
2620 (TOC_SECTION_ASM_OP): Likewise.
2621 (SET_CMODEL): New macro.
2622 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
2624 2017-08-22 Richard Biener <rguenther@suse.de>
2626 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
2627 to allow for free-lang-data replacements similar to verify_type_variant.
2629 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
2630 Alan Hayward <alan.hayward@arm.com>
2631 David Sherwood <david.sherwood@arm.com>
2633 * config/aarch64/aarch64.md (casesi): Use DImode rather than
2634 VOIDmode for the LABEL_REF.
2636 2017-08-22 Richard Biener <rguenther@suse.de>
2638 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
2640 2017-08-22 Richard Biener <rguenther@suse.de>
2642 * common.opt (feliminate-dwarf2-dups): Ignore.
2643 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
2644 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
2645 same_die_p_wrap, compute_section_prefix,
2646 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
2647 (comdat_symbol_id, comdat_symbol_number): Likewise.
2648 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
2650 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
2651 (output_die): Mark unreachable path unreachable.
2652 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
2653 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
2654 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
2655 (dwarf2out_early_finish): Likewise.
2657 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
2659 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
2661 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
2664 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
2665 not VAR_P. Filter attribute warnings with OPT_Wattributes.
2666 (avr_attribute_table) <io, io_low, address>: Initialize
2667 .decl_required with true.
2669 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
2671 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
2672 undocumented debugging options.
2673 (-mvsx-scalar-double): Likewise.
2674 (-mallow-df-permute): Likewise.
2675 (-mvectorize-builtins): Likewise.
2676 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2677 (rs6000_builtin_vectorized_function): Likewise.
2678 (rs6000_builtin_md_vectorized_function): Likewise.
2679 (rs6000_opt_vars): Likewise.
2681 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
2684 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
2685 (*btrq_imm): Rename from *btrq.
2686 (*btcq_imm): Rename from *btcq.
2687 (btsc): New code attribute.
2688 (*<btsc><mode>): New insn pattern.
2689 (*btr<mode>): Ditto.
2690 (*<btsc><mode>_mask): New insn_and_split pattern.
2691 (*btr<mode>_mask): Ditto.
2693 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2694 Alan Hayward <alan.hayward@arm.com>
2695 David Sherwood <david.sherwood@arm.com>
2697 * function.c (pad_below): Simplify padding calculation.
2699 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2700 Alan Hayward <alan.hayward@arm.com>
2701 David Sherwood <david.sherwood@arm.com>
2703 * target.def (function_prologue): Remove frame size argument.
2704 (function_epilogue): Likewise.
2705 * doc/tm.texi: Regenerate.
2706 * final.c (final_start_function): Update call to function_prologue.
2707 (final_end_function): Update call to function_epilogue.
2708 (default_function_pro_epilogue): Remove frame size argument.
2709 * output.h (default_function_pro_epilogue): Likewise.
2710 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
2711 (arm_output_function_prologue): Likewise.
2712 * config/frv/frv.c (frv_function_prologue): Likewise.
2713 (frv_function_epilogue): Likewise.
2714 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
2715 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
2716 (ia64_output_function_epilogue): Likewise.
2717 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
2718 (m32r_output_function_epilogue): Likewise.
2719 * config/microblaze/microblaze.c (microblaze_function_prologue)
2720 (microblaze_function_epilogue): Likewise.
2721 * config/mips/mips.c (mips_output_function_prologue): Likewise.
2722 (mips_output_function_epilogue): Likewise.
2723 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
2724 (mmix_target_asm_function_epilogue): Likewise.
2725 * config/msp430/msp430.c (msp430_start_function): Likewise.
2726 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
2727 (nds32_asm_function_epilogue): Likewise.
2728 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
2729 * config/pa/pa.c (pa_output_function_prologue): Likewise.
2730 (pa_output_function_epilogue): Likewise.
2731 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
2732 (rs6000_output_function_epilogue): Likewise.
2733 * config/rl78/rl78.c (rl78_start_function): Likewise.
2734 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
2735 (rs6000_output_function_epilogue): Likewise.
2736 * config/rx/rx.c (rx_output_function_prologue): Likewise.
2737 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
2738 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
2739 (sparc_asm_function_epilogue): Likewise.
2741 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2743 * tree.h (type_has_mode_precision_p): New function.
2744 * convert.c (convert_to_integer_1): Use it.
2745 * expr.c (expand_expr_real_2): Likewise.
2746 (expand_expr_real_1): Likewise.
2747 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
2748 * match.pd: Likewise.
2749 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
2750 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
2751 * tree-tailcall.c (process_assignment): Likewise.
2752 * tree-vect-loop.c (vectorizable_reduction): Likewise.
2753 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
2754 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
2755 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2756 (vectorizable_assignment): Likewise.
2757 (vectorizable_shift): Likewise.
2758 (vectorizable_operation): Likewise.
2759 * tree-vrp.c (register_edge_assert_for_2): Likewise.
2761 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
2763 * match.pd: Add pow (C, x) simplification.
2765 2017-08-21 Richard Biener <rguenther@suse.de>
2767 PR tree-optimization/81900
2768 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
2769 for blocks with abnormal predecessors.
2770 (compute_antic): Do not set visited flag prematurely.
2772 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
2775 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
2777 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2779 * stor-layout.h (vector_type_mode): Move to...
2780 * tree.h (vector_type_mode): ...here.
2781 * stor-layout.c (vector_type_mode): Move to...
2782 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
2784 2017-08-21 Richard Biener <rguenther@suse.de>
2786 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
2787 register_external_die hooks.
2788 (debug_false_tree_charstarstar_uhwistar): Declare.
2789 (debug_nothing_tree_charstar_uhwi): Likewise.
2790 * debug.c (do_nothing_debug_hooks): Adjust.
2791 (debug_false_tree_charstarstar_uhwistar): New do nothing.
2792 (debug_nothing_tree_charstar_uhwi): Likewise.
2793 * dbxout.c (dbx_debug_hooks): Adjust.
2794 (xcoff_debug_hooks): Likewise.
2795 * sdbout.c (sdb_debug_hooks): Likewise.
2796 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
2797 * dwarf2out.c (macinfo_label_base): New global.
2798 (dwarf2out_register_external_die): New function for the
2799 register_external_die hook.
2800 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
2801 (dwarf2_debug_hooks): Use them.
2802 (dwarf2_lineno_debug_hooks): Adjust.
2803 (struct die_struct): Add with_offset flag.
2804 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
2805 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
2806 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
2807 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
2808 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
2809 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
2810 defining section names for the early LTO debug variants.
2811 (reset_indirect_string): New helper.
2812 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
2813 (print_dw_val): Add support for offsetted symbol references.
2814 (get_ultimate_context): Split out from is_cxx.
2815 (is_cxx): Use get_ultimate_context.
2816 (is_fortran): Add decl overload.
2817 (compute_comp_unit_symbol): Split out worker from
2818 compute_section_prefix.
2819 (compute_section_prefix): Call compute_comp_unit_symbol and
2820 set comdat_type_p here.
2821 (output_die): Skip DIE symbol output for the LTO added one.
2822 Handle DIE symbol references with offset.
2823 (output_comp_unit): Guard section name mangling properly.
2824 For LTO debug sections emit a symbol at the section beginning
2825 which we use to refer to its DIEs.
2826 (add_abstract_origin_attribute): For DIEs registered via
2827 dwarf2out_register_external_die directly refer to the early
2828 DIE rather than indirectly through the shadow one we created.
2829 Remove obsolete call to dwarf2out_abstract_function for
2830 non-function/block origins.
2831 (gen_array_type_die): When generating early LTO debug do
2832 not emit DW_AT_string_length.
2833 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
2834 late when in LTO. As suggested place a gcc_unreachable for
2835 the DECL_ABSTRACT_P case.
2836 (gen_subprogram_die): Avoid another specification DIE
2837 for early built declarations/definitions for the late LTO case.
2838 (gen_variable_die): Add type references for late duplicated VLA dies
2840 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
2841 we have the abstract instance already.
2842 (process_scope_var): Adjust decl DIE contexts in LTO which
2843 first puts them in limbo.
2844 (gen_decl_die): Do not generate type DIEs late apart from
2845 types for VLAs or for decls we do not yet have a DIE. Do not
2846 call dwarf2out_abstract_function late.
2847 (dwarf2out_early_global_decl): Make sure to create DIEs
2848 for abstract instances of a decl first.
2849 (dwarf2out_late_global_decl): Adjust comment.
2850 (output_macinfo_op): With multiple macro sections use
2851 macinfo_label_base to distinguish labels.
2852 (output_macinfo): Likewise. Update macinfo_label_base.
2853 Pass in the line info label.
2854 (note_variable_value_in_expr): When generating LTO resolve
2855 all variable values here by generating DIEs as needed.
2856 (init_sections_and_labels): Add early LTO debug flag parameter
2857 and generate different sections and names if set. Add generation
2858 counter for the labels so we can have multiple of them.
2859 (reset_dies): Helper to allow DIEs to be output multiple times.
2860 (dwarf2out_finish): When outputting DIEs to the fat part of an
2861 LTO object first reset DIEs.
2862 (dwarf2out_early_finish): Output early DIEs when generating LTO.
2863 (modified_type_die): Check for decl_ultimate_origin being self
2865 (gen_type_die_with_usage): Likewise.
2866 (gen_typedef_die): Allow decl_ultimate_origin being self.
2867 (set_decl_abstract_flags): Remove.
2868 (set_block_abstract_flags): Likewise.
2869 (dwarf2out_abstract_function): Treat the early generated DIEs
2870 as the abstract copy and only add DW_AT_inline and
2871 DW_AT_artificial here and call set_decl_origin_self.
2872 If the DIE has an abstract origin don't do anything.
2873 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
2874 if we have none yet (Go fails to build one, PR78628).
2875 (variably_modified_type_p): Prevent endless recursion for Ada
2876 cyclic pointer types.
2877 * lto-streamer-in.c: Include debug.h.
2878 (dref_queue): New global.
2879 (lto_read_tree_1): Stream in DIE references.
2880 (lto_input_tree): Register DIE references.
2881 (input_function): Stream DECL_DEBUG_ARGS.
2882 * lto-streamer-out.c: Include debug.h.
2883 (lto_write_tree_1): Output DIE references.
2884 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
2885 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
2886 (output_function): Stream DECL_DEBUG_ARGS.
2887 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
2888 Stream DECL_ABSTRACT_ORIGIN.
2889 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
2890 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
2891 DECL_CONTEXT for file-scope decls.
2892 * lto-streamer.h (struct dref_entry): Declare.
2893 (dref_queue): Likewise.
2894 * cfgexpand.c (pass_expand::execute): Do not call the
2895 outlining_inline_function hook here.
2896 * lto-wrapper.c (debug_obj): New global.
2897 (tool_cleanup): Unlink it if required.
2898 (debug_objcopy): New function.
2899 (run_gcc): Handle early debug sections in the IL files by
2900 extracting them to separate files, partially linkin them and
2901 feeding the result back as result to the linker.
2902 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
2903 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
2904 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
2905 sections into a separate segment.
2906 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
2908 (darwin_asm_dwarf_section): Likewise.
2909 (darwin_asm_output_dwarf_offset): Likewise.
2910 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
2912 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2913 Alan Hayward <alan.hayward@arm.com>
2914 David Sherwood <david.sherwood@arm.com>
2916 * read-md.h (md_reader::record_potential_iterator_use): Replace
2917 pointer argument with an rtx and an index.
2918 * read-rtl.c (iterator_group::apply_iterator): Likewise.
2919 (apply_mode_iterator): Likewise.
2920 (apply_code_iterator): Likewise.
2921 (apply_int_iterator): Likewise.
2922 (apply_subst_iterator): Likewise.
2923 (record_iterator_use): Likewise.
2924 (record_attribute_use): Likewise.
2925 (md_reader::record_potential_iterator_use): Likewise. Update calls
2926 to record_iterator_use and apply_iterator.
2927 (iterator_use): Replace ptr with x and index.
2928 (attribute_use): Likewise.
2929 (apply_attribute_uses): Update calls to apply_iterator.
2930 (apply_iterators): Likewise. Update initialization of iterator_use.
2931 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
2932 and record_potential_iterator_use.
2933 (rtx_reader::read_rtx_operand): Likewise.
2935 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2936 Alan Hayward <alan.hayward@arm.com>
2937 David Sherwood <david.sherwood@arm.com>
2939 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
2942 2017-08-21 Richard Biener <rguenther@suse.de>
2945 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
2946 at struct end conservatively when comparing common bases.
2948 2017-08-21 Richard Biener <rguenther@suse.de>
2950 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
2951 (mem_ref_in_stmt): Remove.
2952 (determine_max_movement): Use ref index to get at the reference.
2953 (invariantness_dom_walker::before_dom_children): Deal with
2954 lim data already initialized.
2955 (gather_mem_refs_stmt): Initialize lim data and record ref index.
2957 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
2959 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
2960 (TARGET_ISA_ROUND): Ditto.
2961 (TARGET_ROUND): Ditto.
2962 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
2963 * config/i386/i386.md: Ditto.
2964 * config/i386/sse.md: Ditto.
2965 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
2966 with OPTION_MASK_ISA_SSE4_1.
2968 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
2971 * doc/extend.texi (x86 Built-in Functions): Correct the name of
2972 __builtin_ia32_lzcnt_u16.
2974 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
2977 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
2978 (rs6000_set_current_function): Rewrite function to use it.
2980 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
2983 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
2984 and TYPE_WARN_IF_NOT_ALIGN.
2985 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
2986 (handle_warn_if_not_align): New.
2987 (place_union_field): Call handle_warn_if_not_align.
2988 (place_field): Call handle_warn_if_not_align.
2989 Copy TYPE_WARN_IF_NOT_ALIGN.
2990 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
2991 (layout_type): Likewise.
2992 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
2994 (tree_decl_common): Add warn_if_not_align.
2995 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
2996 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
2997 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
2998 (DECL_WARN_IF_NOT_ALIGN): Likewise.
2999 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
3000 * doc/extend.texi: Document warn_if_not_aligned attribute.
3001 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
3003 2017-08-17 Martin Liska <mliska@suse.cz>
3006 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
3007 (get_data_dependence): Use it as pointer type.
3008 (distribute_loop): Likewise.
3010 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3012 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
3013 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
3014 (p8_vmrgow_<mode>_direct): New define_insn.
3015 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
3016 handle endianness for vmrgew and vmrgow permute patterns.
3018 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
3020 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
3021 * config/rs6000/rs6000-cpus.def: Remove comment.
3022 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
3023 (POWERPC_MASKS): Likewise.
3024 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
3025 use of TARGET_VSX_TIMODE.
3026 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
3027 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
3028 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
3029 (rs6000_option_override_internal): Remove dead code.
3030 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
3031 (rs6000_legitimize_reload_address): Likewise.
3032 (rs6000_legitimate_address_p): Likewise.
3033 (rs6000_opt_masks): Delete "vsx-timode".
3034 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
3035 from function comment.
3036 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
3037 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
3038 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
3040 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
3041 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
3042 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
3045 (define_peephole2): Likewise.
3047 2017-08-17 Martin Sebor <msebor@redhat.com>
3050 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
3051 past the end of an array.
3052 (test_pp_format): Add test cases.
3054 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
3056 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
3057 missing ECF_NOTHROW flags.
3059 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
3062 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
3063 operands residing in integer registers.
3064 (*vsx_le_perm_load_<mode>): Likewise.
3065 (*vsx_le_perm_store_<mode>): Likewise.
3066 (define_peephole2): Add peepholes to optimize the above.
3068 2017-08-17 Marek Polacek <polacek@redhat.com>
3071 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
3072 to mimic what shorten_compare did. Change the return type to bool.
3073 (fold_cond_expr_with_comparison): Update call to
3074 operand_equal_for_comparison_p.
3075 (fold_ternary_loc): Likewise.
3077 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
3079 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
3081 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
3083 2017-08-17 Richard Biener <rguenther@suse.de>
3085 * tree-ssa-structalias.c (solve_graph): When propagating
3086 to successors update the graphs succ edges and avoid duplicate work.
3088 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
3091 * config/i386/i386.c (ix86_option_override_internal): Save target
3092 specific options after ix86_stack_protector_guard_reg was changed.
3094 2017-08-17 Richard Biener <rguenther@suse.de>
3096 PR tree-optimization/81827
3097 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
3098 (new_var_info): Initialize it conservatively.
3099 (get_call_vi): Mark register vars.
3100 (new_scalar_tmp_constraint_exp): Likewise.
3101 (handle_rhs_call): Likewise.
3102 (handle_const_call): Likewise.
3103 (create_function_info_for): Likewise.
3104 (solve_constraints): Sort varinfos to separate register from
3105 non-register vars to pack points-to solution bitmaps during
3108 2017-08-17 Marek Polacek <polacek@redhat.com>
3110 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
3112 2017-08-17 Richard Biener <rguenther@suse.de>
3114 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
3115 to true when overflow is undefined and we saturated the result.
3117 2017-08-17 Alan Modra <amodra@gmail.com>
3120 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
3121 Don't use store multiple if only one reg needs saving.
3122 (interesting_frame_related_regno): New function.
3123 (rs6000_frame_related): Don't emit frame info for regs that
3125 (rs6000_emit_epilogue): Likewise.
3127 2017-08-16 Nathan Sidwell <nathan@acm.org>
3129 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
3130 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
3131 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
3132 * tree.c (free_lang_data_in_type): Use else-if chain. Always
3133 clear TYPE_LANG_1. Remove obsolete member-function stripping.
3134 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
3135 (verify_type): Adjust for TYPE_BINFO move.
3136 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
3137 process TYPE_BINFO directly.
3138 (hash_tree): Likewise.
3139 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
3141 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
3144 2017-08-16 David Malcolm <dmalcolm@redhat.com>
3146 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
3148 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
3151 * config/i386/i386.md (*anddi_1_btr): Change predicates of
3152 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
3153 Add ix86_binary_operator_ok to insn constraint.
3154 (*iordi_1_bts): Ditto.
3155 (*xordi_1_btc): Ditto.
3156 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
3157 Update corresponding peephole2 pattern.
3161 2017-08-16 Bin Cheng <bin.cheng@arm.com>
3163 PR tree-optimization/81832
3164 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
3165 copy loop header which has IFN_LOOP_DIST_ALIAS call.
3167 2017-08-16 Marek Polacek <polacek@redhat.com>
3170 * fold-const.c (fold_indirect_ref_1): Restore original behavior
3171 regarding size_zero_node.
3173 2017-08-16 Martin Liska <mliska@suse.cz>
3176 * config.gcc: Respect previously set extra_objs in case
3179 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
3181 PR tree-optimization/81835
3182 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
3183 the phi SSA_NAME. Check that the condition in a COND_EXPR does
3184 not depend on the phi.
3186 2017-08-16 Alan Modra <amodra@gmail.com>
3188 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
3191 2017-08-16 Alan Modra <amodra@gmail.com>
3193 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
3194 (save_reg_p): ..into this. Update all callers.
3195 (first_reg_to_save): Simplify.
3197 2017-08-16 Alan Modra <amodra@gmail.com>
3199 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
3202 2017-08-15 Joseph Myers <joseph@codesourcery.com>
3206 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
3207 constant count if that count is less than 32.
3209 2017-08-15 Nathan Sidwell <nathan@acm.org>
3211 * gcc.c (execute): Emit friendlier message if inferior is killed
3212 by an external cause.
3214 2017-08-15 Richard Biener <rguenther@suse.de>
3216 PR tree-optimization/81790
3217 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
3218 CONSTRUCTORs from simplifying and VN.
3220 2017-08-14 Martin Sebor <msebor@redhat.com>
3222 * builtin-attrs.def: Add comments.
3224 2017-08-14 Martin Sebor <msebor@redhat.com>
3227 * doc/extend.texi (attribute nonstring): Document new attribute.
3229 2017-08-14 Martin Sebor <msebor@redhat.com>
3232 * tree-diagnostic.c (default_tree_printer): Handle %G.
3233 * gimple-pretty-print.h (percent_G_format): Declare new function.
3234 * gimple-pretty-print.c (percent_G_format): Define.
3235 * tree-pretty-print.c (percent_K_format): Add argument.
3237 2017-08-14 Martin Sebor <msebor@redhat.com>
3239 PR translation/79998
3240 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
3241 Remove a stray space.
3243 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
3246 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
3247 (*iordi_1_bts): Ditto.
3248 (*xordi_1_btc): Ditto.
3250 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3253 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
3254 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3256 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
3257 quoted strings, and make more translator-friendly.
3258 (darwin_rs6000_override_options): Likewise.
3259 (rs6000_option_override_internal): Likewise.
3260 (rs6000_return_in_memory): Fix overlong line.
3261 (init_cmulative_args): Use quoted strings, and make more
3262 translator-friendly.
3263 (rs6000_pass_by_reference): Fix overlong line.
3264 (def_builtin): Use quoted strings.
3265 (altivec_expand_predicate_builtin): Use quoted strings, and make
3266 more translator-friendly.
3267 (htm_expand_builtin): Use quoted strings.
3268 (cpu_expand_builtin): Use quoted strings, and make more
3269 translator-friendly.
3270 (altivec_expand_builtin): Likewise.
3271 (paired_expand_predicate_builtin): Likewise.
3272 (rs6000_invalid_builtin): Likewise.
3273 (builtin_function_type): Use quoted strings.
3274 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
3275 more translator-friendly.
3276 (rs6000_trampoline_init): Likewise.
3277 (rs6000_handle_altivec_attribute): Likewise.
3278 (rs6000_inner_target_options): Use quoted strings.
3279 (rs6000_disable_incompatible_switches): Likewise.
3280 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
3281 strings, and make more translator-friendly.
3282 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
3284 2017-08-14 Bin Cheng <bin.cheng@arm.com>
3286 PR tree-optimization/81799
3287 * tree-loop-distribution.c (version_loop_by_alias_check): Force
3288 cond_expr to simple gimple operand.
3290 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
3293 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
3295 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
3299 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
3300 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
3301 TARGET_GASISR_PROLOGUES.
3302 * config/avr/avr.c (avr_option_override): Same.
3303 (avr_pass_pre_proep::execute): Same.
3305 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
3308 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
3309 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
3311 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
3313 * config/i386/i386.md (*load_tp_<mode>): Redefine as
3314 define_insn_and_split. Split to a memory load from 0 in
3315 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
3316 using PTR mode iterator.
3317 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
3318 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
3319 (*add_tp_<mode>): Redefine as define_insn_and_split.
3320 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
3321 address space. Merge with *add_tp_x32 using PTR mode iterator.
3322 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
3323 Split to an add with a memory load from 0 in
3324 DEFAULT_TLS_SEG_REG address space.
3326 2017-08-12 Andrew Pinski <apinski@cavium.com>
3328 * config/aarch64/aarch64-option-extensions.def (rdma):
3329 Fix feature string to what Linux prints out in /proc/cpuinfo.
3331 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
3334 * dwarf2out.c (modified_type_die): For C typedef types that have
3335 an ultimate origin, process the ultimate origin instead of the
3337 (gen_typedef_die): Assert that input DECLs have no ultimate
3339 (gen_type_die_with_usage): For typedef variants that have an
3340 ultimate origin, just call gen_decl_die on the original DECL.
3341 (process_scope_var): Avoid creating DIEs for local typedefs and
3342 concrete static variables.
3344 2017-08-12 Alan Modra <amodra@gmail.com>
3348 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
3349 match gnu-user.h startfile.
3350 (ENDFILE_LINUX_SPEC): Similarly.
3352 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
3355 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
3357 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
3359 2017-08-11 Tamar Christina <tamar.christina@arm.com>
3360 * config/aarch64/aarch64.md (mov<mode>): Change.
3361 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
3362 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
3363 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
3365 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
3367 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
3368 for storage order barriers.
3370 2017-08-11 Martin Liska <mliska@suse.cz>
3373 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
3374 variables of void type.
3376 2017-08-11 Martin Liska <mliska@suse.cz>
3378 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
3379 TARGET_SUPPORTS_ALIASES.
3380 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
3381 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
3382 (optimize_weakref): Likewise.
3383 * symtab.c (symtab_node::noninterposable_alias): Likewise.
3384 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
3385 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
3387 2017-08-11 Martin Liska <mliska@suse.cz>
3390 * config/i386/i386.c (make_resolver_func): Do complete
3391 refactoring of the function.
3393 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
3396 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
3397 * config/i386/i386.c (ix86_stack_protect_guard): Use
3398 ix86_stack_protect_guard_symbol_str to generate varible declaration.
3399 * doc/invoke.texi (x86 Options): Document
3400 -mstack-protector-guard-symbol= option.
3402 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
3404 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
3405 * config/i386/i386.c (ix86_split_stack_guard): New function.
3406 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
3407 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
3408 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
3409 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
3410 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
3411 (split_stack_space_check): Call ix86_split_stack_guard.
3413 2017-08-10 Martin Sebor <msebor@redhat.com>
3415 * print-tree.c (print_node): Print location using the established
3417 Replace spaces with colons.
3418 (debug_raw, debug): Ditto.
3420 2017-08-10 Martin Sebor <msebor@redhat.com>
3423 * pretty-print.c (pp_format): Correct the handling of %s precision.
3425 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
3428 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
3430 (ix86_finalize_stack_frame_flags): This. Also clear
3431 frame_pointer_needed if -fno-omit-frame-pointer is used without
3433 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
3434 with ix86_finalize_stack_frame_flags.
3435 (ix86_expand_epilogue): Likewise.
3436 (ix86_expand_split_stack_prologue): Likewise.
3437 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
3439 2017-08-10 Martin Liska <mliska@suse.cz>
3442 * c-attribs.c (handle_target_attribute):
3443 Report warning for an empty string argument of target attribute.
3445 2017-08-09 Jakub Jelinek <jakub@redhat.com>
3448 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
3450 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
3451 or DECL_NONLOCAL labels.
3452 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
3453 or DECL_NONLOCAL labels here.
3455 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
3457 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
3458 to indicate when early gimple folding has been disabled.
3459 (rs6000_gimple_fold_builtin): Add debug content.
3460 (rs6000_invalid_builtin): Fix whitespace.
3461 (rs6000_expand_builtin): Fix whitespace.
3462 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
3464 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
3467 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
3468 SAVE_MULTIPLE if not all the registers that saves, should be saved.
3470 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
3472 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
3473 (qdf24xx): Likewise.
3474 * config/aarch64/aarch64.md: Include falkor.md.
3475 * config/aarch64/falkor.md: New.
3477 2017-08-09 Marek Polacek <polacek@redhat.com>
3480 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
3481 * diagnostic.c (emit_diagnostic): Add a comment.
3482 (emit_diagnostic_valist): New function.
3484 2017-08-09 Marek Polacek <polacek@redhat.com>
3487 * input.c (make_location): New overload.
3488 * input.h (make_location): Declare.
3490 2017-08-08 Alan Modra <amodra@gmail.com>
3491 H.J. Lu <hongjiu.lu@intel.com>
3494 * gcc.c (NO_PIE_SPEC): Delete.
3495 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
3497 (LINK_PIE_SPEC): ..to here.
3498 (LINK_COMMAND_SPEC): Support -no-pie.
3499 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
3500 chain of crtbegin*.o selection, update for PIE_SPEC changes and
3502 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
3503 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
3504 (ENDFILE_CRTEND_SPEC): Similarly.
3506 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
3509 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
3510 (mstack-protector-guard-offset=): Ditto.
3511 * config/i386/i386.c (ix86_option_override): Handle
3512 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
3514 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
3515 ix86_stack_protect_guard_offset variables.
3516 (TARGET_STACK_PROTECT_GUARD): Always define.
3517 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
3518 and -mstack-protector-guard-offset= options.
3520 2017-08-08 Bin Cheng <bin.cheng@arm.com>
3522 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
3523 boundary case for the last candidate.
3525 2017-08-08 Bin Cheng <bin.cheng@arm.com>
3527 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
3528 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
3530 2017-08-08 Tamar Christina <tamar.christina@arm.com>
3533 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
3534 * config/aarch64/aarch64-builtins.c
3535 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
3536 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
3537 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
3539 2017-08-08 Tamar Christina <tamar.christina@arm.com>
3540 Andrew Pinski <pinskia@gmail.com>
3543 * internal-fn.def (XORSIGN): New.
3544 * optabs.def (xorsign_optab): New.
3545 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
3546 (convert_expand_mult_copysign): New.
3547 (pass_optimize_widening_mul::execute): Call
3548 convert_expand_mult_copysign.
3550 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3552 PR tree-optimization/81354
3553 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
3554 Insert on edges rather than explicitly creating landing pads.
3555 (analyze_candidates_and_replace): Commit edge inserts.
3557 2017-08-08 Richard Biener <rguenther@suse.de>
3560 * tree-ssa-loop-niter.c: Include tree-dfa.h.
3561 (expand_simple_operations): Also look through ADDR_EXPRs with
3562 MEM_REF bases treating them as POINTER_PLUS_EXPR.
3564 2017-08-08 Richard Biener <rguenther@suse.de>
3566 PR tree-optimization/81723
3567 * tree-vect-slp.c (struct bst_traits): New hash traits.
3568 (bst_fail): New global.
3569 (vect_build_slp_tree_2): New worker, split out from ...
3570 (vect_build_slp_tree): ... this now wrapping it with using
3571 bst_fail set to cache SLP tree build fails. Properly handle
3573 (vect_analyze_slp_instance): Allocate and free bst_fail.
3575 2017-08-08 Martin Liska <mliska@suse.cz>
3578 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
3579 LABEL_DECLs that can be from a different function.
3581 2017-08-08 Bin Cheng <bin.cheng@arm.com>
3583 PR tree-optimization/81744
3584 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
3585 loop's number of iterations.
3587 2017-08-08 Martin Liska <mliska@suse.cz>
3589 * asan.c: Include header files.
3590 * attribs.c (build_decl_attribute_variant): New function moved
3592 (build_type_attribute_qual_variant): Likewise.
3593 (cmp_attrib_identifiers): Likewise.
3594 (simple_cst_list_equal): Likewise.
3595 (omp_declare_simd_clauses_equal): Likewise.
3596 (attribute_value_equal): Likewise.
3597 (comp_type_attributes): Likewise.
3598 (build_type_attribute_variant): Likewise.
3599 (lookup_ident_attribute): Likewise.
3600 (remove_attribute): Likewise.
3601 (merge_attributes): Likewise.
3602 (merge_type_attributes): Likewise.
3603 (merge_decl_attributes): Likewise.
3604 (merge_dllimport_decl_attributes): Likewise.
3605 (handle_dll_attribute): Likewise.
3606 (attribute_list_equal): Likewise.
3607 (attribute_list_contained): Likewise.
3608 * attribs.h (lookup_attribute): New function moved from tree.[ch].
3609 (lookup_attribute_by_prefix): Likewise.
3610 * bb-reorder.c: Include header files.
3611 * builtins.c: Likewise.
3612 * calls.c: Likewise.
3613 * cfgexpand.c: Likewise.
3614 * cgraph.c: Likewise.
3615 * cgraphunit.c: Likewise.
3616 * convert.c: Likewise.
3617 * dwarf2out.c: Likewise.
3618 * final.c: Likewise.
3619 * fold-const.c: Likewise.
3620 * function.c: Likewise.
3621 * gimple-expr.c: Likewise.
3622 * gimple-fold.c: Likewise.
3623 * gimple-pretty-print.c: Likewise.
3624 * gimple.c: Likewise.
3625 * gimplify.c: Likewise.
3626 * hsa-common.c: Likewise.
3627 * hsa-gen.c: Likewise.
3628 * internal-fn.c: Likewise.
3629 * ipa-chkp.c: Likewise.
3630 * ipa-cp.c: Likewise.
3631 * ipa-devirt.c: Likewise.
3632 * ipa-fnsummary.c: Likewise.
3633 * ipa-inline.c: Likewise.
3634 * ipa-visibility.c: Likewise.
3636 * lto-cgraph.c: Likewise.
3637 * omp-expand.c: Likewise.
3638 * omp-general.c: Likewise.
3639 * omp-low.c: Likewise.
3640 * omp-offload.c: Likewise.
3641 * omp-simd-clone.c: Likewise.
3642 * opts-global.c: Likewise.
3643 * passes.c: Likewise.
3644 * predict.c: Likewise.
3645 * sancov.c: Likewise.
3646 * sanopt.c: Likewise.
3647 * symtab.c: Likewise.
3648 * toplev.c: Likewise.
3649 * trans-mem.c: Likewise.
3650 * tree-chkp.c: Likewise.
3651 * tree-eh.c: Likewise.
3652 * tree-into-ssa.c: Likewise.
3653 * tree-object-size.c: Likewise.
3654 * tree-parloops.c: Likewise.
3655 * tree-profile.c: Likewise.
3656 * tree-ssa-ccp.c: Likewise.
3657 * tree-ssa-live.c: Likewise.
3658 * tree-ssa-loop.c: Likewise.
3659 * tree-ssa-sccvn.c: Likewise.
3660 * tree-ssa-structalias.c: Likewise.
3661 * tree-ssa.c: Likewise.
3662 * tree-streamer-in.c: Likewise.
3663 * tree-vectorizer.c: Likewise.
3664 * tree-vrp.c: Likewise.
3666 * ubsan.c: Likewise.
3667 * varasm.c: Likewise.
3668 * varpool.c: Likewise.
3669 * tree.c: Remove functions moved to attribs.[ch].
3671 * config/aarch64/aarch64.c: Add attrs.h header file.
3672 * config/alpha/alpha.c: Likewise.
3673 * config/arc/arc.c: Likewise.
3674 * config/arm/arm.c: Likewise.
3675 * config/avr/avr.c: Likewise.
3676 * config/bfin/bfin.c: Likewise.
3677 * config/c6x/c6x.c: Likewise.
3678 * config/cr16/cr16.c: Likewise.
3679 * config/cris/cris.c: Likewise.
3680 * config/darwin.c: Likewise.
3681 * config/epiphany/epiphany.c: Likewise.
3682 * config/fr30/fr30.c: Likewise.
3683 * config/frv/frv.c: Likewise.
3684 * config/ft32/ft32.c: Likewise.
3685 * config/h8300/h8300.c: Likewise.
3686 * config/i386/winnt.c: Likewise.
3687 * config/ia64/ia64.c: Likewise.
3688 * config/iq2000/iq2000.c: Likewise.
3689 * config/lm32/lm32.c: Likewise.
3690 * config/m32c/m32c.c: Likewise.
3691 * config/m32r/m32r.c: Likewise.
3692 * config/m68k/m68k.c: Likewise.
3693 * config/mcore/mcore.c: Likewise.
3694 * config/microblaze/microblaze.c: Likewise.
3695 * config/mips/mips.c: Likewise.
3696 * config/mmix/mmix.c: Likewise.
3697 * config/mn10300/mn10300.c: Likewise.
3698 * config/moxie/moxie.c: Likewise.
3699 * config/msp430/msp430.c: Likewise.
3700 * config/nds32/nds32-isr.c: Likewise.
3701 * config/nds32/nds32.c: Likewise.
3702 * config/nios2/nios2.c: Likewise.
3703 * config/nvptx/nvptx.c: Likewise.
3704 * config/pa/pa.c: Likewise.
3705 * config/pdp11/pdp11.c: Likewise.
3706 * config/powerpcspe/powerpcspe.c: Likewise.
3707 * config/riscv/riscv.c: Likewise.
3708 * config/rl78/rl78.c: Likewise.
3709 * config/rx/rx.c: Likewise.
3710 * config/s390/s390.c: Likewise.
3711 * config/sh/sh.c: Likewise.
3712 * config/sol2.c: Likewise.
3713 * config/sparc/sparc.c: Likewise.
3714 * config/spu/spu.c: Likewise.
3715 * config/stormy16/stormy16.c: Likewise.
3716 * config/tilegx/tilegx.c: Likewise.
3717 * config/tilepro/tilepro.c: Likewise.
3718 * config/v850/v850.c: Likewise.
3719 * config/vax/vax.c: Likewise.
3720 * config/visium/visium.c: Likewise.
3721 * config/xtensa/xtensa.c: Likewise.
3723 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
3726 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
3727 constraints since the -mupper-regs-* switches have been
3729 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
3730 into a vector from a double word element that was extracted from
3731 another vector, and eliminate extra XXPERMDI instructions.
3732 (vsx_concat_<mode>_2): Likewise.
3733 (vsx_concat_<mode>_3): Likewise.
3734 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
3735 concat to allow optimizing inserts from previous extracts.
3737 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
3739 * config/i386/i386.c (ix86_stack_protect_guard): Generate
3740 memory reference to a SSP offset in TLS address space.
3741 (ix86_print_operand) <case '@'>: Remove.
3742 (ix86_print_operand_punct_valid_p): Remove '@' code.
3743 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
3745 (stack_tls_protect_set_<mode>): Remove.
3746 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
3747 (stack_tls_protect_test_<mode>): Remove.
3748 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
3750 2017-08-07 Olivier Hainque <hainque@adacore.com>
3753 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
3755 2017-08-07 Douglas Rupp <rupp@adacore.com>
3757 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
3758 variable was referenced as multidir in command.
3760 2017-08-07 Jakub Jelinek <jakub@redhat.com>
3763 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
3766 2017-08-07 Martin Liska <mliska@suse.cz>
3768 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
3769 * config/rl78/rl78.c: Add include of attribs.h.
3770 * config/sh/sh.c: Likewise.
3771 * config/v850/v850.c: Likewise.
3773 2017-08-07 Tom de Vries <tom@codesourcery.com>
3776 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
3778 2017-08-07 Martin Liska <mliska@suse.cz>
3780 * config/mips/mips.c: Include attribs.h.
3782 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
3785 * doc/invoke.texi: Document change in behvaior for -Ofast for
3788 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
3790 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
3792 (aarch64_pop_regs): Likewise.
3793 (aarch64_gen_load_pair): Likewise.
3794 (aarch64_save_callee_saves): Likewise.
3795 (aarch64_restore_callee_saves): Likewise.
3797 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
3799 * config/i386/i386.c: Revert the last change.
3801 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
3804 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
3806 (ix86_finalize_stack_frame_flags): This. Also clear
3807 frame_pointer_needed if -fno-omit-frame-pointer is used without
3809 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
3810 with ix86_finalize_stack_frame_flags.
3811 (ix86_expand_epilogue): Likewise.
3812 (ix86_expand_split_stack_prologue): Likewise.
3814 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
3817 * config/i386/i386.c (get_builtin_code_for_version): Set priority
3818 to P_AES for Westmere.
3820 2017-08-07 Jonathan Yong <10walls@gmail.com>
3822 * config/i386/mingw.opt (fset-stack-executable): Removed.
3823 * config/i386/cygming.opt (fset-stack-executable): Moved
3825 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
3827 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
3829 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
3831 2017-08-07 Marek Polacek <polacek@redhat.com>
3834 * fold-const.c (fold_indirect_ref_1): Check type_domain.
3836 2017-08-07 Martin Liska <mliska@suse.cz>
3838 * attribs.h (canonicalize_attr_name): New function.
3839 (cmp_attribs): Move from c-format.c and adjusted.
3840 (is_attribute_p): Moved from tree.h.
3841 * tree-inline.c: Add new includes.
3842 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
3843 (private_is_attribute_p): Remove.
3844 (private_lookup_attribute): Likewise.
3845 (private_lookup_attribute_by_prefix): Simplify.
3846 (remove_attribute): Use is_attribute_p.
3847 * tree.h: Remove removed declarations.
3849 2017-08-07 Jakub Jelinek <jakub@redhat.com>
3852 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
3853 instead of computing it in the function. Formatting fix.
3854 (expand_case): Don't rely on default_edge being the first edge,
3855 clear it if removing it, pass default_edge to
3856 emit_case_dispatch_table.
3857 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
3860 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
3862 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
3863 insn in the function, emit NOP after the insn.
3865 2017-08-06 Tom de Vries <tom@codesourcery.com>
3867 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
3870 2017-08-06 Tom de Vries <tom@codesourcery.com>
3872 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
3875 2017-08-04 Yury Gribov <tetra2005@gmail.com>
3877 PR tree-optimization/57371
3878 * match.pd: New pattern.
3880 2017-08-04 Marek Polacek <polacek@redhat.com>
3883 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
3884 perform the computation in offset_int.
3886 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3888 PR tree-optimization/81136
3889 * tree-vectorizer.h: Include tree-hash-traits.h.
3890 (vec_base_alignments): New typedef.
3891 (vec_info): Add a base_alignments field.
3892 (vect_record_base_alignments): Declare.
3893 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
3895 (DR_IS_CONDITIONAL_IN_STMT): New macro.
3896 (create_data_ref): Add an is_conditional_in_stmt argument.
3897 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
3898 the is_conditional_in_stmt field.
3899 (data_ref_loc): Add an is_conditional_in_stmt field.
3900 (get_references_in_stmt): Set the is_conditional_in_stmt field.
3901 (find_data_references_in_stmt): Update call to create_data_ref.
3902 (graphite_find_data_references_in_stmt): Likewise.
3903 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
3904 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
3905 (vect_record_base_alignment): New function.
3906 (vect_record_base_alignments): Likewise.
3907 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
3908 for nested statements even if we fail to compute a misalignment.
3909 Use pooled base alignments for unconditional references.
3910 (vect_find_same_alignment_drs): Compare base addresses instead
3912 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
3913 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
3915 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3917 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
3918 Add an explicit name for the enum. Use auto_vec for slp_instances
3920 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
3922 (_bb_vec_info): Add a constructor and destructor.
3923 (vinfo_for_stmt): Return NULL for uids of -1 as well.
3924 (destroy_loop_vec_info): Delete.
3925 (vect_destroy_datarefs): Likewise.
3926 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
3927 (vec_info::vec_info): New function.
3928 (vec_info::~vec_info): Likewise.
3929 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
3930 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
3931 destroy_loop_vec_info.
3932 * tree-vect-loop.c (new_loop_vec_info): Replace with...
3933 (_loop_vec_info::_loop_vec_info): ...this.
3934 (destroy_loop_vec_info): Replace with...
3935 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
3936 the stmt_vec_infos. Leave handling of vec_info information to its
3937 destructor. Remove explicit vector releases.
3938 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
3939 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
3940 * tree-vect-slp.c (new_bb_vec_info): Replace with...
3941 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
3942 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
3943 (destroy_bb_vec_info): Replace with...
3944 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
3945 information to its destructor.
3946 (vect_slp_analyze_bb_1): Use new and delete instead of
3947 new_bb_vec_info and destroy_bb_vec_info.
3948 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
3951 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3953 * tree-data-ref.h (subscript): Add access_fn field.
3954 (data_dependence_relation): Add could_be_independent_p.
3955 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
3956 (same_access_functions): Move to tree-data-ref.c.
3957 * tree-data-ref.c (ref_contains_union_access_p): New function.
3958 (access_fn_component_p): Likewise.
3959 (access_fn_components_comparable_p): Likewise.
3960 (dr_analyze_indices): Add a reference to access_fn_component_p.
3961 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
3963 (constant_access_functions): Likewise.
3964 (add_other_self_distances): Likewise.
3965 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
3966 (initialize_data_dependence_relation): Use XCNEW and remove
3967 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
3968 of access functions that have the same type. Allow the
3969 subsequence to end with different bases in some circumstances.
3970 Record the chosen access functions in SUB_ACCESS_FN.
3971 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
3972 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
3973 (subscript_dependence_tester_1): Likewise dra and drb.
3974 (build_classic_dist_vector): Update calls accordingly.
3975 (subscript_dependence_tester): Likewise.
3976 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
3977 DDR_COULD_BE_INDEPENDENT_P.
3978 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
3979 comp_alias_ddrs instead of may_alias_ddrs.
3980 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
3982 (vect_analyze_data_ref_dependence): Use it if
3983 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
3984 distance vectors if that fails.
3985 (dependence_distance_ge_vf): New function.
3986 (vect_prune_runtime_alias_test_list): Use it. Don't clear
3987 LOOP_VINFO_MAY_ALIAS_DDRS.
3989 2017-08-04 Richard Biener <rguenther@suse.de>
3992 * fold-const.c (fold_binary_loc): Properly restrict
3993 minus_var0 && minus_var1 case when associating undefined overflow
3996 2017-08-04 Tom de Vries <tom@codesourcery.com>
3998 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
4000 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4002 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4003 Don't start diagnostic messages with a capital letter.
4004 * config/rs6000/rs6000.c (rs6000_option_override_internal):
4006 (rs6000_invalid_builtin): Likewise.
4007 (rs6000_trampoline_init): Likewise.
4009 2017-08-03 Jakub Jelinek <jakub@redhat.com>
4012 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
4013 after setting changeable df flags.
4015 2017-08-03 Richard Biener <rguenther@suse.de>
4017 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
4018 up if the use is in USE - X.
4020 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
4022 * toplev.c (dumpfile.h): New include.
4023 (internal_error_reentered): New static function. Use it...
4024 (internal_error_function): ...here to handle reentered internal_error.
4026 2017-08-03 Richard Biener <rguenther@suse.de>
4029 * fold-const.c (split_tree): Add minus_var and minus_con
4030 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
4031 here but always use minus_*.
4032 (associate_trees): Assert we never associate with MINUS_EXPR
4033 and NULL first operand. Do not recurse for PLUS_EXPR operands
4034 when associating as MINUS_EXPR either.
4035 (fold_binary_loc): Track minus_var and minus_con.
4037 2017-08-03 Tom de Vries <tom@codesourcery.com>
4040 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
4041 ACCEL_COMPILER, apply finish_options on
4042 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
4044 2017-08-03 Tom de Vries <tom@codesourcery.com>
4047 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
4048 function_entry_patch_area_size > 0.
4050 2017-08-03 Jakub Jelinek <jakub@redhat.com>
4053 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
4054 instead of 10??LU, perform unit multiplication in wide_int,
4055 don't change alloc_object_size_limit if the limit is larger
4058 PR tree-optimization/81655
4059 PR tree-optimization/81588
4060 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
4061 the case when ranges[i].low and high are 1 for unsigned type with
4065 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
4066 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
4068 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4070 * tree-vrp.h: Add include guard.
4072 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
4075 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
4076 (ud2): New insn pattern.
4077 * config/i386/i386.c (ix86_expand_epilogue):
4078 For naked functions, generate ud2 instead of trap insn.
4080 2017-08-02 Marek Polacek <polacek@redhat.com>
4083 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
4085 2017-08-02 Tom de Vries <tom@codesourcery.com>
4086 Cesar Philippidis <cesar@codesourcery.com>
4088 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
4089 Add missing edge probabilities.
4091 2017-08-02 Tamar Christina <tamar.christina@arm.com>
4093 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
4096 2017-08-02 Jakub Jelinek <jakub@redhat.com>
4099 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
4100 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
4101 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
4102 prologue_seq sequences - if any.
4104 2017-08-02 Richard Biener <rguenther@suse.de>
4106 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
4107 via vectors if supported, integer extracts via punning if supported
4108 or otherwise vector extracts.
4110 2017-08-02 Richard Biener <rguenther@suse.de>
4112 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
4114 (bitmap_insert_into_set): ... this.
4116 2017-08-02 Richard Biener <rguenther@suse.de>
4118 PR tree-optimization/81633
4120 2015-08-17 Alan Hayward <alan.hayward@arm.com>
4122 PR tree-optimization/71752
4123 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
4125 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
4127 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
4128 (machine_function::call_ms2sysv_pad_out): Remove field.
4129 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
4130 (ix86_compute_frame_layout): Likewise.
4132 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
4135 * config/i386/i386.c (ix86_set_func_type): Disallow naked
4136 attribute with interrupt attribute.
4138 2017-08-01 Andrew Pinski <apinski@cavium.com>
4140 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
4141 BIT_INSERT_EXPR's operand 1
4142 to see if the types precision matches.
4144 2017-08-01 Martin Liska <mliska@suse.cz>
4147 * builtins.c (expand_builtin_memcpy_args): Remove.
4148 (expand_builtin_memcpy): Call newly added function
4149 expand_builtin_memory_copy_args.
4150 (expand_builtin_memcpy_with_bounds): Likewise.
4151 (expand_builtin_mempcpy): Remove last argument.
4152 (expand_builtin_mempcpy_with_bounds): Likewise.
4153 (expand_builtin_memory_copy_args): New function created from
4154 expand_builtin_mempcpy_args with small modifications.
4155 (expand_builtin_mempcpy_args): Remove.
4156 (expand_builtin_stpcpy): Remove unused argument.
4157 (expand_builtin): Likewise.
4158 (expand_builtin_with_bounds): Likewise.
4160 2017-08-01 Martin Liska <mliska@suse.cz>
4163 Make mempcpy more optimal (PR middle-end/70140).
4165 2017-08-01 Jakub Jelinek <jakub@redhat.com>
4168 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
4169 __builtin_vec_cmpne verify both arguments are compatible vectors
4170 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
4171 verify arg1_type is a pointer or array type. For __builtin_vec_st,
4172 move computation of aligned to after checking the argument types.
4176 * config/rs6000/vsx.md (vextract_fp_from_shorth,
4177 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
4180 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
4181 Doug Rupp <rupp@adacore.com>
4182 Olivier Hainque <hainque@adacore.com>
4184 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
4185 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
4187 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
4188 for TARGET_OS_CPP_BUILTIN.
4189 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
4190 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
4192 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
4193 passing required abi options to the assembler for EABI configurations.
4194 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
4195 of .text.hot and .text.unlikely sections for kernel modules when
4196 using ARM style exceptions.
4197 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
4198 options. Add EXTRA_CC1_SPEC.
4199 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
4201 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
4203 (ARM_TARGET2_DWARF_FORMAT): Define.
4204 * config/arm/t-vxworks: Adjust multilib control to removal of the
4205 Diab command line options.
4207 2017-08-01 Martin Liska <mliska@suse.cz>
4209 PR gcov-profile/81561
4210 * gcov.c (unblock): Make unblocking safe as we need to preserve
4211 index correspondence of blocks and block_lists.
4213 2017-08-01 Richard Biener <rguenther@suse.de>
4215 PR tree-optimization/81181
4216 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4217 (compute_antic): ... end of iteration here.
4219 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
4221 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
4222 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
4223 (ftree-slp-vectorize): Likewise.
4224 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
4225 can no longer be set independent of flag_tree_loop_vectorize.
4226 * omp-general.c (emp_max_vf): Likewise.
4227 * opts.c (enable_fdo_optimizations): Remove references to
4228 flag_tree_vectorize, these are now implicit.
4229 (common_handle_option): Remove handling for OPT_ftree_vectorize,
4230 and leave it for the options machinery.
4232 2017-08-01 Martin Liska <mliska@suse.cz>
4235 * builtins.c (expand_builtin_memcpy_args): Remove.
4236 (expand_builtin_memcpy): Call newly added function
4237 expand_builtin_memory_copy_args.
4238 (expand_builtin_memcpy_with_bounds): Likewise.
4239 (expand_builtin_mempcpy): Remove last argument.
4240 (expand_builtin_mempcpy_with_bounds): Likewise.
4241 (expand_builtin_memory_copy_args): New function created from
4242 expand_builtin_mempcpy_args with small modifications.
4243 (expand_builtin_mempcpy_args): Remove.
4244 (expand_builtin_stpcpy): Remove unused argument.
4245 (expand_builtin): Likewise.
4246 (expand_builtin_with_bounds): Likewise.
4248 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
4251 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
4252 print "ds:" only for immediates in generic address space.
4254 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
4257 * config/i386/i386.c (ix86_funciton_naked): New prototype.
4258 (ix86_function_ok_for_sibcall): Return false for naked functions.
4260 2017-08-01 Richard Biener <rguenther@suse.de>
4262 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
4263 (compute_antic): Seed worklist with exit block predecessors.
4264 * cfganal.c (dfs_find_deadend): For a cycle return the source
4265 of the edge closing it.
4267 2017-08-01 Tamar Christina <tamar.christina@arm.com>
4269 * config/aarch64/aarch64.c
4270 (aarch64_can_const_movi_rtx_p): Move 0 check.
4272 2017-08-01 Bin Cheng <bin.cheng@arm.com>
4274 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
4275 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
4277 * match.pd: Ditto in address comparison pattern.
4279 2017-08-01 Bin Cheng <bin.cheng@arm.com>
4281 PR tree-optimization/81627
4282 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
4283 closed ssa form for store-store chain.
4285 2017-08-01 Bin Cheng <bin.cheng@arm.com>
4287 PR tree-optimization/81620
4288 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
4289 for store-store chain.
4291 2017-08-01 Jakub Jelinek <jakub@redhat.com>
4293 PR tree-optimization/81588
4294 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
4295 ranges[i].in_p, invert comparison code ccode. For >/>=,
4296 swap rhs1 and rhs2 and comparison code unconditionally,
4297 for </<= don't do that. Don't swap rhs1/rhs2 again if
4298 ranges[i].in_p, instead invert comparison code ccode if
4299 opcode or oe->rank is BIT_IOR_EXPR.
4302 * optabs.def (vec_extract_optab, vec_init_optab): Change from
4303 a direct optab to conversion optab.
4304 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
4305 with GET_MODE_INNER as last argument instead of optab_handler.
4306 * expmed.c (extract_bit_field_1): Likewise. Use vector from
4307 vector extraction if possible and optab is available.
4308 * expr.c (store_constructor): Use convert_optab_handler instead
4309 of optab_handler. Use vector initialization from smaller
4310 vectors if possible and optab is available.
4311 * tree-vect-stmts.c (vectorizable_load): Likewise.
4312 * doc/md.texi (vec_extract, vec_init): Document that the optabs
4314 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
4315 of vec_init from half-sized vectors with the same element mode.
4316 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
4317 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
4318 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
4319 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
4320 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
4321 after mode in gen_vec_extract* calls.
4322 (vec_extract<mode>): Renamed to ...
4323 (vec_extract<mode><ssescalarmodelower>): ... this.
4324 (vec_extract<mode><ssehalfvecmodelower>): New expander.
4325 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
4326 element mode after mode in gen_vec_init* calls.
4327 (VEC_INIT_HALF_MODE): New mode iterator.
4328 (vec_init<mode>): Renamed to ...
4329 (vec_init<mode><ssescalarmodelower>): ... this.
4330 (vec_init<mode><ssehalfvecmodelower>): New expander.
4331 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
4332 (vec_extractv2sfsf): ... this.
4333 (vec_initv2sf): Renamed to ...
4334 (vec_initv2sfsf): ... this.
4335 (vec_extractv2si): Renamed to ...
4336 (vec_extractv2sisi): ... this.
4337 (vec_initv2si): Renamed to ...
4338 (vec_initv2sisi): ... this.
4339 (vec_extractv4hi): Renamed to ...
4340 (vec_extractv4hihi): ... this.
4341 (vec_initv4hi): Renamed to ...
4342 (vec_initv4hihi): ... this.
4343 (vec_extractv8qi): Renamed to ...
4344 (vec_extractv8qiqi): ... this.
4345 (vec_initv8qi): Renamed to ...
4346 (vec_initv8qiqi): ... this.
4347 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
4348 (vec_init<mode>): Renamed to ...
4349 (vec_init<mode><VEC_base_l>): ... this.
4350 (vec_extract<mode>): Renamed to ...
4351 (vec_extract<mode><VEC_base_l>): ... this.
4352 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
4353 (vec_initv2sfsf): ... this.
4354 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
4355 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
4356 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
4357 element mode after mode in gen_vec_init* calls.
4358 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
4359 (vec_init<mode><Vel>): ... this.
4360 (vec_extract<mode>): Renamed to ...
4361 (vec_extract<mode><Vel>): ... this.
4362 * config/aarch64/iterators.md (Vel): New mode attribute.
4363 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
4364 Add element mode after mode in gen_vec_extract* calls.
4365 * config/s390/vector.md (non_vec_l): New mode attribute.
4366 (vec_extract<mode>): Renamed to ...
4367 (vec_extract<mode><non_vec_l>): ... this.
4368 (vec_init<mode>): Renamed to ...
4369 (vec_init<mode><non_vec_l>): ... this.
4370 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
4371 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
4373 * config/arm/iterators.md (V_elem_l): New mode attribute.
4374 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
4375 (vec_extract<mode><V_elem_l>): ... this.
4376 (vec_extractv2di): Renamed to ...
4377 (vec_extractv2didi): ... this.
4378 (vec_init<mode>): Renamed to ...
4379 (vec_init<mode><V_elem_l>): ... this.
4380 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
4381 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
4382 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
4383 Add element mode after gen_vec_extract* calls.
4384 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
4385 (vec_init<mode><unitmode>): ... this.
4386 (vec_extract<mode>): Renamed to ...
4387 (vec_extract<mode><unitmode>): ... this.
4388 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
4389 (vec_init<mode><unitmode>): ... this.
4390 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
4391 (vec_initv2sfsf): ... this.
4392 (vec_extractv2sf): Renamed to ...
4393 (vec_extractv2sfsf): ... this.
4394 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
4395 Add element mode after gen_vec_extract* calls.
4396 * config/mips/mips.md (unitmode): New mode iterator.
4397 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
4398 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
4399 * config/spu/spu.md (inner_l): New mode attribute.
4400 (vec_init<mode>): Renamed to ...
4401 (vec_init<mode><inner_l>): ... this.
4402 (vec_extract<mode>): Renamed to ...
4403 (vec_extract<mode><inner_l>): ... this.
4404 * config/sparc/sparc.md (veltmode): New mode iterator.
4405 (vec_init<VMALL:mode>): Renamed to ...
4406 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
4407 * config/ia64/vect.md (vec_initv2si): Renamed to ...
4408 (vec_initv2sisi): ... this.
4409 (vec_initv2sf): Renamed to ...
4410 (vec_initv2sfsf): ... this.
4411 (vec_extractv2sf): Renamed to ...
4412 (vec_extractv2sfsf): ... this.
4413 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
4414 (vec_init<mode>): Renamed to ...
4415 (vec_init<mode><VEC_base_l>): ... this.
4416 (vec_extract<mode>): Renamed to ...
4417 (vec_extract<mode><VEC_base_l>): ... this.
4418 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
4419 (vec_initv2sfsf): ... this.
4420 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
4421 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
4422 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
4423 gen_vec_init* calls.
4425 2017-08-01 Richard Biener <rguenther@suse.de>
4427 PR tree-optimization/81297
4428 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
4429 TREE_OVERFLOW from INTEGER_CSTs.
4431 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
4433 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
4435 2017-07-31 Carl Love <cel@us.ibm.com>
4437 * config/rs6000/rs6000-c: Add support for built-in functions
4438 vector signed char vec_xl_be (signed long long, signed char *);
4439 vector unsigned char vec_xl_be (signed long long, unsigned char *);
4440 vector signed int vec_xl_be (signed long long, signed int *);
4441 vector unsigned int vec_xl_be (signed long long, unsigned int *);
4442 vector signed long long vec_xl_be (signed long long, signed long long *);
4443 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
4444 vector signed short vec_xl_be (signed long long, signed short *);
4445 vector unsigned short vec_xl_be (signed long long, unsigned short *);
4446 vector double vec_xl_be (signed long long, double *);
4447 vector float vec_xl_be (signed long long, float *);
4448 * config/rs6000/altivec.h (vec_xl_be): Add #define.
4449 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
4450 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
4452 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
4453 (altivec_expand_builtin): Add switch statement to call
4454 altivec_expand_xl_be for each builtin.
4455 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
4456 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
4457 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
4458 __builtin_vsx_le_be_v16qi.
4459 * doc/extend.texi: Update the built-in documentation file for the
4460 new built-in functions.
4462 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
4465 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
4467 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
4469 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4471 * config.gcc: Add z14.
4472 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
4473 CPU model numbers for z13s and z14.
4474 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
4476 * config/s390/s390-opts.h (enum processor_type): Rename
4477 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
4478 * config/s390/s390.c (processor_table): Add field for CPU name to
4479 be passed to Binutils.
4480 (s390_asm_output_machine_for_arch): Use the new field in
4481 processor_table for Binutils.
4482 (s390_expand_builtin): Replace arch12 with z14.
4483 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
4484 (s390_get_sched_attrmask): Likewise.
4485 (s390_get_unit_mask): Likewise.
4486 * config/s390/s390.opt: Add z14 to processor_type enum.
4488 2017-07-31 Martin Jambor <mjambor@suse.cz>
4491 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
4492 regardless of optimization level.
4494 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
4495 Martin Liska <mliska@suse.cz>
4497 * predict.def: Remove old comment and adjust probability.
4498 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
4501 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
4504 * config/i386/i386.c (ix86_function_naked): New function.
4505 (ix86_can_use_return_insn_p): Return false for naked functions.
4506 (ix86_expand_prologue): Skip prologue for naked functions.
4507 (ix86_expand_epilogue): Skip epilogue for naked functions
4508 and emit trap instruction.
4509 (ix86_warn_func_return): New function.
4510 (ix86_attribute_table): Add "naked" attribute specification.
4511 (TARGET_WARN_FUNC_RETURN): Define.
4512 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
4514 2017-07-31 Martin Liska <mliska@suse.cz>
4516 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
4517 (dump_gimple_bb_header): Always dump BB info.
4518 (pp_cfg_jump): Do not append info about BB when dumping a jump.
4520 2017-07-31 Martin Liska <mliska@suse.cz>
4523 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
4524 also with current_function_decl non-null equality.
4526 2017-07-31 Jakub Jelinek <jakub@redhat.com>
4529 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
4530 change type to the element type, instead add eltype variable and
4531 use it where we are interested in the element type.
4533 PR tree-optimization/81603
4534 * ipa-polymorphic-call.c
4535 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
4536 offset arithmetic in offset_int, bail out if the resulting bit offset
4537 doesn't fit into shwi.
4539 2017-07-31 Martin Liska <mliska@suse.cz>
4541 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
4542 (gimplify_save_expr): Fix comment.
4544 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
4547 * config/i386/i386.c (ix86_function_arg): Update arguments for
4549 (ix86_compute_frame_layout): Set the initial stack offset to
4550 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
4551 INCOMING_FRAME_SP_OFFSET.
4552 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
4553 stack before exception handler returns.
4554 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
4555 the 'ERROR_CODE' for exception handler.
4557 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
4559 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
4560 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
4561 (ASM_OUTPUT_REG_POP): Ditto.
4562 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
4563 instead of asm_fprintf to output pure string.
4565 2017-07-29 Jakub Jelinek <jakub@redhat.com>
4567 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
4568 to imported_module_or_decl hook.
4569 (debug_nothing_tree_tree_tree_bool): Remove.
4570 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
4571 * debug.c (do_nothing_debug_hooks): Use
4572 debug_nothing_tree_tree_tree_bool_bool instead of
4573 debug_nothing_tree_tree_tree_bool.
4574 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
4575 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
4576 * sdbout.c (sdb_debug_hooks): Likewise.
4577 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
4578 (gen_namespace_die): Add DW_AT_export_symbols attribute if
4580 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
4581 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
4582 attribute, don't add anything.
4584 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4586 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
4587 (fold_build2_stat_loc): Likewise.
4588 (fold_build3_stat_loc): Likewise.
4589 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
4590 (fold_build1_loc): Remove macro.
4591 (fold_build2_loc): Likewise.
4592 (fold_build3_loc): Likewise.
4594 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4596 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
4597 (gimple_build_debug_bind_source_stat): Likewise.
4598 * gimple.h (gimple_build_debug_bind): Remove macro.
4599 (gimple_build_debug_bind_source): Likewise.
4601 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4603 * bitmap.c (bitmap_alloc): Adjust.
4604 (bitmap_gc_alloc): Likewise.
4605 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
4607 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4609 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
4610 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
4611 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
4612 (bitmap_gc_alloc_stat): Likewise.
4613 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
4615 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4617 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
4618 * rtl.h (shallow_copy_rtx): Remove macro.
4620 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4622 * emit-rtl.c (gen_raw_REG): Adjust.
4623 * gengenrtl.c (gendef): Likewise.
4624 * rtl.c (rtx_alloc_stat): Remove _stat from name.
4625 * rtl.h (rtx_alloc): Remove macro.
4627 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4629 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
4630 (build_tree_list_stat): Likewise.
4631 * tree.h (build_tree_list): Remove macro.
4632 (build_tree_list_vec): Likewise.
4634 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4636 * tree.c (make_vector_stat): Remove _stat from name.
4637 (build_vector_stat): Likewise.
4638 * tree.h (make_vector_stat): Remove macro.
4639 (build_vector_stat): Likewise.
4641 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4643 * tree.h (build_var_debug_value): Remove prototype.
4645 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4647 * tree.c (tree_cons_stat): Remove _stat from name.
4648 * tree.h (tree_cons): Remove macro.
4650 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4652 * tree.c (build_vl_exp_stat): Remove _stat from name.
4653 * tree.h (build_vl_exp): Remove macro.
4655 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4657 * tree.c (build_decl_stat): Remove _stat from name.
4658 * tree.h (build_decl): Remove macro.
4660 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4662 * gimple.c (gimple_build_with_ops_stat): Adjust.
4663 (gimple_alloc_stat): Remove _stat from name.
4664 * gimple.h (gimple_alloc): Remove macro.
4666 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4668 * tree.c (make_tree_vec_stat): Remove _stat from name.
4669 (grow_tree_vec_stat): Likewise.
4670 * tree.h (make_tree_vec_stat): Adjust prototype.
4671 (grow_tree_vec_stat): Likewise.
4672 (make_tree_vec): Remove macro.
4673 (grow_tree_vec): Likewise.
4675 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4677 * fold-const.c (fold_build1_stat_loc): Adjust.
4678 (fold_build2_stat_loc): Likewise.
4679 (fold_build3_stat_loc): Likewise.
4680 * tree.c (build0_stat): Remove _stat from name.
4681 (build1_stat): Likewise.
4682 (build2_stat): Likewise.
4683 (build3_stat): Likewise.
4684 (build4_stat): Likewise.
4685 (build5_stat): Likewise.
4686 * tree.h (build1_loc): Remove macro, and rename _stat function
4688 (build2_loc): Likewise.
4689 (build3_loc): Likewise.
4690 (build4_loc): Likewise.
4691 (build5_loc): Likewise.
4693 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4695 * tree.c (make_int_cst_stat): Remove _stat from name.
4696 * tree.h (make_int_cst_stat): Adjust prototype.
4697 (make_int_cst): Remove macro.
4699 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4701 * tree.c (make_tre_binfo_stat): Remove _stat from name.
4702 * tree.h (make_tree_binfo_stat): Adjust prototype.
4703 (make_tree_binfo): Remove.
4705 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4707 * tree.c (copy_node_stat): Rename to copy_node.
4708 (build_distinct_type_copy): Adjust.
4709 * tree.h (copy_node_stat): Adjust prototype.
4710 (copy_node): Remove macro.
4712 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4714 * tree.c (make_node_stat): rename to make_node.
4715 (build_tree_list_stat): Adjust.
4716 (build0_stat): Likewise.
4717 (build2_stat): Likewise.
4718 (build3_stat): Likewise.
4719 (build4_stat): Likewise.
4720 (build5_stat): Likewise.
4721 (build_decl_stat): Likewise.
4722 * tree.h (make_node_stat): Adjust prototype.
4723 (make_node): remove macro.
4725 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
4727 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
4728 (PPC_FEATURE2_SCV): Likewise.
4729 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
4731 2017-07-28 Tamar Christina <tamar.christina@arm.com>
4733 * config/aarch64/aarch64.c
4734 (aarch64_internal_mov_immediate): Add new special pattern.
4735 * config/aarch64/aarch64.md (*movdi_aarch64):
4736 Add reg/32bit const mov case.
4738 2017-07-28 Tamar Christina <tamar.christina@arm.com>
4739 Richard Sandiford <richard.sandiford@linaro.org>
4741 * config/aarch64/aarch64.md (mov<mode>): Generalize.
4742 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
4743 Add integer and movi cases.
4744 (movi-split-hf-df-sf split, fp16): New.
4745 (enabled): Added TARGET_FP_F16INST.
4746 * config/aarch64/iterators.md (GPF_HF): New.
4747 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
4749 2017-07-28 Tamar Christina <tamar.christina@arm.com>
4751 * config/aarch64/aarch64.c
4752 (aarch64_simd_container_mode): Add prototype.
4753 (aarch64_expand_mov_immediate): Add HI support.
4754 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
4755 (aarch64_can_const_movi_rtx_p): New.
4756 (aarch64_preferred_reload_class):
4757 Remove restrictions of using FP registers for certain SIMD operations.
4758 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
4759 (aarch64_valid_floating_const): Add integer move validation.
4760 (aarch64_simd_imm_scalar_p): Remove.
4761 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
4762 (aarch64_legitimate_constant_p): Expand list of supported cases.
4763 * config/aarch64/aarch64-protos.h
4764 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
4765 (aarch64_reinterpret_float_as_int): New.
4766 (aarch64_simd_imm_scalar_p): Remove.
4767 * config/aarch64/constraints.md (Uvi): New.
4768 (Dd): Split into Ds and new Dd.
4769 * config/aarch64/aarch64.md (*movsi_aarch64):
4771 (*movdi_aarch64): Add SIMD mov case.
4773 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4775 * tree-predcom.c: (struct chain): Handle store-store chain in which
4776 stores for elimination only store loop invariant values.
4777 (execute_pred_commoning_chain): Ditto.
4778 (prepare_initializers_chain_store_elim): Ditto.
4779 (prepare_finalizers): Ditto.
4780 (is_inv_store_elimination_chain): New function.
4781 (initialize_root_vars_store_elim_1): New function.
4783 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4785 * tree-predcom.c: Revise general description of the pass.
4786 (enum chain_type): New enum type for store elimination.
4787 (struct chain): New field supporting store elimination.
4788 (struct component): Ditto.
4789 (dump_chain): Dump store-stores chain.
4790 (release_chain): Release resources.
4791 (split_data_refs_to_components): Compute and create component
4792 contains only stores for elimination.
4793 (get_chain_last_ref_at): New function.
4794 (make_invariant_chain): Initialization.
4795 (make_rooted_chain): Specify chain type in parameter and record it.
4796 (add_looparound_copies): Skip for store-stores chain.
4797 (determine_roots_comp): Compute type of chain and pass it to
4799 (initialize_root_vars_store_elim_2): New function.
4800 (finalize_eliminated_stores): New function.
4801 (remove_stmt): Handle store for elimination.
4802 (execute_pred_commoning_chain): Execute predictive commoning on
4804 (determine_unroll_factor): Skip unroll for store-stores chain.
4805 (prepare_initializers_chain_store_elim): New function.
4806 (prepare_initializers_chain): Hanlde store-store chain.
4807 (prepare_finalizers_chain, prepare_finalizers): New function.
4808 (tree_predictive_commoning_loop): Return integer value indicating
4809 if loop is unrolled or lcssa form is corrupted.
4810 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
4812 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4814 * tree-predcom.c (initialize_root): Delete.
4815 (execute_pred_commoning_chain): Initialize root vars and replace
4816 reference of non-combined chain directly, rather than call above
4819 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4821 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
4822 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
4824 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4826 * tree-predcom.c (struct chain): New field init_seq.
4827 (release_chain): Release init_seq.
4828 (prepare_initializers_chain): Record intialization stmts in above
4830 (insert_init_seqs): New function.
4831 (tree_predictive_commoning_loop): Call insert_init_seqs.
4833 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4835 * tree-predcom.c (determine_roots_comp): Skip trivial components.
4837 2017-07-28 Richard Biener <rguenther@suse.de>
4839 * match.pd: Remove superfluous :c.
4840 * genmatch.c (simplify::id): Add member.
4841 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
4843 (current_id): New global.
4844 (dt_node::parent): Move from ...
4845 (dt_operand::parent): ... here. Add for_id member.
4846 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
4847 (decision_tree::find_node): Relax order requirement when
4848 merging DT_TRUE nodes to ones inbetween the current simplify
4849 and the one we try to merge with. Add diagnostic whenever
4850 we need to enforce pattern order by not merging.
4851 (decision_tree::insert): Set current_id.
4852 (decision_tree::print_node): Dump parent node and for_id.
4853 (parser::last_id): Add member.
4854 (parser::push_simplify): Assign unique id.
4855 (parser::parser): Initialize last_id.
4857 2017-07-28 Martin Liska <mliska@suse.cz>
4860 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
4861 gimple_build_debug_bind.
4863 2017-07-28 Richard Biener <rguenther@suse.de>
4865 PR tree-optimization/81502
4866 * match.pd: Add pattern combining BIT_INSERT_EXPR with
4868 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
4870 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
4871 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
4872 for BIT_FIELD_REF args.
4873 * fold-const.c (make_bit_field_ref): Likewise.
4874 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
4876 2017-07-28 Jakub Jelinek <jakub@redhat.com>
4879 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
4880 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
4881 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
4882 Or it into SANITIZER_UNDEFINED.
4883 * ubsan.c: Include gimple-fold.h and varasm.h.
4884 (ubsan_expand_ptr_ifn): New function.
4885 (instrument_pointer_overflow): New function.
4886 (maybe_instrument_pointer_overflow): New function.
4887 (instrument_object_size): Formatting fix.
4888 (pass_ubsan::execute): Call instrument_pointer_overflow
4889 and maybe_instrument_pointer_overflow.
4890 * internal-fn.c (expand_UBSAN_PTR): New function.
4891 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
4892 * sanitizer.def (__ubsan_handle_pointer_overflow,
4893 __ubsan_handle_pointer_overflow_abort): New builtins.
4894 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
4895 * internal-fn.def (UBSAN_PTR): New internal function.
4896 * opts.c (sanitizer_opts): Add pointer-overflow.
4897 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
4898 * fold-const.c (build_range_check): Compute pointer range check in
4899 integral type if pointer arithmetics would be needed. Formatting
4902 2017-07-28 Martin Liska <mliska@suse.cz>
4905 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
4906 parameters that are of a variable-length.
4908 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
4910 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
4912 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
4913 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
4914 (CRT_CALL_STATIC_FUNCTION): Likewise.
4915 (ASM_DEFAULT_SPEC): New define.
4916 (ASM_SPEC32): Likewise.
4917 (ASM_SPEC64): Likewise.
4918 (ASM_SPEC_COMMON): Likewise.
4919 (ASM_SPEC): Likewise.
4920 (INVALID_64BIT): Likewise.
4921 (LINK_OS_DEFAULT_SPEC): Likewise.
4922 (LINK_OS_SPEC32): Likewise.
4923 (LINK_OS_SPEC64): Likewise.
4924 (POWERPC_LINUX): Likewise.
4925 (PTRDIFF_TYPE): Likewise.
4926 (RESTORE_FP_PREFIX): Likewise.
4927 (RESTORE_FP_SUFFIX): Likewise.
4928 (SAVE_FP_PREFIX): Likewise.
4929 (SAVE_FP_SUFFIX): Likewise.
4930 (SIZE_TYPE): Likewise.
4931 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
4932 (TARGET_64BIT): Likewise.
4933 (TARGET_64BIT): Likewise.
4934 (TARGET_AIX): Likewise.
4935 (WCHAR_TYPE_SIZE): Likewise.
4936 (WCHAR_TYPE): Undefine.
4937 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
4938 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
4939 (CPP_OS_RTEMS_SPEC): Delete.
4940 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
4941 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
4943 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
4945 2017-07-28 Jakub Jelinek <jakub@redhat.com>
4947 PR tree-optimization/81578
4948 * tree-parloops.c (build_new_reduction): Bail out if
4949 reduction_code isn't one of the standard OpenMP reductions.
4950 Move the details printing after that decision.
4952 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4954 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
4955 related to reload_in_progress.
4956 (splat_input_operand): Likewise.
4957 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
4959 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
4961 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
4962 (TARGET_INSTANTIATE_DECLS): Likewise.
4963 (legitimate_indexed_address_p): Delete reload_in_progress code.
4964 (rs6000_debug_legitimate_address_p): Likewise.
4965 (rs6000_eliminate_indexed_memrefs): Likewise.
4966 (rs6000_emit_le_vsx_store): Likewise.
4967 (rs6000_emit_move_si_sf_subreg): Likewise.
4968 (rs6000_emit_move): Likewise.
4969 (register_to_reg_type): Likewise.
4970 (rs6000_pre_atomic_barrier): Likewise.
4971 (rs6000_machopic_legitimize_pic_address): Likewise.
4972 (rs6000_allocate_stack_temp): Likewise.
4973 (rs6000_address_for_fpconvert): Likewise.
4974 (rs6000_address_for_altivec): Likewise.
4975 (rs6000_secondary_memory_needed_rtx): Delete function.
4976 (rs6000_check_sdmode): Likewise.
4977 (rs6000_alloc_sdmode_stack_slot): Likewise.
4978 (rs6000_instantiate_decls): Likewise.
4979 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
4980 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
4981 Delete reload_in_progress.
4982 (*vec_reload_and_plus_<mptrsize>): Likewise.
4983 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
4984 (vsx_div_v2di): Likewise.
4985 (vsx_udiv_v2di): Likewise.
4987 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4989 * config/rs6000/rs6000.opt (mlra): Replace with stub.
4990 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
4991 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
4992 (rs6000_debug_reg_global): Delete print of LRA status.
4993 (rs6000_option_override_internal): Delete dead LRA related code.
4994 (rs6000_lra_p): Delete function.
4995 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
4997 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
4999 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
5000 * config/riscv/rtems.h: New file.
5002 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5003 Sudakshina Das <sudi.das@arm.com>
5005 * config/aarch64/aarch64.md
5006 (define_split for and<mode>3nr_compare): Move
5007 non aarch64_logical_operand to a register.
5008 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
5009 register immediate operand to a register.
5010 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
5012 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
5015 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
5017 2017-07-27 Richard Biener <rguenther@suse.de>
5019 PR tree-optimization/81573
5020 PR tree-optimization/81494
5021 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
5022 multi defuse cycle case.
5024 2017-07-27 Richard Biener <rguenther@suse.de>
5026 PR tree-optimization/81571
5027 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
5030 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
5032 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
5033 earlier and only if MASK_FPU is set. Adjust formatting.
5035 2017-07-27 Martin Liska <mliska@suse.cz>
5037 * opt-functions.awk: Add validation of value of Init.
5038 * optc-gen.awk: Pass new argument.
5040 2017-07-27 Martin Liska <mliska@suse.cz>
5042 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
5043 Fix wrong condition.
5045 2017-07-27 Martin Liska <mliska@suse.cz>
5047 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
5048 BBs and edges seen by autoFDO.
5050 2017-07-27 Richard Biener <rguenther@suse.de>
5052 PR tree-optimization/81502
5053 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
5054 with incompatible but same sized type.
5055 (execute_update_addresses_taken): Likewise.
5057 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
5059 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
5060 flag_tree_loop_vectorize rather than flag_tree_vectorize.
5062 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5065 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
5066 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
5067 Change s_operand to memory_operand.
5069 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
5071 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
5072 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
5073 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
5074 Emit instructions rather than returning an expression. Handle TFmode
5075 and KFmode by casting to TImode.
5076 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
5077 (rs6000_emit_le_vsx_store): Likewise.
5078 * config/rs6000/vsx.md (VSX_TI): New iterator.
5079 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
5080 (*vsx_le_undo_permute_<mode>): Likewise.
5081 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
5082 emit the split sequence.
5083 (*vsx_le_perm_store_<mode>): Likewise.
5085 2017-07-27 Jakub Jelinek <jakub@redhat.com>
5087 PR tree-optimization/81555
5088 PR tree-optimization/81556
5089 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
5090 if true, force CHANGED for the recursive invocation.
5091 (reassociate_bb): Remember original length of ops array, pass
5092 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
5094 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
5095 attributes for noipa attribute. For naked attribute use
5096 lookup_attribute first before lookup_attribute_spec.
5097 * final.c (rest_of_handle_final): Disable IPA RA for functions with
5099 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
5100 for functions with noipa attribute.
5101 (cgraph_externally_visible_p): Return true for functions with noipa
5103 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
5104 for functions with noipa attribute.
5105 * doc/extend.texi: Document noipa function attribute.
5106 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
5107 also for functions with noipa attribute.
5108 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
5110 2017-07-26 Andrew Pinski <apinski@cavium.com>
5112 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
5113 vec_unalign_load_cost and vec_unalign_store_cost.
5115 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5117 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
5118 -mvsx-small-integer option.
5119 (ISA_3_0_MASKS_IEEE): Likewise.
5120 (OTHER_VSX_VECTOR_MASKS): Likewise.
5121 (POWERPC_MASKS): Likewise.
5122 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
5123 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
5124 code, only testing for DImode being allowed in non-VSX floating
5126 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
5127 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
5129 (rs6000_option_override_internal): Delete -mvsx-small-integer.
5130 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
5131 TARGET_P8_VECTOR test.
5132 (rs6000_secondary_reload_simple_move): Likewise.
5133 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
5134 since TARGET_P9_VECTOR was already tested.
5135 (rs6000_opt_masks): Remove -mvsx-small-integer.
5136 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
5137 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
5139 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
5140 test for TARGET_VEXTRACTUB was used, and that uses
5142 (p9 extract splitter): Likewise.
5143 (vsx_extract_<mode>_di_p9): Likewise.
5144 (vsx_extract_<mode>_store_p9): Likewise.
5145 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
5146 for TARGET_P9_VECTOR was used. Delete code that is now dead with
5147 the elimination of TARGET_VSX_SMALL_INTEGER.
5148 (vsx_extract_<mode>_p8): Likewise.
5149 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
5150 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
5151 (vsx_set_<mode>_p9): Likewise.
5152 (vsx_set_v4sf_p9): Likewise.
5153 (vsx_set_v4sf_p9_zero): Likewise.
5154 (vsx_insert_extract_v4sf_p9): Likewise.
5155 (vsx_insert_extract_v4sf_p9_2): Likewise.
5156 * config/rs6000/rs6000.md (sign extend splitter): Change
5157 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
5158 (floatsi<mode>2_lfiwax_mem): Likewise.
5159 (floatunssi<mode>2_lfiwzx_mem): Likewise.
5160 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
5161 since a test for TARGET_P9_VECTOR was used.
5162 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
5163 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
5164 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
5165 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
5166 TARGET_P8_VECTOR test.
5167 (fix_trunc<mode>si2_stfiwx): Likewise.
5168 (fix_trunc<mode>si2_internal): Likewise.
5169 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
5170 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
5172 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
5173 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
5174 TARGET_P8_VECTOR test.
5175 (fixuns_trunc<mode>si2_stfiwx): Likewise.
5176 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
5177 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
5179 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
5180 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
5181 since a test for TARGET_P9_VECTOR was used.
5182 (splitter for loading small constants): Likewise.
5184 2017-07-26 Andrew Pinski <apinski@cavium.com>
5186 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
5189 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
5192 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
5193 (fp_valid_at): Likewise.
5195 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
5197 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
5198 (qdf24xx_addrcost_table): Likewise.
5199 (cortexa57_tunings): Update to use generic_branch_cost.
5200 (cortexa72_tunings): Likewise.
5201 (cortexa73_tunings): Likewise.
5202 (qdf24xx_tunings): Likewise.
5204 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
5206 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
5207 (thunderx2t99_branch_cost): Likewise.
5208 (cortexa35_tunings): Update to use generic_branch_cost.
5209 (cortexa53_tunings): Likewise.
5210 (cortexa57_tunings): Likewise.
5211 (cortexa72_tunings): Likewise.
5212 (cortexa73_tunings): Likewise.
5213 (thunderx2t99_tunings): Likewise.
5215 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
5217 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
5218 (sparc_option_override): Honour MASK_FSMULD.
5219 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
5220 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
5221 * config/sparc/sparc.opt (mfsmuld): New option.
5222 * doc/invoke.texi (mfsmuld): Document option.
5224 2017-07-26 Marek Polacek <polacek@redhat.com>
5227 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
5229 2017-07-26 Richard Biener <rguenther@suse.de>
5231 * gimple-match-head.c (do_valueize): Return OP if valueize
5233 (get_def): New helper to get at the def stmt of a SSA name
5235 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
5236 do_valueize to get at the def stmt.
5237 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
5239 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
5242 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
5244 2017-07-26 Martin Liska <mliska@suse.cz>
5247 * function.c (expand_function_start): Make expansion of
5248 nonlocal_goto_save_area after parm_birth_insn.
5250 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
5252 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
5253 from all CPU target flags enable members.
5255 2017-07-26 Richard Biener <rguenther@suse.de>
5257 * genmatch.c (dt_simplify::gen): Make iterator vars const.
5258 (decision_tree::gen): Make 'type' const.
5259 (write_predicate): Likewise.
5261 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
5263 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
5264 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
5265 (rs6000_option_override_internal): Likewise.
5266 (rs6000_expand_vector_set): Likewise.
5267 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
5268 (TARGET_UPPER_REGS_SF): Likewise.
5269 (TARGET_UPPER_REGS_DI): Likewise.
5270 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
5271 (TARGET_DIRECT_MOVE_64BIT): Likewise.
5272 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
5273 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
5274 (Splitters for DI constants in Altivec registers): Likewise.
5275 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
5276 (vsx_set_v4sf_p9): Likewise.
5277 (vsx_set_v4sf_p9_zero): Likewise.
5278 (vsx_insert_extract_v4sf_p9): Likewise.
5279 (vsx_insert_extract_v4sf_p9_2): Likewise.
5281 2017-07-25 Carl Love <cel@us.ibm.com>
5283 * doc/extend.texi: Update the built-in documentation file for the
5284 existing built-in functions
5285 vector signed char vec_cnttz (vector signed char);
5286 vector unsigned char vec_cnttz (vector unsigned char);
5287 vector signed short vec_cnttz (vector signed short);
5288 vector unsigned short vec_cnttz (vector unsigned short);
5289 vector signed int vec_cnttz (vector signed int);
5290 vector unsigned int vec_cnttz (vector unsigned int);
5291 vector signed long long vec_cnttz (vector signed long long);
5292 vector unsigned long long vec_cnttz (vector unsigned long long);
5294 2017-07-25 Andrew Pinski <apinski@cavium.com>
5296 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
5297 accesses where the use is for the first operand of a BIT_INSERT.
5299 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
5302 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
5303 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
5305 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
5307 * config/i386/gstabs.h: Delete.
5308 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
5310 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
5312 * config/i386/i386.c (ix86_decompose_address): Do not check for
5313 register RTX when looking at index_reg or base_reg.
5314 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
5316 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
5318 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
5319 to update EH info here.
5321 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
5323 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
5325 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
5327 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
5329 2017-07-25 Torsten Duwe <duwe@suse.de>
5331 * common.opt: Introduce -fpatchable-function-entry
5332 command line option, and its variables function_entry_patch_area_size
5333 and function_entry_patch_area_start.
5334 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
5335 including a two-value parser.
5336 * target.def (print_patchable_function_entry): New target hook.
5337 * targhooks.h (default_print_patchable_function_entry): New function.
5338 * targhooks.c (default_print_patchable_function_entry): Likewise.
5339 * toplev.c (process_options): Switch off IPA-RA if
5340 patchable function entries are being generated.
5341 * varasm.c (assemble_start_function): Look at the
5342 patchable-function-entry command line switch and current
5343 function attributes and maybe generate NOP instructions by
5344 calling the print_patchable_function_entry hook.
5345 * doc/extend.texi: Document patchable_function_entry attribute.
5346 * doc/invoke.texi: Document -fpatchable_function_entry
5347 command line option.
5348 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
5350 * doc/tm.texi: Re-generate.
5352 2017-07-25 Jakub Jelinek <jakub@redhat.com>
5355 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
5356 TARGET_AVX512DQ rather than TARGET_AVX512BW.
5358 2017-07-25 Tamar Christina <tamar.christina@arm.com>
5360 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
5362 2017-07-25 Richard Biener <rguenther@suse.de>
5364 PR tree-optimization/81455
5365 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
5366 not walk in cycles when looking for guards.
5368 2017-07-25 Richard Biener <rguenther@suse.de>
5370 PR tree-optimization/81529
5371 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
5372 when optimizing backedge uses.
5374 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
5376 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
5378 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
5379 to dl_section_ref. On AIX, append an expression to subtract
5380 the size of the section length to dl_section_ref.
5382 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
5384 * configure.ac: If any of the config.* scripts fail, exit 1.
5385 * configure: Regenerate.
5387 2017-07-25 Richard Biener <rguenther@suse.de>
5390 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
5391 of immediate uses, be more verbose on errors.
5393 2017-07-25 Richard Biener <rguenther@suse.de>
5395 PR tree-optimization/81510
5396 * tree-vect-loop.c (vect_is_simple_reduction): When the
5397 reduction stmt is not inside the loop bail out.
5399 2017-07-25 Richard Biener <rguenther@suse.de>
5401 PR tree-optimization/81303
5402 * tree-vect-loop-manip.c (vect_loop_versioning): Build
5403 profitability check against LOOP_VINFO_NITERSM1.
5405 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
5407 * domwalk.c (cmp_bb_postorder): Simplify.
5408 (sort_bbs_postorder): New function. Use it...
5409 (dom_walker::walk): ...here to optimize common cases.
5411 2017-07-25 Martin Liska <mliska@suse.cz>
5414 * ipa-visibility.c (function_and_variable_visibility): Make the
5415 redirection just on target that supports aliasing.
5416 Fix GNU coding style.
5418 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
5421 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
5423 * config/arm/rtems.h: Likewise.
5424 * config/bfin/rtems.h: Likewise.
5425 * config/i386/rtemself.h: Likewise.
5426 * config/lm32/rtems.h: Likewise.
5427 * config/m32c/rtems.h: Likewise.
5428 * config/m68k/rtemself.h: Likewise.
5429 * config/microblaze/rtems.h: Likewise.
5430 * config/mips/rtems.h: Likewise.
5431 * config/moxie/rtems.h: Likewise.
5432 * config/nios2/rtems.h: Likewise.
5433 * config/powerpcspe/rtems.h: Likewise.
5434 * config/rs6000/rtems.h: Likewise.
5435 * config/rtems.h: Likewise.
5436 * config/sh/rtems.h: Likewise.
5437 * config/sh/rtemself.h: Likewise.
5438 * config/sparc/rtemself.h: Likewise.
5440 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
5443 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
5444 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
5445 * tree-ssa-structalias.c (alias_get_name): Same.
5447 2017-07-25 Bin Cheng <bin.cheng@arm.com>
5450 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
5451 instructions if no du chain is found.
5453 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
5455 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
5457 2017-07-25 Richard Biener <rguenther@suse.de>
5460 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
5463 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
5465 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
5467 (ISA_2_7_MASKS_SERVER): Likewise.
5468 (ISA_3_0_MASKS_IEEE): Likewise.
5469 (OTHER_P8_VECTOR_MASKS): Likewise.
5470 (OTHER_VSX_VECTOR_MASKS): Likewise.
5471 (POWERPC_MASKS): Likewise.
5472 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
5473 duplicate list of options.
5474 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
5475 explicit -mupper-regs options.
5476 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
5477 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
5478 alias for -mupper-regs-df.
5479 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
5480 (rs6000_init_hard_regno_mode_ok): Likewise.
5481 (rs6000_option_override_internal): Likewise.
5482 (rs6000_opt_masks): Likewise.
5483 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
5484 options in terms of whether -mvsx or -mpower8-vector was used.
5485 (TARGET_UPPER_REGS_DI): Likewise.
5486 (TARGET_UPPER_REGS_SF): Likewise.
5487 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
5488 -mupper-regs-* options.
5490 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
5492 * passes.c (emergency_dump_function): Print some empty lines and a
5493 header before the RTL dump.
5495 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
5497 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
5499 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
5502 * config/aarch64/aarch64.c (aarch64_classify_symbol):
5503 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
5505 2017-07-24 Carl Love <cel@us.ibm.com>
5507 * config/rs6000/rs6000-c.c: Add support for built-in functions
5508 vector float vec_extract_fp32_from_shorth (vector unsigned short);
5509 vector float vec_extract_fp32_from_shortl (vector unsigned short);
5510 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
5511 vec_extract_fp_from_shortl): Add defines for the two builtins.
5512 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
5513 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
5515 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
5516 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
5517 * doc/extend.texi: Update the built-in documentation file for the
5518 new built-in function.
5520 2017-07-24 Jakub Jelinek <jakub@redhat.com>
5523 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
5525 * doc/generic.texi: Likewise.
5526 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
5527 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
5529 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
5531 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
5532 (aarch64_mls_elt_merge<mode>): Likewise.
5534 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
5536 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
5537 having __cxa_atexit.
5539 2017-07-23 Michael Collison <michael.collison@arm.com>
5541 * config/arm/arm.c (arm_option_override): Deprecate
5542 use of -mstructure-size-boundary.
5543 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
5544 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
5546 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5549 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
5550 Reduce cost estimate for direct moves.
5552 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
5555 * config/i386/i386.c (ix86_option_override_internal): Disable
5556 BMI, BMI2 and TBM instructions for -m16.
5558 2017-07-21 Carl Love <cel@us.ibm.com>
5560 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5561 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5562 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5563 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5564 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5565 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
5566 VMULOSW): New enum "unspec" values.
5567 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
5568 altivec_vmulosw): New patterns.
5569 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5570 VMULOSW): Add definitions.
5572 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
5574 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
5575 (qdf24xx): Likewise.
5576 * config/aarch64/aarch64-options-extensions.def (rdma); New.
5577 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
5578 (AARCH64_FL_V8_1): Renumber.
5579 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
5580 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
5581 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
5582 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
5583 rdma to feature modifiers list.
5585 2017-07-21 Yury Gribov <tetra2005@gmail.com>
5588 * ipa-visibility (function_and_variable_visibility): Convert
5589 recursive PLT call to direct call if appropriate.
5591 2017-07-21 Andrew Pinski <apinski@cavium.com>
5593 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
5594 operand 1 to see if the types precision matches.
5595 * fold-const.c (operand_equal_p): Likewise.
5597 2017-07-21 Richard Biener <rguenther@suse.de>
5599 PR tree-optimization/81303
5600 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
5601 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
5602 (vect_peeling_hash_get_lowest_cost): Adjust.
5603 (vect_enhance_data_refs_alignment): Likewise. Use
5604 vect_get_peeling_costs_all_drs to compute the penalty for no
5605 peeling to match up costs.
5607 2017-07-21 Richard Biener <rguenther@suse.de>
5609 PR tree-optimization/81500
5610 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
5611 we didn't identify a reduction path.
5613 2017-07-21 Tom de Vries <tom@codesourcery.com>
5614 Cesar Philippidis <cesar@codesourcery.com>
5616 PR gcov-profile/81442
5617 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
5620 2017-07-21 Tom de Vries <tom@codesourcery.com>
5623 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
5625 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
5626 nvptx_override_options_after_change.
5628 2017-07-21 Ulrich Drepper <drepper@redhat.com>
5630 * dwarf2out.c (output_file_names): Avoid double testing for
5633 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
5635 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
5637 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
5639 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
5641 (try_crossjump_to_edge): Do not punt on partitioned functions.
5643 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
5645 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
5646 Put all BBs reachable only via paths crossing cold region to cold
5648 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
5650 2016-07-21 Richard Biener <rguenther@suse.de>
5652 PR tree-optimization/81303
5653 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
5654 into account prologue and epilogue iterations when raising
5655 min_profitable_iters to sth at least covering one vector iteration.
5657 2017-07-21 Tamar Christina <tamar.christina@arm.com>
5659 * config/arm/arm.c (arm_test_cpu_arch_dat):
5662 2017-07-20 Nathan Sidwell <nathan@acm.org>
5664 Remove TYPE_METHODS.
5665 * tree.h (TYPE_METHODS): Delete.
5666 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
5667 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
5668 (dbxout_type_methods): Scan TYPE_FIELDS.
5669 (dbxout_type): Don't check TYPE_METHODS here.
5670 * function.c (use_register_for_decl): Always ignore register for
5671 class types when not optimizing.
5672 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
5673 * tree.c (free_lang_data_in_type): Stitch out member functions and
5674 templates from TYPE_FIELDS.
5675 (build_distinct_type_copy, verify_type_variant,
5676 verify_type): Member fns are on TYPE_FIELDS.
5677 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
5678 * tree-pretty-print.c (dump_generic_node): Likewise.
5680 2017-07-20 Jakub Jelinek <jakub@redhat.com>
5683 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
5684 V2TImode and V4TImode.
5685 (ix86_expand_vector_extract): Likewise.
5686 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
5687 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
5688 (ssescalarmode): Handle V4TImode and V2TImode.
5689 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
5690 (*vec_extractv2ti, *vec_extractv4ti): New insns.
5691 (VEXTRACTI128_MODE): New mode iterator.
5692 (splitter for *vec_extractv?ti first element): New.
5693 (VEC_INIT_MODE): New mode iterator.
5694 (vec_init<mode>): Consolidate 3 expanders into one using
5695 VEC_INIT_MODE mode iterator.
5697 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
5699 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
5700 non_spilled_static_chain_regno_p.
5702 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
5704 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
5706 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
5708 * bb-reorder.c (connect_traces): Allow copying of blocks within
5711 2017-07-20 Richard Biener <rguenther@suse.de>
5713 * gimple.h (gimple_phi_result): Add gphi * overload.
5714 (gimple_phi_result_ptr): Likewise.
5715 (gimple_phi_arg): Likewise. Adjust index assert to only
5716 allow actual argument accesses rather than all slots available
5718 (gimple_phi_arg_def): Add gphi * overload.
5719 * tree-phinodes.c (make_phi_node): Initialize only actual
5721 (resize_phi_node): Clear memory not covered by old node,
5722 do not initialize excess argument slots.
5723 (reserve_phi_args_for_new_edge): Initialize new argument slot
5726 2017-07-20 Bin Cheng <bin.cheng@arm.com>
5728 PR tree-optimization/81388
5730 2016-07-21 Bin Cheng <bin.cheng@arm.com>
5732 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
5733 by removing computation of may_be_zero.
5735 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5736 Tom de Vries <tom@codesourcery.com>
5739 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
5740 when gimple level profile disagrees with what RTL expander did.
5742 2017-07-20 Richard Biener <rguenther@suse.de>
5744 PR tree-optimization/61171
5745 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
5746 (vect_analyze_stmt): Add slp instance parameter.
5747 (vectorizable_reduction): Likewise.
5748 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
5749 (vect_is_simple_reduction): Deal with chains not detected
5750 as SLP reduction chain, specifically not properly associated
5751 chains containing a mix of plus/minus.
5752 (get_reduction_op): Remove.
5753 (get_initial_defs_for_reduction): Simplify, pass in whether
5754 this is a reduction chain, pass in the SLP node for the PHIs.
5755 (vect_create_epilog_for_reduction): Get the SLP instance as
5757 (vectorizable_reduction): Get the SLP instance as arg.
5758 During analysis remember the SLP node with the PHIs in the
5759 instance. Simplify getting at the vectorized reduction PHIs.
5760 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
5761 through SLP instance.
5762 (vect_slp_analyze_operations): Likewise.
5763 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
5764 (vect_transform_stmt): Likewise.
5766 2017-07-20 Tom de Vries <tom@codesourcery.com>
5768 PR tree-optimization/81489
5769 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
5770 read of phi arg location to before loop that modifies phi.
5772 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
5774 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
5777 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
5780 * except.c (execute): Fix ordering issue.
5782 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
5784 PR rtl-optimization/81423
5785 * combine.c (make_compound_operation_int): Don't try to optimize
5786 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
5788 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
5790 PR rtl-optimization/81423
5791 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
5792 with a constant that is -1 in the truncated to mode.
5794 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
5796 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
5797 (determine_unlikely_bbs): ... here.
5798 * predict.h (propagate_unlikely_bbs_forward): Declare.
5799 * cfgexpand.c (pass_expand::execute): Use it.
5800 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
5802 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
5803 propagate_unlikely_bbs_forward.
5805 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
5808 * except.c (maybe_add_nop_after_section_switch): New function.
5811 2017-07-19 Tom de Vries <tom@codesourcery.com>
5813 * gimple.h (gimple_phi_set_arg): Make assert more strict.
5815 2017-07-19 Tom de Vries <tom@codesourcery.com>
5817 * gimple.h (gimple_phi_arg): Make assert more strict.
5819 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
5821 * config.gcc (powerpc*-*-*): Add mmintrin.h.
5822 * config/rs6000/mmintrin.h: New file.
5823 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
5825 2017-07-19 Jakub Jelinek <jakub@redhat.com>
5827 PR tree-optimization/81346
5828 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
5830 2017-07-19 Tom de Vries <tom@codesourcery.com>
5832 * config/nvptx/nvptx.md (VECIM): Add V2DI.
5834 2017-07-19 Tom de Vries <tom@codesourcery.com>
5836 * config/nvptx/nvptx-modes.def: Add V2DImode.
5837 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
5838 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
5839 (nvptx_output_mov_insn): Handle lack of mov.b128.
5840 (nvptx_print_operand): Handle 'H' and 'L' codes.
5841 (nvptx_vector_mode_supported): Allow V2DImode.
5842 (nvptx_preferred_simd_mode): New function.
5843 (nvptx_data_alignment): New function.
5844 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
5845 nvptx_preferred_simd_mode.
5846 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
5848 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
5850 2017-07-19 Tom de Vries <tom@codesourcery.com>
5852 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
5853 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
5854 (nvptx_vector_mode_supported): New function. Allow V2SImode.
5855 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
5856 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
5857 (mov<VECIM>_insn): New define_insn.
5858 (define_expand "mov<VECIM>): New define_expand.
5860 2017-07-19 Tom de Vries <tom@codesourcery.com>
5862 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
5864 2017-07-19 Jakub Jelinek <jakub@redhat.com>
5866 PR tree-optimization/81346
5867 * fold-const.h (fold_div_compare, range_check_type): Declare.
5868 * fold-const.c (range_check_type): New function.
5869 (build_range_check): Use range_check_type.
5870 (fold_div_compare): No longer static, rewritten into
5871 a match.pd helper function.
5872 (fold_comparison): Don't call fold_div_compare here.
5873 * match.pd (X / C1 op C2): New optimization using fold_div_compare
5876 2017-07-19 Nathan Sidwell <nathan@acm.org>
5878 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
5879 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
5880 * tree.c (find_decls_types_r, verify_type): Use
5881 TYPE_{MIN,MAX}_VALUE_RAW.
5882 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
5883 (hash_tree): Likewise.
5884 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
5886 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
5889 2017-07-18 Tom de Vries <tom@codesourcery.com>
5892 * omp-expand.c (expand_omp_for_static_chunk): Handle
5893 equal-argument loop exit phi.
5895 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
5898 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
5899 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
5900 operand 2 predicate.
5901 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
5902 operand 2 predicate.
5903 (ror,rol -> rorx splitters): Use const_int_operand as
5904 operand 2 predicate.
5906 2017-06-18 Richard Biener <rguenther@suse.de>
5908 PR tree-optimization/81410
5909 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
5910 the gap in the ! slp_perm SLP case after each group.
5912 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5915 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
5918 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5921 * predict.c (set_even_probabilities): Cleanup; do not affect
5922 probabilities that are already known.
5923 (combine_predictions_for_bb): Call even when count is set.
5925 2017-07-18 Nathan Sidwell <nathan@acm.org>
5927 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
5930 2017-07-18 Bin Cheng <bin.cheng@arm.com>
5933 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
5934 optimization for loop niter analysis.
5936 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
5939 * config/avr/avr.c (avr_optimize_casesi): Don't use
5940 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
5942 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5944 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
5945 body_cost_vec from _vect_peel_extended_info.
5946 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
5947 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
5950 2017-07-18 Bin Cheng <bin.cheng@arm.com>
5952 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
5954 2017-07-18 Richard Biener <rguenther@suse.de>
5956 PR tree-optimization/80620
5957 PR tree-optimization/81403
5958 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
5959 info when re-using a VN table entry.
5961 2017-07-18 Richard Biener <rguenther@suse.de>
5963 PR tree-optimization/81418
5964 * tree-vect-loop.c (vectorizable_reduction): Properly compute
5965 vectype_in. Verify that with lane-reducing reduction operations
5966 we have a single def-use cycle.
5968 2017-07-17 Carl Love <cel@us.ibm.com>
5970 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
5972 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5973 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5974 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5975 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5976 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5977 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
5978 VMULOSW): New enum "unspec" values.
5979 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
5980 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
5981 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
5982 altivec_vmulosw): New patterns.
5983 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5984 VMULOSW): Add definitions.
5985 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
5987 * config/alpha/alpha.c: Include predict.h.
5989 2017-07-17 Yury Gribov <tetra2005@gmail.com>
5991 * tree-vrp.c (compare_assert_loc): Fix comparison function
5992 to return predictable results.
5994 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5996 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
5999 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
6000 * doc/invoke.texi (mexpand-adddi): Update text.
6002 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6004 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
6005 that also clobbers the CC register. The old expand code is moved
6007 (*arc_clzsi2): ... here.
6008 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
6009 the CC register. The old expand code is moved to ...
6010 (arc_ctzsi2): ... here.
6012 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6014 * config/arc/arc.opt (mindexed-loads): Use initial value
6015 TARGET_INDEXED_LOADS_DEFAULT.
6016 (mauto-modify-reg): Use initial value
6017 TARGET_AUTO_MODIFY_REG_DEFAULT.
6018 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
6019 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
6020 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
6021 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
6023 2017-07-17 Martin Liska <mliska@suse.cz>
6026 * opts.c (finish_options): Do not allow -fgnu-tm
6027 w/ -fsanitize={kernel-,}address. Say sorry.
6029 2017-07-17 Bin Cheng <bin.cheng@arm.com>
6032 * tree-loop-distribution.c (classify_partition): Only assert on
6033 numer of iterations.
6034 (merge_dep_scc_partitions): Delete prameter. Update function call.
6035 (distribute_loop): Remove code handling loop with unknown niters.
6036 (pass_loop_distribution::execute): Skip loop with unknown niters.
6038 2017-07-17 Bin Cheng <bin.cheng@arm.com>
6041 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
6042 function sort_partitions_by_post_order.
6044 2017-07-17 Bin Cheng <bin.cheng@arm.com>
6046 PR tree-optimization/81374
6047 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
6048 the max index of basic blocks, rather than number of basic blocks.
6050 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6052 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
6054 (arc_legitimate_pic_operand_p): Likewise.
6055 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
6057 (arc_needs_pcl_p): Likewise.
6058 (arc_legitimate_pc_offset_p): Likewise.
6059 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
6060 function is also used in constrains.md.
6061 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
6062 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
6063 PLUS. Only return true/false in known cases, otherwise assert.
6064 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
6065 is already called in arc_legitimate_constant_p.
6066 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
6068 (LEGITIMATE_PIC_OPERAND_P): Use
6069 arc_raw_symbolic_reference_mentioned_p function.
6070 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
6075 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6076 Andrew Burgess <andrew.burgess@embecosm.com>
6078 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
6079 (arc_return_address_register): New function.
6080 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
6081 (arc_handle_fndecl_attribute): Add naked attribute.
6082 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
6083 (TARGET_WARN_FUNC_RETURN): Likewise.
6084 (arc_allocate_stack_slots_for_args): New function.
6085 (arc_warn_func_return): Likewise.
6086 (machine_function): Change type fn_type.
6087 (arc_compute_function_type): Consider new naked function type,
6088 change function return type.
6089 (arc_must_save_register): Adapt to handle new
6090 arc_compute_function_type's return type.
6091 (arc_expand_prologue): Likewise.
6092 (arc_expand_epilogue): Likewise.
6093 (arc_return_address_regs): Delete.
6094 (arc_return_address_register): New function.
6095 (arc_epilogue_uses): Use above function.
6096 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
6097 (arc_function_type): Change encoding, add naked type.
6098 (ARC_INTERRUPT_P): Change to handle the new encoding.
6099 (ARC_FAST_INTERRUPT_P): Likewise.
6100 (ARC_NORMAL_P): Define.
6101 (ARC_NAKED_P): Likewise.
6102 (arc_compute_function_type): Delete prototype.
6103 * config/arc/arc.md (in_ret_delay_slot): Use
6104 arc_return_address_register function.
6105 (simple_return): Likewise.
6106 (p_return_i): Likewise.
6108 2017-07-17 Jakub Jelinek <jakub@redhat.com>
6110 PR tree-optimization/81428
6111 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
6112 can't be built for those types.
6114 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
6116 Remove stuff dead since r239246.
6118 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
6119 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
6120 (avr_inform_devices): Remove dead stuff.
6122 2017-07-17 Tamar Christina <tamar.christina@arm.com>
6124 * config/arm/arm_neon.h: Fix softp typo.
6126 2017-07-17 Jakub Jelinek <jakub@redhat.com>
6128 PR tree-optimization/81365
6129 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
6130 aggregate moves onto bb predecessor edges, make sure there are no
6131 loads that could alias the lhs in between the start of bb and the
6134 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
6137 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
6138 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
6139 [LSHIFTRT, outer_code = TRUNCATE]: Same.
6141 2017-07-17 Jakub Jelinek <jakub@redhat.com>
6143 PR tree-optimization/81396
6144 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
6145 (init_symbolic_number): Initialize it to 1.
6146 (perform_symbolic_merge): Add n_ops from both operands into the new
6148 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
6149 without base_addr as useless if they need more than one operation.
6150 (bswap_replace): Handle !bswap case for NULL base_addr.
6152 2017-07-17 Tom de Vries <tom@codesourcery.com>
6155 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
6158 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
6160 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
6161 conditional builtin define __FIX_LEON3FT_B2BST.
6163 2017-07-17 Daniel Cederman <cederman@gaisler.com>
6165 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
6166 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
6169 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
6171 PR rtl-optimization/81424
6172 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
6173 to remove potential trapping from operands if -fnon-call-exceptions.
6175 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6177 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
6178 profile_proability for scalling.
6179 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
6181 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6183 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
6185 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6187 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
6188 fixpoint arithmetics.
6190 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6192 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
6193 fixpoint arithmetics.
6195 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6197 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
6198 fixpoint arithmetics.
6200 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6202 * profile-count.h (profile_probability::from_reg_br_prob_note,
6203 profile_probability::to_reg_br_prob_note): New functions.
6204 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
6205 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
6206 * predict.c (probability_reliable_p): Update.
6207 (edge_probability_reliable_p): Update.
6208 (br_prob_note_reliable_p): Update.
6209 (invert_br_probabilities): Update.
6210 (add_reg_br_prob_note): New function.
6211 (combine_predictions_for_insn): Update.
6212 * asan.c (asan_clear_shadow): Update.
6213 * cfgbuild.c (compute_outgoing_frequencies): Update.
6214 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
6215 (update_br_prob_note): Update.
6216 (rtl_verify_edges): Update.
6217 (purge_dead_edges): Update.
6218 (fixup_reorder_chain): Update.
6219 * emit-rtl.c (try_split): Update.
6220 * ifcvt.c (cond_exec_process_insns): Update.
6221 (cond_exec_process_if_block): Update.
6222 (dead_or_predicable): Update.
6223 * internal-fn.c (expand_addsub_overflow): Update.
6224 (expand_neg_overflow): Update.
6225 (expand_mul_overflow): Update.
6226 * loop-doloop.c (doloop_modify): Update.
6227 * loop-unroll.c (compare_and_jump_seq): Update.
6228 * optabs.c (emit_cmp_and_jump_insn_1): Update.
6229 * predict.h: Update.
6230 * reorg.c (mostly_true_jump): Update.
6232 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
6233 * config/alpha/alpha.c (emit_unlikely_jump): Update.
6234 * config/arc/arc.c: (emit_unlikely_jump): Update.
6235 * config/arm/arm.c: (emit_unlikely_jump): Update.
6236 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
6237 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
6238 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
6239 (ix86_print_operand): Update.
6240 (ix86_split_fp_branch): Update.
6241 (predict_jump): Update.
6242 * config/ia64/ia64.c (ia64_print_operand): Update.
6243 * config/mmix/mmix.c (mmix_print_operand): Update.
6244 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
6245 (rs6000_expand_split_stack_prologue): Update.
6246 * config/rs6000/rs6000.c: Update.
6247 * config/s390/s390.c (s390_expand_vec_strlen): Update.
6248 (s390_expand_vec_movstr): Update.
6249 (s390_expand_cs_tdsi): Update.
6250 (s390_expand_split_stack_prologue): Update.
6251 * config/sh/sh.c (sh_print_operand): Update.
6252 (expand_cbranchsi4): Update.
6253 (expand_cbranchdi4): Update.
6254 * config/sparc/sparc.c (output_v9branch): Update.
6255 * config/spu/spu.c (get_branch_target): Update.
6256 (ea_load_store_inline): Update.
6257 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
6258 * config/tilepro/tilepro.c: Update.
6260 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
6262 * gimplify.c (mostly_copy_tree_r): Revert latest change.
6263 (gimplify_save_expr): Likewise.
6265 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6267 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
6269 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6271 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
6273 * timevar.def (TV_IPA_FNSUMMARY): Define.
6275 2017-07-16 Daniel Cederman <cederman@gaisler.com>
6277 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
6278 to back store errata sensitive sequence from being generated.
6279 (sqrtdf2_fix): Likewise.
6281 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6283 * tree-ssa-threadupdate.c (compute_path_counts,
6284 update_joiner_offpath_counts): Use profile_probability.
6286 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
6289 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
6291 * config/arm/arm-c.c (arm_cpu_builtins): Define
6292 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
6294 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
6296 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6297 array entries to represent __ieee128 versions of the
6298 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
6299 scalar_extract_sig, and scalar_insert_exp built-in functions.
6300 (altivec_resolve_overloaded_builtin): Add special case handling
6301 for the __builtin_scalar_insert_exp function, as represented by
6302 the P9V_BUILTIN_VEC_VSIEDP constant.
6303 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
6304 exponent support for __ieee128 argument.
6305 (VSESQP): Add scalar extract signature support for __ieee128
6307 (VSTDCNQP): Add scalar test negative support for __ieee128
6309 (VSIEQP): Add scalar insert exponent support for __int128 argument
6310 with __ieee128 result.
6311 (VSIEQPF): Add scalar insert exponent support for __ieee128
6312 argument with __ieee128 result.
6313 (VSTDCQP): Add scalar test data class support for __ieee128
6315 (VSTDCNQP): Add overload support for scalar test negative with
6317 (VSTDCQP): Add overload support for scalar test data class
6319 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
6321 (UNSPEC_VSX_SIEXPQP): New constant.
6322 (xsxexpqp): New insn for VSX scalar extract exponent quad
6324 (xsxsigqp): New insn for VSX scalar extract significand quad
6326 (xsiexpqpf): New insn for VSX scalar insert exponent quad
6327 precision with floating point argument.
6328 (xststdcqp): New expand for VSX scalar test data class quad
6330 (xststdcnegqp): New expand for VSX scalar test negative quad
6332 (xststdcqp): New insn to match expansions for VSX scalar test data
6333 class quad precision and VSX scalar test negative quad precision.
6334 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6335 special case operand checking to enforce that second operand of
6336 VSX scalar test data class with quad precision argument is a 7-bit
6338 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
6339 prototypes and descriptions of __ieee128 versions of
6340 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
6341 scalar_test_data_class, and scalar_test_neg built-in functions.
6343 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6345 PR tree-optimization/81162
6346 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
6347 replace a negate with an add.
6349 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
6351 * doc/invoke.texi (arm/-mcpu): Document +crypto.
6353 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
6355 * config/arm/arm-c.c (arm_cpu_builtins): Define
6356 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
6358 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
6360 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
6361 (armv8-r): Set ARM Cortex-R52 as default CPU.
6362 * config/arm/arm-tables.opt: Regenerate.
6363 * config/arm/arm-tune.md: Regenerate.
6364 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
6366 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
6367 extension for -mcpu=cortex-r52.
6369 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
6371 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
6372 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
6373 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
6374 (fp-armv8): Define it as FP_ARMv8 only.
6375 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
6376 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
6378 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
6379 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
6380 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
6381 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
6382 than TARGET_FPU_ARMV8.
6383 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
6384 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
6385 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
6387 * config/arm/neon.md (neon_vrint): Likewise.
6388 (neon_vcvt): Likewise.
6389 (neon_<fmaxmin_op><mode>): Likewise.
6390 (<fmaxmin><mode>3): Likewise.
6391 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
6392 * config/arm/predicates.md (arm_cond_move_operator): Check against
6393 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
6395 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
6397 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
6400 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6402 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
6403 loop in comment with memset.
6405 2017-07-14 Martin Liska <mliska@suse.cz>
6407 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
6408 * dwarf2out.c (is_java): Remove the function.
6409 (output_pubname): Remove usage of the function.
6410 (lower_bound_default): Remove usage of DW_LANG_Java.
6411 (gen_compile_unit_die): Likewise.
6412 * gcc.c: Remove compiler defaults for .java and .zip files.
6413 * gimple-expr.c (remove_suffix): Change as there's no longer
6414 extension than 4-letter one.
6415 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
6416 (gimplify_save_expr): Likewise.
6417 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
6418 as it's possible even for other languages than Java.
6419 * langhooks.h (struct lang_hooks): Remove Java from a comment.
6420 * lto-opts.c (lto_write_options): Remove reference to Java.
6421 * opts.c (strip_off_ending): Update file extension handling.
6422 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
6423 * tree-eh.c (lower_resx): Likewise.
6424 * tree.c (free_lang_data_in_type): Remove dead code.
6425 (find_decls_types_r): Likewise.
6426 (build_common_builtin_nodes): Remove Java from a comment.
6427 (verify_type): Remove dead code.
6428 * varasm.c (assemble_external): Remove Java from a comment.
6430 2017-07-14 Martin Liska <mliska@suse.cz>
6432 * opts.c (finish_options): Add quotes.
6433 (common_handle_option): Likewise.
6435 2017-07-14 Martin Liska <mliska@suse.cz>
6437 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
6438 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
6440 * dwarf2out.c (lower_bound_default): Do not handle
6442 (gen_compile_unit_die): Likewise.
6443 * gcc.c: Remove default extension binding for GNU Pascal.
6444 * stmt.c: Remove Pascal language from a comment.
6445 * xcoffout.c: Likewise.
6447 2017-07-13 David Malcolm <dmalcolm@redhat.com>
6450 * diagnostic-show-locus.c (fixit_cmp): New function.
6451 (layout::layout): Sort m_fixit_hints.
6452 (column_range::column_range): Assert that the values are valid.
6453 (struct char_span): New struct.
6454 (correction::overwrite): New method.
6455 (struct source_line): New struct.
6456 (line_corrections::add_hint): Add assertions. Reimplement memcpy
6457 calls in terms of classes source_line and char_span, and
6458 correction::overwrite.
6459 (selftest::test_overlapped_fixit_printing_2): New function.
6460 (selftest::diagnostic_show_locus_c_tests): Call it.
6462 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
6464 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
6465 early if there is no lhs.
6467 2017-07-13 Martin Liska <mliska@suse.cz>
6469 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
6470 (gen_reference_type_die): Likewise.
6471 * stor-layout.c: Remove Pascal-related comment.
6473 2017-07-13 Martin Liska <mliska@suse.cz>
6475 * opts.c (finish_options): Add quotes to error messages.
6476 (parse_sanitizer_options): Likewise.
6478 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
6480 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
6482 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
6484 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
6486 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
6488 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
6490 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
6492 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
6495 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
6496 provides the hardware capability bits, define the macro
6497 __BUILTIN_CPU_SUPPORTS__.
6498 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
6499 if GLIBC does not provide the hardware capability bits. Add a
6500 gcc_unreachable call if the built-in cpu function is neither
6501 __builtin_cpu_is nor __builtin_cpu_supports.
6502 (rs6000_get_function_versions_dispatcher): Change the warning
6503 that an old GLIBC is used which does not export the capability
6504 bits to be an error.
6505 * doc/extend.texi (target_clones attribute): Document the
6506 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
6507 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
6508 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
6509 the macros defined by GCC if the newer GLIBC is available.
6511 2017-07-12 Jeff Law <law@redhat.com>
6513 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
6514 remaining includes slightly.
6515 * config/riscv/riscv-builtins.c: Include profile-count.h.
6517 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
6520 * config/avr/avr.c (avr_set_current_function): In diagnostic
6521 messages: Quote keywords and (parts of) identifiers.
6522 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
6525 2017-07-12 Carl Love <cel@us.ibm.com>
6527 * config/rs6000/rs6000-c.c: Add support for built-in functions
6528 vector bool char vec_revb (vector bool char);
6529 vector bool short vec_revb (vector short char);
6530 vector bool int vec_revb (vector bool int);
6531 vector bool long long vec_revb (vector bool long long);
6532 * doc/extend.texi: Update the built-in documentation file for the
6533 new built-in functions.
6535 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6537 * config/s390/s390.md: Remove movcc splitter.
6539 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6541 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
6542 load/store on condition.
6544 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
6547 * config/avr/avr.c (avr_encode_section_info)
6548 [progmem && !TREE_READONLY]: Error if progmem object needs
6551 2017-07-11 Michael Collison <michael.collison@arm.com>
6553 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
6556 2017-07-11 Carl Love <cel@us.ibm.com>
6558 * config/rs6000/rs6000-c.c: Add support for builtins
6559 vector unsigned int vec_parity_lsbb (vector signed int);
6560 vector unsigned int vec_parity_lsbb (vector unsigned int);
6561 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
6562 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
6563 vector unsigned long long vec_parity_lsbb (vector signed long long);
6564 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
6565 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
6566 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
6567 * doc/extend.texi: Update the built-in documentation file for the
6568 new built-in functions.
6570 2017-07-11 David Malcolm <dmalcolm@redhat.com>
6572 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
6573 (layout::m_primary_loc): New field.
6574 (layout::layout): Initialize new field. Move location filtering
6575 logic from here to...
6576 (layout::maybe_add_location_range): ...this new method. Add
6577 support for filtering to just the lines already specified by other
6579 (layout::will_show_line_p): New method.
6580 (gcc_rich_location::add_location_if_nearby): New method.
6581 (selftest::test_add_location_if_nearby): New test function.
6582 (selftest::diagnostic_show_locus_c_tests): Call it.
6583 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
6586 2017-07-11 Tom de Vries <tom@codesourcery.com>
6588 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
6589 (bb_first_real_insn): New function.
6590 (nvptx_single): Add extra initialization of broadcasted condition
6593 2017-07-11 Nathan Sidwell <nathan@acm.org>
6595 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
6597 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
6599 * doc/extend.texi (AVR Function Attributes): Remove weblink to
6600 Binutils doc as TEXI will mess them up.
6601 * doc/invoke.texi (AVR Options): Same here.
6603 2017-07-11 Daniel Cederman <cederman@gaisler.com>
6605 * config/sparc/sparc.opt (mfix-ut700): New option.
6606 (mfix-gr712rc): Likewise.
6607 (sparc_fix_b2bst): New variable.
6608 * doc/invoke.texi (SPARC options): Document them.
6609 (ARM options): Fix warnings.
6610 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
6611 instructions to prevent sequences that can trigger the store-store
6612 errata for certain LEON3FT processors.
6613 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
6614 (sparc_option_override): Set sparc_fix_b2bst appropriately.
6615 * config/sparc/sparc.md (fix_b2bst): New attribute.
6616 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
6618 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
6621 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
6623 (*rsqrtsf2_sse): Ditto.
6625 (div<mode>3): Macroize insn from divdf3 and divsf3
6626 using MODEF mode iterator.
6628 2017-07-10 Martin Sebor <msebor@redhat.com>
6630 PR tree-optimization/80397
6631 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
6632 instead of testing for equality to INTEGER_TYPE.
6634 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
6636 * config.gcc: Remove uclibc from arc target spec.
6638 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
6640 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
6642 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6645 * lto-wrapper.c (remove_option): New function.
6646 (merge_and_complain): Merge PIC/PIE options more realistically.
6648 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
6650 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
6654 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
6655 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
6656 * config.in: Regenerate.
6657 * configure: Regenerate.
6658 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
6659 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
6660 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
6661 (TARGET_GASISR_PROLOGUES): ...target mask.
6662 * common/config/avr/avr-common.c
6663 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
6664 Set -mgas-isr-prologues.
6665 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
6666 INSERT_PASS_BEFORE for it.
6667 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
6668 * config/avr/avr.c (avr_option_override)
6669 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
6670 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
6671 (avr_attribute_table) <no_gccisr>: Add new function attribute.
6672 (avr_set_current_function) <is_no_gccisr>: Init machine field.
6673 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
6675 (make_avr_pass_pre_proep): New function.
6676 (emit_push_sfr) <treg>: Add argument to function and use it
6678 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
6679 and set machine->gasisr.yes.
6680 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
6681 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
6682 __gcc_isr.n_pushed to .L__stack_usage.
6683 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
6684 (avr_asm_final_postscan_insn): ...this new static function.
6685 * config/avr/avr.h (machine_function)
6686 <is_no_gccisr, use_L__stack_usage>: New fields.
6687 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
6688 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
6689 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
6690 (gasisr, *gasisr): New expander and insn.
6691 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
6692 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
6693 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
6695 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
6697 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
6700 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
6702 Move jump-tables out of .text again.
6705 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
6706 (ASM_OUTPUT_ADDR_VEC): New function.
6707 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
6708 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
6709 INSN_ADDRESSes as asm comment.
6710 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
6711 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
6712 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
6713 * config/avr/avr.md (*tablejump): Adjust comment.
6714 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
6715 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
6717 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
6718 (avr_output_addr_vec): New proto.
6719 (avr_log_t) <insn_addresses>: New field.
6721 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
6724 * config/i386/i386.c (ix86_function_arg_advance): Set
6725 outgoing_args_on_stack to true if there are outgoing arguments
6727 (ix86_function_arg): Likewise.
6728 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
6729 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
6730 * config/i386/i386.h (machine_function): Add
6731 outgoing_args_on_stack.
6733 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
6735 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
6737 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
6739 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
6741 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
6742 (REAL_H): Remove $(MACHMODE_H).
6743 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
6745 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
6746 $(MACHMODE_H) and double-int.h.
6747 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
6749 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
6752 2017-07-07 Andrew Pinski <apinski@cavium.com>
6754 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
6755 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
6757 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6759 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
6760 Add warning if GCC was not configured to link against a GLIBC that
6761 exports the hardware capability bits.
6762 (make_resolver_func): Make resolver function private and not a
6763 COMDAT function. Create the name with clone_function_name instead
6764 of make_unique_name.
6767 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
6768 correct operand in doing the split.
6770 2017-07-07 Carl Love <cel@us.ibm.com>
6772 * config/rs6000/rs6000-c: Add support for built-in function
6773 vector unsigned short vec_pack_to_short_fp32 (vector float,
6775 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
6776 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
6777 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
6778 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
6779 (convert_4f32_8i16): Add define_expand.
6780 * doc/extend.texi: Update the built-in documentation file for the
6781 new built-in function.
6783 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6785 * config/sparc/m8.md: New file.
6786 * config/sparc/sparc.md: Include m8.md.
6788 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6790 * config/sparc/sparc.opt: New option -mvis4b.
6791 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
6792 (sparc_option_override): Handle VIS4B.
6793 (enum sparc_builtins): Define
6794 SPARC_BUILTIN_DICTUNPACK{8,16,32},
6795 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
6796 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
6797 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
6798 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
6799 (check_constant_argument): New function.
6800 (sparc_vis_init_builtins): Define builtins
6801 __builtin_vis_dictunpack{8,16,32},
6802 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
6803 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
6804 __builtin_vis_fpcmpde{8,16,32}shl and
6805 __builtin_vis_fpcmpur{8,16,32}shl.
6806 (sparc_expand_builtin): Check that the constant operands to
6807 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
6808 constant and in range.
6809 * config/sparc/sparc-c.c (sparc_target_macros): Handle
6811 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
6812 (SPARC_IMM5_P): Likewise.
6813 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
6814 (enabled): Handle vis4b.
6815 (UNSPEC_DICTUNPACK): New unspec.
6816 (UNSPEC_FPCMPSHL): Likewise.
6817 (UNSPEC_FPUCMPSHL): Likewise.
6818 (UNSPEC_FPCMPDESHL): Likewise.
6819 (UNSPEC_FPCMPURSHL): Likewise.
6820 (cpu_feature): New CPU feature `vis4b'.
6821 (dictunpack{8,16,32}): New insns.
6822 (FPCSMODE): New mode iterator.
6823 (fpcscond): New code iterator.
6824 (fpcsucond): Likewise.
6825 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
6826 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
6827 (fpcmpde{8,16,32}{si,di}shl): Likewise.
6828 (fpcmpur{8,16,32}{si,di}shl): Likewise.
6829 * config/sparc/constraints.md: Define constraints `q' for unsigned
6830 2-bit integer constants and `t' for unsigned 5-bit integer
6832 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
6834 (imm5_operand_dictunpack16): Likewise.
6835 (imm5_operand_dictunpack32): Likewise.
6836 (imm2_operand): Likewise.
6837 * doc/invoke.texi (SPARC Options): Document -mvis4b.
6838 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
6839 ditunpack* and fpcmp*shl builtins.
6841 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6843 * config.gcc: Handle m8 in --with-{cpu,tune} options.
6844 * config.in: Add HAVE_AS_SPARC6 define.
6845 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
6847 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
6849 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
6850 (CPP_CPU_SPEC): Handle m8.
6851 (ASM_CPU_SPEC): Likewise.
6852 * config/sparc/sparc-opts.h (enum processor_type): Add
6854 * config/sparc/sparc.c (m8_costs): New struct.
6855 (sparc_option_override): Handle TARGET_CPU_m8.
6856 (sparc32_initialize_trampoline): Likewise.
6857 (sparc64_initialize_trampoline): Likewise.
6858 (sparc_issue_rate): Likewise.
6859 (sparc_register_move_cost): Likewise.
6860 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
6861 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
6862 (ASM_CPU64_DEFAULT_SPEC): Likewise.
6863 (CPP_CPU_SPEC): Handle M8.
6864 (ASM_CPU_SPEC): Likewise.
6865 (AS_M8_FLAG): Define.
6866 * config/sparc/sparc.md: Add m8 to the cpu attribute.
6867 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
6868 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
6870 * configure: Regenerate.
6871 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
6874 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6876 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
6878 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
6879 ("*movdi_insn_sp32"): Do not set v3pipe.
6880 ("*movsi_insn"): Likewise.
6881 ("*movdi_insn_sp64"): Likewise.
6882 ("*movsf_insn"): Likewise.
6883 ("*movdf_insn_sp32"): Likewise.
6884 ("*movdf_insn_sp64"): Likewise.
6885 ("*zero_extendsidi2_insn_sp64"): Likewise.
6886 ("*sign_extendsidi2_insn"): Likewise.
6887 ("*mov<VM32:mode>_insn"): Likewise.
6888 ("*mov<VM64:mode>_insn_sp64"): Likewise.
6889 ("*mov<VM64:mode>_insn_sp32"): Likewise.
6890 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
6891 ("<vlop:code><VL:mode>3"): Likewise.
6892 ("*not_<vlop:code><VL:mode>3"): Likewise.
6893 ("*nand<VL:mode>_vis"): Likewise.
6894 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
6895 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
6896 ("one_cmpl<VL:mode>2"): Likewise.
6897 ("faligndata<VM64:mode>_vis"): Likewise.
6898 ("alignaddrsi_vis"): Likewise.
6899 ("alignaddrdi_vis"): Likweise.
6900 ("alignaddrlsi_vis"): Likewise.
6901 ("alignaddrldi_vis"): Likewise.
6902 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
6903 ("bmaskdi_vis"): Likewise.
6904 ("bmasksi_vis"): Likewise.
6905 ("bshuffle<VM64:mode>_vis"): Likewise.
6906 ("cmask8<P:mode>_vis"): Likewise.
6907 ("cmask16<P:mode>_vis"): Likewise.
6908 ("cmask32<P:mode>_vis"): Likewise.
6909 ("pdistn<P:mode>_vis"): Likewise.
6910 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
6912 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6914 * config/sparc/sparc.md ("subtype"): New insn attribute.
6915 ("*wrgsr_sp64"): Set insn subtype.
6916 ("*rdgsr_sp64"): Likewise.
6917 ("alignaddrsi_vis"): Likewise.
6918 ("alignaddrdi_vis"): Likewise.
6919 ("alignaddrlsi_vis"): Likewise.
6920 ("alignaddrldi_vis"): Likewise.
6921 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
6922 ("fexpand_vis"): Likewise.
6923 ("fpmerge_vis"): Likewise.
6924 ("faligndata<VM64:mode>_vis"): Likewise.
6925 ("bshuffle<VM64:mode>_vis"): Likewise.
6926 ("cmask8<P:mode>_vis"): Likewise.
6927 ("cmask16<P:mode>_vis"): Likewise.
6928 ("cmask32<P:mode>_vis"): Likewise.
6929 ("fchksm16_vis"): Likewise.
6930 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
6931 ("fmean16_vis"): Likewise.
6932 ("fp<plusminus_insn>64_vis"): Likewise.
6933 ("<plusminus_insn>v8qi3"): Likewise.
6934 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
6935 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
6936 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
6937 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
6938 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
6939 ("*movqi_insn"): Likewise.
6940 ("*movhi_insn"): Likewise.
6941 ("*movsi_insn"): Likewise.
6942 ("movsi_pic_gotdata_op"): Likewise.
6943 ("*movdi_insn_sp32"): Likewise.
6944 ("*movdi_insn_sp64"): Likewise.
6945 ("movdi_pic_gotdata_op"): Likewise.
6946 ("*movsf_insn"): Likewise.
6947 ("*movdf_insn_sp32"): Likewise.
6948 ("*movdf_insn_sp64"): Likewise.
6949 ("*zero_extendhisi2_insn"): Likewise.
6950 ("*zero_extendqihi2_insn"): Likewise.
6951 ("*zero_extendqisi2_insn"): Likewise.
6952 ("*zero_extendqidi2_insn"): Likewise.
6953 ("*zero_extendhidi2_insn"): Likewise.
6954 ("*zero_extendsidi2_insn_sp64"): Likewise.
6955 ("ldfsr"): Likewise.
6956 ("prefetch_64"): Likewise.
6957 ("prefetch_32"): Likewise.
6958 ("tie_ld32"): Likewise.
6959 ("tie_ld64"): Likewise.
6960 ("*tldo_ldub_sp32"): Likewise.
6961 ("*tldo_ldub1_sp32"): Likewise.
6962 ("*tldo_ldub2_sp32"): Likewise.
6963 ("*tldo_ldub_sp64"): Likewise.
6964 ("*tldo_ldub1_sp64"): Likewise.
6965 ("*tldo_ldub2_sp64"): Likewise.
6966 ("*tldo_ldub3_sp64"): Likewise.
6967 ("*tldo_lduh_sp32"): Likewise.
6968 ("*tldo_lduh1_sp32"): Likewise.
6969 ("*tldo_lduh_sp64"): Likewise.
6970 ("*tldo_lduh1_sp64"): Likewise.
6971 ("*tldo_lduh2_sp64"): Likewise.
6972 ("*tldo_lduw_sp32"): Likewise.
6973 ("*tldo_lduw_sp64"): Likewise.
6974 ("*tldo_lduw1_sp64"): Likewise.
6975 ("*tldo_ldx_sp64"): Likewise.
6976 ("*mov<VM32:mode>_insn"): Likewise.
6977 ("*mov<VM64:mode>_insn_sp64"): Likewise.
6978 ("*mov<VM64:mode>_insn_sp32"): Likewise.
6980 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6982 * config/sparc/sparc.md ("type"): New insn type viscmp.
6983 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
6985 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
6986 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
6987 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
6988 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
6990 ("n7_vis_logical_11cycle"): Likewise.
6991 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
6992 * config/sparc/niagara2.md ("niag3_vis": Likewise.
6993 * config/sparc/niagara.md ("niag_vis"): Likewise.
6994 * config/sparc/ultra3.md ("us3_fga"): Likewise.
6995 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
6997 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6999 * config/sparc/sparc.md: New instruction type `bmask'.
7000 (bmaskdi_vis): Use the `bmask' type.
7001 (bmasksi_vis): Likewise.
7002 * config/sparc/ultra3.md (us3_array): Likewise.
7003 * config/sparc/niagara7.md (n7_array): Likewise.
7004 * config/sparc/niagara4.md (n4_array): Likewise.
7005 * config/sparc/niagara2.md (niag2_vis): Likewise.
7006 (niag3_vis): Likewise.
7007 * config/sparc/niagara.md (niag_vis): Likewise.
7009 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
7011 * ipa-comdats.c: Remove optimize check from gate.
7012 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
7013 for functions not optimized.
7014 (ipa_fn_summary_read): Skip optimize check.
7015 (ipa_fn_summary_write): Likewise.
7016 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
7018 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
7020 (can_inline_edge_p): Check flag_pcc_struct_return for match.
7021 (check_callers): Give up on caller which is not optimized.
7022 (inline_small_functions): Likewise.
7023 (ipa_inline): Do not give up when not optimizing.
7024 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
7025 away unoptimizes cdtors.
7026 (whole_program_function_and_variable_visibility): Do
7027 ipa_discover_readonly_nonaddressable_vars in LTO mode.
7028 * ipa.c (process_references): Do not check optimize.
7029 (symbol_table::remove_unreachable_nodes): Update optimize check.
7030 (set_writeonly_bit): Update optimize check.
7031 (pass_ipa_cdtor_merge::gate): Do not check optimize.
7032 (pass_ipa_single_use::gate): Remove.
7034 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
7036 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
7037 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
7038 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
7039 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
7040 permute_load, permute_store, adjust_extract, adjust_splat,
7041 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
7042 replace_swap_with_copy, dump_swap_insn_table,
7043 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
7044 recombine_lvx_pattern, recombine_stvx_pattern,
7045 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
7046 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
7047 to file rs6000-p8swap.c.
7048 * config/rs6000/rs6000-p8swap.c: New file.
7049 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
7050 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
7051 and rs6000*-*-* targets.
7053 2017-07-06 David Malcolm <dmalcolm@redhat.com>
7055 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
7057 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
7059 * lto-wrapper.c (merge_and_complain): Do not merge
7060 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
7061 fsigned_zeros, ftrapping_math, fwrapv.
7062 (append_compiler_options): Do not track these options.
7063 (append_linker_options): Likewie
7065 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
7067 * cgraphunit.c (cgraph_node::finalize_function): When
7068 !flag_toplevel_reorde set no_reorder flag.
7069 (varpool_node::finalize_decl): Likewise.
7070 (symbol_table::compile): Drop no toplevel reorder path.
7072 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
7074 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
7075 edges; zero probability is not better than uninitialized.
7077 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
7079 * asan.h (asan_sanitize_allocas_p): Declare.
7080 * asan.c (asan_sanitize_allocas_p): New function.
7081 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
7082 (handle_builtin_alloca): Likewise.
7083 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
7084 if !asan_sanitize_allocas_p.
7085 * params.def (asan-instrument-allocas): Add new option.
7086 * params.h (ASAN_PROTECT_ALLOCAS): Define.
7087 * opts.c (common_handle_option): Disable allocas sanitization for
7090 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
7092 * asan.c: Include gimple-fold.h.
7093 (get_last_alloca_addr): New function.
7094 (handle_builtin_stackrestore): Likewise.
7095 (handle_builtin_alloca): Likewise.
7096 (asan_emit_allocas_unpoison): Likewise.
7097 (get_mem_refs_of_builtin_call): Add new parameter, remove const
7098 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
7099 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
7100 (instrument_builtin_call): Pass gimple iterator to
7101 get_mem_refs_of_builtin_call.
7102 (last_alloca_addr): New global.
7103 * asan.h (asan_emit_allocas_unpoison): Declare.
7104 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
7105 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
7106 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
7107 if function calls alloca.
7108 * gimple-fold.c (replace_call_with_value): Remove static keyword.
7109 * gimple-fold.h (replace_call_with_value): Declare.
7110 * internal-fn.c: Include asan.h.
7111 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
7112 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
7114 2017-07-06 David Malcolm <dmalcolm@redhat.com>
7116 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
7117 (C_SELFTEST_FLAGS): New.
7118 (CPP_SELFTEST_FLAGS): New.
7119 (SELFTEST_DEPS): New, from deps of s-selftest.
7120 (C_SELFTEST_DEPS): New, from deps of s-selftest.
7121 (CPP_SELFTEST_DEPS): New.
7122 (selftest): Add dependency on s-selftest-c++.
7123 (s-selftest): Rename to...
7124 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
7125 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
7126 than SELFTEST_FLAGS.
7127 (selftest-gdb): Rename to...
7128 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
7130 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
7131 (selftest-valgrind): Rename to...
7132 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
7134 (selftest-valgrind): Reintroduce as an alias for
7135 selftest-c-valgrind.
7136 (s-selftest-c++): New.
7137 (selftest-c++-gdb): New.
7138 (selftest-c++-valgrind): New.
7140 2017-07-06 Olivier Hainque <hainque@adacore.com>
7142 * gcc.c (process_command): When deciding if undefined variables
7143 should be ignored when processing specs, accept "gcc -v" as well.
7145 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
7147 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
7148 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
7150 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
7152 * config/arm/arm-cpus.in (armv8-r): Add new entry.
7153 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
7154 * config/arm/arm-tables.opt: Regenerate.
7155 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
7157 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
7159 2017-07-06 Carl Love <cel@us.ibm.com>
7161 * ChangeLog: Clean up from mid air collision
7163 2017-07-06 Carl Love <cel@us.ibm.com>
7165 * config/rs6000/rs6000-c.c: Add support for built-in functions
7166 vector signed int vec_subc (vector signed int, vector signed int);
7167 vector signed __int128 vec_subc (vector signed __int128,
7168 vector signed __int128);
7169 vector unsigned __int128 vec_subc (vector unsigned __int128,
7170 vector unsigned __int128);
7171 vector signed int vec_sube (vector signed int, vector signed int,
7173 vector unsigned int vec_sube (vector unsigned int,
7174 vector unsigned int,
7175 vector unsigned int);
7176 vector signed __int128 vec_sube (vector signed __int128,
7177 vector signed __int128,
7178 vector signed__int128);
7179 vector unsigned __int128 vec_sube (vector unsigned __int128,
7180 vector unsigned __int128,
7181 vector unsigned __int128);
7182 vector signed int vec_subec (vector signed int, vector signed int,
7184 vector unsigned int vec_subec (vector unsigned int,
7185 vector unsigned int,
7186 vector unsigned int);
7187 vector signed __int128 vec_subec (vector signed __int128,
7188 vector signed __int128,
7189 vector signed__int128);
7190 vector unsigned __int128 vec_subec (vector unsigned __int128,
7191 vector unsigned __int128,
7192 vector unsigned __int128);
7193 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
7194 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
7195 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
7196 BU_ALTIVEC_OVERLOAD_X definitions.
7197 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
7198 * doc/extend.texi: Update the built-in documentation file for the new
7201 2017-07-06 David Malcolm <dmalcolm@redhat.com>
7204 * diagnostic-show-locus.c (layout::layout): Use start and finish
7205 spelling location for the start and finish of each range.
7206 * genmatch.c (linemap_client_expand_location_to_spelling_point):
7207 Add unused aspect param.
7208 * input.c (expand_location_1): Add "aspect" param, and use it
7209 to access the correct part of the location.
7210 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
7212 (expand_location_to_spelling_point): Likewise.
7213 (linemap_client_expand_location_to_spelling_point): Add "aspect"
7214 param, and pass it to expand_location_1.
7216 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
7218 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
7219 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
7220 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
7221 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
7222 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
7223 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
7224 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
7225 _mm_maskz_getmant_ss): New intrinsics.
7226 (__builtin_ia32_getexpss128_mask): Changed to ...
7227 __builtin_ia32_getexpss128_round ... this.
7228 (__builtin_ia32_getexpsd128_mask): Changed to ...
7229 __builtin_ia32_getexpsd128_round ... this.
7230 * config/i386/i386-builtin-types.def
7231 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
7232 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
7233 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
7234 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
7235 __builtin_ia32_getmantss_mask_round): New builtins.
7236 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
7237 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
7238 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
7239 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
7240 * config/i386/sse.md
7241 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
7242 avx512f_sgetexp<mode><mask_scalar_name>
7243 <round_saeonly_scalar_name> ... this.
7244 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
7245 %0, %1, %2<round_saeonly_op3>}): Changed to ...
7246 vgetexp<ssescalarmodesuffix>
7247 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
7248 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
7249 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
7250 avx512f_vgetmant<mode><mask_scalar_name>
7251 <round_saeonly_scalar_name> ... this.
7252 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
7253 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
7254 vgetmant<ssescalarmodesuffix>
7255 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
7256 %0<mask_scalar_operand4>, %1, %2
7257 <round_saeonly_scalar_mask_op4>, %3} ... this.
7258 * config/i386/subst.md (mask_scalar_operand4,
7259 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
7260 round_saeonly_scalar_nimm_predicate): New subst attributes.
7262 2017-07-06 Julia Koval <julia.koval@intel.com>
7264 * config/i386/i386.c (ix86_erase_embedded_rounding):
7265 Remove code for old rounding pattern.
7267 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
7269 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
7271 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
7273 * doc/sourcebuild.texi (Test Directives, Variants of
7274 dg-require-support): Add documentation for dg-require-stack-check.
7276 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
7278 * config/i386/subst.md (mask_scalar, round_scalar,
7279 round_saeonly_scalar): New meta-templates.
7280 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
7281 round_scalar_mask_operand3, round_scalar_mask_op3,
7282 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
7283 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
7284 round_saeonly_scalar_constraint,
7285 round_saeonly_scalar_prefix): New subst attribute.
7286 * config/i386/sse.md
7287 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
7288 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
7289 <round_scalar_name> ... this.
7290 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
7291 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
7292 <round_scalar_name> ... this.
7293 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
7294 <sse>_vm<code><mode>3<mask_scalar_name>
7295 <round_saeonly_scalar_name> ... this.
7296 (v<plusminus_mnemonic><ssescalarmodesuffix>
7297 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
7298 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
7299 v<plusminus_mnemonic><ssescalarmodesuffix>
7300 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
7301 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
7302 (v<multdiv_mnemonic><ssescalarmodesuffix>
7303 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
7304 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
7305 v<multdiv_mnemonic><ssescalarmodesuffix>
7306 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
7307 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
7308 (v<maxmin_float><ssescalarmodesuffix>
7309 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
7310 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
7311 v<maxmin_float><ssescalarmodesuffix>
7312 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
7313 %0<mask_scalar_operand3>, %1, %<iptr>2
7314 <round_saeonly_scalar_mask_op3>} ... this.
7316 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
7318 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
7319 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
7321 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
7322 Alan Hayward <alan.hayward@arm.com>
7323 David Sherwood <david.sherwood@arm.com>
7325 * combine.c (simplify_if_then_else): Remove "enum" before
7327 * compare-elim.c (can_eliminate_compare): Likewise.
7328 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
7330 (aarch64_lookup_simd_builtin_type): Likewise.
7331 (aarch64_simd_builtin_type): Likewise.
7332 (aarch64_init_simd_builtin_types): Likewise.
7333 (aarch64_simd_expand_args): Likewise.
7334 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
7336 (aarch64_reverse_mask): Likewise.
7337 (aarch64_simd_emit_reg_reg_move): Likewise.
7338 (aarch64_gen_adjusted_ldpstp): Likewise.
7339 (aarch64_ccmp_mode_to_code): Likewise.
7340 (aarch64_operands_ok_for_ldpstp): Likewise.
7341 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
7342 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
7344 (aarch64_min_divisions_for_recip_mul): Likewise.
7345 (aarch64_reassociation_width): Likewise.
7346 (aarch64_get_condition_code_1): Likewise.
7347 (aarch64_simd_emit_reg_reg_move): Likewise.
7348 (aarch64_simd_attr_length_rglist): Likewise.
7349 (aarch64_reverse_mask): Likewise.
7350 (aarch64_operands_ok_for_ldpstp): Likewise.
7351 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
7352 (aarch64_gen_adjusted_ldpstp): Likewise.
7353 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
7355 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
7356 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
7357 (arm_lookup_simd_builtin_type): Likewise.
7358 (arm_simd_builtin_type): Likewise.
7359 (arm_init_simd_builtin_types): Likewise.
7360 (arm_expand_builtin_args): Likewise.
7361 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
7362 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
7363 (ft32_setup_incoming_varargs): Likewise.
7364 (ft32_function_arg): Likewise.
7365 (ft32_function_arg_advance): Likewise.
7366 (ft32_pass_by_reference): Likewise.
7367 (ft32_arg_partial_bytes): Likewise.
7368 (ft32_valid_pointer_mode): Likewise.
7369 (ft32_addr_space_pointer_mode): Likewise.
7370 (ft32_addr_space_legitimate_address_p): Likewise.
7371 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
7373 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
7374 (ix86_emit_outlined_ms2sysv_restore): Likewise.
7375 (iamcu_alignment): Likewise.
7376 (canonicalize_vector_int_perm): Likewise.
7377 (ix86_noce_conversion_profitable_p): Likewise.
7378 (ix86_mpx_bound_mode): Likewise.
7379 (ix86_operands_ok_for_move_multiple): Likewise.
7380 * config/microblaze/microblaze-protos.h
7381 (microblaze_expand_conditional_branch_reg): Likewise.
7382 * config/microblaze/microblaze.c
7383 (microblaze_expand_conditional_branch_reg): Likewise.
7384 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
7386 (rs6000_reassociation_width): Likewise.
7387 (rs6000_invalid_binary_op): Likewise.
7388 (fusion_p9_p): Likewise.
7389 (emit_fusion_p9_load): Likewise.
7390 (emit_fusion_p9_store): Likewise.
7391 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
7393 (riscv_hard_regno_mode_ok_p): Likewise.
7394 (riscv_address_insns): Likewise.
7395 (riscv_split_symbol): Likewise.
7396 (riscv_legitimize_move): Likewise.
7397 (riscv_function_value): Likewise.
7398 (riscv_hard_regno_nregs): Likewise.
7399 (riscv_expand_builtin): Likewise.
7400 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
7401 (riscv_build_integer): Likewise.
7402 (riscv_split_integer): Likewise.
7403 (riscv_legitimate_constant_p): Likewise.
7404 (riscv_cannot_force_const_mem): Likewise.
7405 (riscv_regno_mode_ok_for_base_p): Likewise.
7406 (riscv_valid_base_register_p): Likewise.
7407 (riscv_valid_offset_p): Likewise.
7408 (riscv_valid_lo_sum_p): Likewise.
7409 (riscv_classify_address): Likewise.
7410 (riscv_legitimate_address_p): Likewise.
7411 (riscv_address_insns): Likewise.
7412 (riscv_load_store_insns): Likewise.
7413 (riscv_force_binary): Likewise.
7414 (riscv_split_symbol): Likewise.
7415 (riscv_force_address): Likewise.
7416 (riscv_legitimize_address): Likewise.
7417 (riscv_move_integer): Likewise.
7418 (riscv_legitimize_const_move): Likewise.
7419 (riscv_legitimize_move): Likewise.
7420 (riscv_address_cost): Likewise.
7421 (riscv_subword): Likewise.
7422 (riscv_output_move): Likewise.
7423 (riscv_canonicalize_int_order_test): Likewise.
7424 (riscv_emit_int_order_test): Likewise.
7425 (riscv_function_arg_boundary): Likewise.
7426 (riscv_pass_mode_in_fpr_p): Likewise.
7427 (riscv_pass_fpr_single): Likewise.
7428 (riscv_pass_fpr_pair): Likewise.
7429 (riscv_get_arg_info): Likewise.
7430 (riscv_function_arg): Likewise.
7431 (riscv_function_arg_advance): Likewise.
7432 (riscv_arg_partial_bytes): Likewise.
7433 (riscv_function_value): Likewise.
7434 (riscv_pass_by_reference): Likewise.
7435 (riscv_setup_incoming_varargs): Likewise.
7436 (riscv_print_operand): Likewise.
7437 (riscv_elf_select_rtx_section): Likewise.
7438 (riscv_save_restore_reg): Likewise.
7439 (riscv_for_each_saved_reg): Likewise.
7440 (riscv_register_move_cost): Likewise.
7441 (riscv_hard_regno_mode_ok_p): Likewise.
7442 (riscv_hard_regno_nregs): Likewise.
7443 (riscv_class_max_nregs): Likewise.
7444 (riscv_memory_move_cost): Likewise.
7445 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
7446 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
7447 (rl78_addr_space_address_mode): Likewise.
7448 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7450 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
7451 (rs6000_reassociation_width): Likewise.
7452 (rs6000_invalid_binary_op): Likewise.
7453 (fusion_p9_p): Likewise.
7454 (emit_fusion_p9_load): Likewise.
7455 (emit_fusion_p9_store): Likewise.
7456 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
7457 (ok_for_simple_move_operands): Likewise.
7458 (ok_for_simple_move_strict_operands): Likewise.
7459 (ok_for_simple_arith_logic_operands): Likewise.
7460 (visium_legitimize_reload_address): Likewise.
7461 (visium_select_cc_mode): Likewise.
7462 (output_cbranch): Likewise.
7463 (visium_split_double_move): Likewise.
7464 (visium_expand_copysign): Likewise.
7465 (visium_expand_int_cstore): Likewise.
7466 (visium_expand_fp_cstore): Likewise.
7467 * config/visium/visium.c (visium_pass_by_reference): Likewise.
7468 (visium_function_arg): Likewise.
7469 (visium_function_arg_advance): Likewise.
7470 (visium_libcall_value): Likewise.
7471 (visium_setup_incoming_varargs): Likewise.
7472 (visium_legitimate_constant_p): Likewise.
7473 (visium_legitimate_address_p): Likewise.
7474 (visium_legitimize_address): Likewise.
7475 (visium_secondary_reload): Likewise.
7476 (visium_register_move_cost): Likewise.
7477 (visium_memory_move_cost): Likewise.
7478 (prepare_move_operands): Likewise.
7479 (ok_for_simple_move_operands): Likewise.
7480 (ok_for_simple_move_strict_operands): Likewise.
7481 (ok_for_simple_arith_logic_operands): Likewise.
7482 (visium_function_value_1): Likewise.
7483 (rtx_ok_for_offset_p): Likewise.
7484 (visium_legitimize_reload_address): Likewise.
7485 (visium_split_double_move): Likewise.
7486 (visium_expand_copysign): Likewise.
7487 (visium_expand_int_cstore): Likewise.
7488 (visium_expand_fp_cstore): Likewise.
7489 (visium_split_cstore): Likewise.
7490 (visium_select_cc_mode): Likewise.
7491 (visium_split_cbranch): Likewise.
7492 (output_cbranch): Likewise.
7493 (visium_print_operand_address): Likewise.
7494 * expmed.c (flip_storage_order): Likewise.
7495 * expmed.h (emit_cstore): Likewise.
7496 (flip_storage_order): Likewise.
7497 * genrecog.c (validate_pattern): Likewise.
7498 * hsa-gen.c (gen_hsa_addr): Likewise.
7499 * internal-fn.c (expand_arith_overflow): Likewise.
7500 * ira-color.c (allocno_copy_cost_saving): Likewise.
7501 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7502 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
7503 (process_invariant_for_inheritance): Likewise.
7504 * lra-eliminations.c (move_plus_up): Likewise.
7505 * omp-low.c (lower_oacc_reductions): Likewise.
7506 * simplify-rtx.c (simplify_subreg): Likewise.
7507 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
7508 (TARGET_CHKP_BOUND_MODE): Likewise..
7509 * targhooks.c (default_chkp_bound_mode): Likewise.
7510 (default_setup_incoming_vararg_bounds): Likewise.
7511 * targhooks.h (default_chkp_bound_mode): Likewise.
7512 (default_setup_incoming_vararg_bounds): Likewise.
7513 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
7514 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
7515 (have_whole_vector_shift): Likewise.
7516 * tree-vect-stmts.c (vectorizable_load): Likewise.
7517 * doc/tm.texi: Regenerate.
7519 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
7521 Graceful degrade if Binutils PR21472 is not available.
7524 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
7525 .rodata in flash test fails.
7526 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
7527 * confgure: Regenerate.
7528 * config.in: Regenerate.
7529 * config/avr/avr.c (avr_asm_named_section)
7530 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
7531 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
7532 (avr_asm_init_sections): Same.
7534 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
7536 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
7537 (fma<VH:mode>4_intrinsic): Likewise.
7538 (*fmsub<VCVTF:mode>4): Likewise.
7539 (*fmsub<VH:mode>4_intrinsic): Likewise.
7541 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
7544 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
7545 Don't depend on "optimize > 0".
7546 (out_movhi_r_mr, out_movqi_mr_r): Same.
7547 (out_movhi_mr_r, out_movqi_r_mr): Same.
7548 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
7549 io_address_operand on "optimize > 0".
7551 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7553 * tree-loop-distribution.c: Add general explanantion on the pass.
7554 (generate_loops_for_partition): Mark distributed loop.
7555 (pg_add_dependence_edges): New parameter. Handle alias data
7556 dependence specially and record it in the parameter if asked.
7557 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
7558 (init_partition_graph_vertices, add_partition_graph_edge): New.
7559 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
7560 (free_partition_graph_vdata, build_partition_graph): New.
7561 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
7562 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
7563 (data_ref_segment_size, latch_dominated_by_data_ref): New.
7564 (compute_alias_check_pairs, version_loop_by_alias_check): New.
7565 (version_for_distribution_p, finalize_partitions): New.
7566 (distribute_loop): Handle alias data dependence specially. Factor
7567 out loop fusion code as functions and call these functions.
7569 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7571 * tree-loop-distribution.c (classify_partition): New parameter and
7572 better handle reduction statement.
7573 (rdg_build_partitions): Revise comment.
7574 (distribute_loop): Compute statements in all partitions and pass it
7575 to classify_partition.
7577 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7579 * tree-loop-distribution.c (enum partition_type): New.
7580 (struct partition): New field type.
7581 (partition_merge_into): Add parameter. Update partition type.
7582 (data_dep_in_cycle_p, update_type_for_merge): New functions.
7583 (build_rdg_partition_for_vertex): Compute partition type.
7584 (rdg_build_partitions): Dump partition type.
7585 (distribute_loop): Update calls to partition_merge_into.
7587 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7589 * tree-loop-distribution.c (struct ddr_hasher): New.
7590 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
7592 (classify_partition): Call get_data_dependence.
7593 (pg_add_dependence_edges): Ditto.
7594 (distribute_loop): Release data dependence hash table.
7596 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7598 * tree-loop-distribution.c (ref_base_address): Delete.
7599 (similar_memory_accesses): Rename ...
7600 (share_memory_accesses): ... to this. Check if partitions access
7601 the same memory reference.
7602 (distribute_loop): Call share_memory_accesses.
7604 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7606 * tree-loop-distribution.c (struct partition): New field recording
7608 (partition_alloc, partition_free): Init and release data refs.
7609 (partition_merge_into): Merge data refs.
7610 (build_rdg_partition_for_vertex): Collect data refs for partition.
7611 (pg_add_dependence_edges): Change parameters from vector to bitmap.
7613 (distribute_loop): Remve data refs from vertice data of partition
7616 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7618 * tree-loop-distribution.c (params.h): Include header file.
7619 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
7620 (datarefs_vec): New global var.
7621 (create_rdg_vertices): Use datarefs_vec directly.
7622 (free_rdg): Don't free data references.
7623 (build_rdg): Update use. Don't free data references.
7624 (distribute_loop): Compute global variable for data references.
7625 Bail out if there are too many data references.
7627 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7629 * tree-loop-distribution.c (loop_nest): New global var.
7630 (build_rdg): Use loop directly, rather than loop nest.
7631 (pg_add_dependence_edges): Remove loop nest parameter. Use global
7633 (distribute_loop): Compute global variable loop nest. Update use.
7635 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7637 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
7638 (partition_merge_into): New parameter. Dump reason for fusion.
7639 (distribute_loop): Update use of partition_merge_into.
7641 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7643 * tree-loop-distribution.c (bb_top_order_index): New.
7644 (bb_top_order_index_size, bb_top_order_cmp): New.
7645 (stmts_from_loop): Use topological order.
7646 (pass_loop_distribution::execute): Compute and release topological
7647 order for basic blocks.
7649 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7651 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
7654 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7656 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
7657 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
7658 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
7659 * internal-fn.def (LOOP_DIST_ALIAS): New.
7660 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
7661 (fold_loop_internal_call): ... this.
7662 (vect_loop_dist_alias_call): New function.
7663 (set_uid_loop_bbs): Call fold_loop_internal_call.
7664 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
7667 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
7670 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
7671 Require dead FLAGS_REG at the beginning of a peephole.
7673 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
7676 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
7677 arguments in the call to __builtin_ia32_sbb_u32.
7678 (_subborrow_u64): Swap _X and _Y arguments in the call to
7679 __builtin_ia32_sbb_u64.
7681 2017-07-04 Jakub Jelinek <jakub@redhat.com>
7684 * tree-vrp.c (compare_assert_loc): Turn into a function template
7685 with stable template parameter. Only test if a->e is NULL,
7686 !a->e == !b->e has been verified already. Use e == NULL or
7687 e != NULL instead of e or ! e tests. If stable is true, don't use
7688 iterative_hash_expr, on the other side allow a or b or both NULL
7689 and sort the NULLs last.
7690 (process_assert_insertions): Sort using compare_assert_loc<false>
7691 instead of compare_assert_loc, later sort using
7692 compare_assert_loc<true> before calling process_assert_insertions_for
7693 in a loop. Use break instead of continue once seen NULL pointer.
7695 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
7697 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
7698 Cortex-R7 and Cortex-R8 processors.
7700 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
7702 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
7703 uninitialized while src is not.
7705 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
7707 * common/config/arm/arm-common.c: Adjust include path for
7709 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
7710 (arm-cpu.h): Create in build directory. Adjust dependency rules.
7711 (arm-cpu-data.h): Likewise.
7712 (arm-cpu-cdata.h): Likewise.
7713 * config/arm/arm-cpu.h: Delete.
7714 * config/arm/arm-cpu-cdata.h: Delete.
7715 * config/arm/arm-cpu-data.h: Delete.
7717 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
7719 * config/arm/arm-cpus.in (cortex-a55): New.
7720 (cortex-a75): Likewise.
7721 (cortex-a75.cortex-a55): Likewise.
7722 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
7724 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
7725 * config/arm/arm-cpu-cdata.h: Regenerate.
7726 * config/arm/arm-cpu-data.h: Regenerate.
7727 * config/arm/arm-cpu.h: Regenerate.
7728 * config/arm/arm-tables.opt: Regenerate.
7729 * config/arm/arm-tune.md: Regenerate.
7731 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
7733 * haifa-sched.c (sched_create_recovery_edges): Update profile.
7735 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
7737 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
7740 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
7742 PR tree-optimization/81292
7743 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
7744 full_string_p, also call adjust_related_strinfos if the adjustment
7745 is simple, otherwise invalidate related strinfos.
7747 2017-07-04 Martin Liska <mliska@suse.cz>
7750 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
7751 newly created variable as DECL_IGNORED_P.
7753 2017-07-04 Martin Liska <mliska@suse.cz>
7756 * ipa-inline.c (inline_small_functions):
7757 Use xstrdup_for_dump.
7759 2017-07-04 Tom de Vries <tom@codesourcery.com>
7761 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
7763 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
7766 * config/darwin.c (darwin_function_switched_text_sections):
7769 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
7771 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
7773 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
7775 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
7777 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7779 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
7780 min_profitable_iters, and th as inclusive lower bounds.
7781 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
7782 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
7783 for min_profitable_iters and min_profitable_estimate.
7784 (vect_transform_loop): Treat th as an inclusive lower bound.
7785 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
7787 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
7790 * config/darwin.c (darwin_function_switched_text_sections):
7791 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
7792 in two pieces, and suppress the use of buf.
7794 2017-07-03 Nathan Sidwell <nathan@acm.org>
7796 * hash-table.h (hash_table_mod1): Fix indentation.
7798 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7801 * predict.c (force_edge_cold): Be more careful about propagation
7803 * profile-count.h (profile_probability::guessed,
7804 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
7806 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
7808 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
7810 * doc/invoke.texi (rcpc architecture extension): Document it.
7812 2017-07-03 Richard Biener <rguenther@suse.de>
7814 PR tree-optimization/60510
7815 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
7816 the scalar reduction PHI and use it.
7817 (vectorizable_reduction): Properly guard the single_defuse_cycle
7818 path for non-SLP reduction chains where we cannot use it.
7819 Rework reduc_def/index and vector type deduction. Rework
7820 vector operand gathering during reduction op code-gen.
7821 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
7822 chains dissolve the chain and leave it to non-SLP reduction
7825 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7827 * tree-data-ref.h (dr_alignment): Declare.
7828 * tree-data-ref.c (dr_alignment): New function.
7829 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
7830 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
7832 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
7834 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7836 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
7837 and base_misalignment fields.
7838 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
7839 * tree-data-ref.c: Include builtins.h.
7840 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
7841 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
7842 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
7843 * tree-vect-data-refs.c: Include tree-cfg.h.
7844 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
7845 fields instead of calculating an alignment here.
7846 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
7847 innermost_loop_behavior fields.
7849 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7851 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
7853 (DR_STEP_ALIGNMENT): New macro.
7854 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
7855 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
7856 (create_data_ref): Print it.
7857 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
7858 to tell whether the step preserves vector (mis)alignment.
7859 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
7860 Move the check for an integer step and generalise to all INTEGER_CST.
7861 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
7862 Print the outer step alignment.
7864 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7866 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
7867 with offset_alignment.
7868 (DR_ALIGNED_TO): Delete.
7869 (DR_OFFSET_ALIGNMENT): New macro.
7870 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
7871 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
7872 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
7873 (create_data_ref): Likewise.
7874 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
7875 (vect_analyze_data_refs): Likewise.
7876 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
7877 creating dummy innermost behavior.
7879 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7881 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
7882 with a "innermost_loop_behavior *" and refeence tree.
7883 * tree-data-ref.c (dr_analyze_innermost): Likewise.
7884 (create_data_ref): Update call accordingly.
7885 * tree-predcom.c (find_looparound_phi): Likewise.
7887 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7889 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
7890 fields with dr_wrt_vec_loop.
7891 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
7892 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
7893 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
7894 (vect_dr_behavior): New function.
7895 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
7896 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
7897 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
7898 track whether the step preserves the misalignment.
7899 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
7900 Use vect_dr_behavior.
7901 (vect_setup_realignment): Update call accordingly.
7902 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
7903 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
7904 call to vect_create_addr_base_for_vector_ref.
7905 (vect_create_cond_for_align_checks): Likewise.
7906 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
7907 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
7908 (vect_recog_mask_conversion_pattern): Likewise.
7909 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
7910 (new_stmt_vec_info): Remove redundant zeroing.
7912 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
7914 * common/config/arm/arm-common.c (arm_be8_option): New function.
7915 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
7916 (ISA_ARMv6): Add isa_bit_be8.
7917 * config/arm/arm.h (arm_be8_option): Add prototype.
7918 (BE8_SPEC_FUNCTION): New define.
7919 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
7920 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
7921 (mlittle-endian): Similarly.
7922 (mbe8, mbe32): New options.
7923 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
7924 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
7926 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7928 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
7930 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7932 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
7933 (cleanup_tree_cfg_bb): Use it.
7934 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
7936 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
7938 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7941 * loop-doloop.c (add_test): Update profile.
7943 2017-07-03 Martin Liska <mliska@suse.cz>
7946 * sanopt.c (rewrite_usage_of_param): New function.
7947 (sanitize_rewrite_addressable_params): Likewise.
7948 (pass_sanopt::execute): Call rewrite_usage_of_param.
7950 2017-07-03 Richard Biener <rguenther@suse.de>
7952 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
7953 back to using VIEW_CONVERT_EXPR.
7955 2017-07-03 Martin Liska <mliska@suse.cz>
7958 * doc/extend.texi: Document when a resolver function is
7959 generated for target_clones.
7961 2017-07-03 Martin Liska <mliska@suse.cz>
7963 * asan.c (asan_emit_stack_protection): Unpoison just red zones
7964 and shadow memory of auto variables which are subject of
7965 use-after-scope sanitization.
7966 (asan_expand_mark_ifn): Add do set only when is_poison.
7968 2016-07-03 Richard Biener <rguenther@suse.de>
7970 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
7972 (vect_force_simple_reduction): Record reduction def -> phi mapping.
7973 (vectorizable_reduction): Perform reduction PHI creation when
7974 visiting a reduction PHI and adjust and simplify code generation
7975 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
7976 (vect_transform_loop): Visit reduction PHIs.
7977 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
7978 defs into the SLP tree.
7979 (vect_build_slp_tree): Reduction defs terminate the recursion.
7980 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
7982 (vect_get_vec_defs_for_stmt_copy): Export.
7983 (vect_get_vec_defs): Likewise.
7984 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
7986 (vect_get_vec_defs_for_stmt_copy): Declare.
7987 (vect_get_vec_defs): Likewise.
7989 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7991 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
7992 parameter with a "loop" parameter and use it instead of the
7993 loop containing DR_STMT. Don't check simple_iv when doing
7994 BB analysis. Describe the two analysis modes in the comment.
7996 2017-07-03 Tom de Vries <tom@codesourcery.com>
7998 PR tree-optimization/69468
7999 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
8000 (find_same_succ_bb): Handle ignore_edge_flags.
8002 2017-07-03 Tom de Vries <tom@codesourcery.com>
8004 PR tree-optimization/81192
8005 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
8007 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
8009 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
8011 2017-07-03 Tom de Vries <tom@codesourcery.com>
8013 PR tree-optimization/81192
8014 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
8015 BB_SAME_SUCC (bb) == NULL.
8017 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8019 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
8022 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8024 * dumpfile.c: Include profile-count.h
8025 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
8027 (insert_cond_bb): Update profile.
8028 * tree-cfg.h (insert_cond_bb): Update prototype.
8029 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
8030 * tree-dump.c: Do not include tree-cfg.
8032 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8034 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
8036 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8038 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
8041 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8043 * tree-complex.c (expand_complex_div_wide): update profile.
8045 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
8046 Alan Hayward <alan.hayward@arm.com>
8047 David Sherwood <david.sherwood@arm.com>
8049 * Makefile.in (MACHMODE_H): Remove insn-modes.h
8050 (CORETYPES_H): New define.
8051 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
8052 (insn-modes-inline.h, s-modes-inline-h): New rules.
8053 (generated_files): Add insn-modes-inline.h.
8054 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
8055 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
8056 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
8057 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
8058 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
8059 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
8060 (build/gencodes.o, build/genconditions.o): Likewise.
8061 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
8062 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
8063 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
8064 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
8065 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
8066 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
8067 * coretypes.h: Include everything up to real.h for generators.
8068 Include insn-modes.h first. Include wide-int-print.h after
8069 wide-int.h. Include insn-modes-inline.h and then machmode.h.
8070 * machmode.h: Don't include insn-modes.h here.
8071 * function-tests.c: Remove includes of signop.h, machmode.h,
8072 double-int.h and wide-int.h.
8074 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
8076 * optc-save-gen.awk: Likewise.
8077 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
8078 * godump.c: Remove include of wide-int-print.h.
8079 * pretty-print.h: Likewise.
8080 * wide-int-print.cc: Likewise.
8081 * wide-int.cc: Likewise.
8082 * hash-map-tests.c: Remove include of signop.h.
8083 * hash-set-tests.c: Likewise.
8084 * rtl-tests.c: Likewise.
8085 * mkconfig.sh: Remove include of machmode.h.
8086 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
8088 (emit_insn_modes_inline_h): ...this new function. Emit the code
8089 into an insn-modes-inline.h header file, adding appropriate
8090 include guards and end comments.
8091 (emit_insn_modes_c_header): Remove include of machmode.h.
8092 (emit_min_insn_modes_c_header): Include coretypes.h rather than
8094 (main): Handle -i flag and call emit_insn_modes_inline_h when
8097 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
8099 * tree-ssa-strlen.c (strinfo): Rename the length field to
8100 nonzero_chars. Add a full_string_p field.
8101 (compare_nonzero_chars, zero_length_string_p): New functions.
8102 (get_addr_stridx): Add an offset_out parameter.
8103 Use compare_nonzero_chars.
8104 (get_stridx): Update accordingly. Use compare_nonzero_chars.
8105 (new_strinfo): Update after above changes to strinfo.
8106 (set_endptr_and_length): Set full_string_p.
8107 (get_string_length): Update after above changes to strinfo.
8108 (unshare_strinfo): Update call to new_strinfo.
8109 (maybe_invalidate): Likewise.
8110 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
8111 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
8112 as a uhwi instead of an shwi. Update after above changes to
8113 strinfo and new_strinfo.
8114 (zero_length_string): Assert that chainsi contains full strings.
8115 Use zero_length_string_p. Update call to new_strinfo.
8116 (adjust_related_strinfos): Update after above changes to strinfo.
8117 Copy full_string_p from origsi.
8118 (adjust_last_stmt): Use zero_length_string_p.
8119 (handle_builtin_strlen): Update after above changes to strinfo and
8120 new_strinfo. Install the lhs as the string length if the previous
8121 entry didn't describe a full string.
8122 (handle_builtin_strchr): Update after above changes to strinfo
8124 (handle_builtin_strcpy): Likewise.
8125 (handle_builtin_strcat): Likewise.
8126 (handle_builtin_malloc): Likewise.
8127 (handle_pointer_plus): Likewise.
8128 (handle_builtin_memcpy): Likewise. Track nonzero characters
8129 that aren't necessarily followed by a nul terminator.
8130 (handle_char_store): Likewise.
8132 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
8134 PR tree-optimization/80769
8135 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
8136 for malloc and calloc. Document the new invariant that all related
8137 strinfos have delayed lengths or none do.
8138 (verify_related_strinfos): Move earlier in file.
8139 (set_endptr_and_length): New function, split out from...
8140 (get_string_length): ...here. Also set the lengths of related
8142 (zero_length_string): Assert that chainsi has known (rather than
8144 (adjust_related_strinfos): Likewise.
8146 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
8148 PR tree-optimization/81136
8149 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
8150 assert that two references with the same misalignment have the same
8151 compile-time misalignment if those compile-time misalignments
8154 2017-07-01 Andi Kleen <ak@linux.intel.com>
8156 * print-tree.c (print_node): Print all attributes.
8158 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8160 * cfg.c (scale_bbs_frequencies): New function.
8161 * cfg.h (scale_bbs_frequencies): Declare it.
8162 * cfgloopanal.c (single_likely_exit): Cleanup.
8163 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
8165 (scale_loop_profile): Likewise.
8166 (loop_version): Likewise.
8167 (create_empty_loop_on_edge): Update.
8168 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
8169 scale_loop_frequencies, scale_loop_profile, loopify,
8170 loop_version): Update prototypes.
8171 * modulo-sched.c (sms_schedule): Update.
8172 * predict.c (unlikely_executed_edge_p): Also check probability.
8173 (probably_never_executed_edge_p): Fix typo.
8174 * tree-if-conv.c (version_loop_for_if_conversion): Update.
8175 * tree-parloops.c (gen_parallel_loop): Update.
8176 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
8177 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8178 * tree-ssa-loop-split.c (split_loop): Update.
8179 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
8180 * tree-vect-loop-manip.c (vect_do_peeling): Update.
8181 (vect_loop_versioning): Update.
8182 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8184 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8186 * trans-mem.c (split_bb_make_tm_edge): Update profile.
8188 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8190 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
8191 to keep profile consistent.
8193 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8195 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
8196 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
8197 * profile-count.h (max_safe_multiplier): Make unsigned.
8198 (profile_count::guessed_zero): New.
8200 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8202 * bb-reorder.c (fix_up_crossing_landing_pad,
8203 fix_crossing_conditional_branches): Use make_single_succ_edge
8204 to keep profile consistent.
8206 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8208 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
8211 2017-07-01 Jakub Jelinek <jakub@redhat.com>
8214 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
8215 the right scopes, make sure cond_jump isn't preserved between multiple
8216 iterations. Search for fallthru edge whenever there are 3+ edges and
8217 use find_fallthru_edge for it.
8219 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8221 Patch by Alexander Monakov <amonakov@ispras.ru>
8222 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
8223 probabilities consistently.
8225 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8227 * pa.c (pa_expand_compare_and_swap_loop): Update call of
8228 emit_cmp_and_jump_insns.
8230 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8233 * tree-inline.c (expand_call_inline): Combine profile statuses.
8235 2017-06-30 Andrew Pinski <apinski@cavium.com>
8237 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
8238 fold_stmt returned true.
8240 2017-06-30 Nathan Sidwell <nathan@acm.org>
8242 * ggc.h (empty_string): Delete.
8243 * cfgexpand.c (expand_asm_stmt): Use plain "".
8244 * optabs.c (expand_asm_memory_barrier): Likewise.
8245 * stringpool.c (empty_string): Delete.
8246 (digit_vector, digit_string): Delete.
8247 (ggc_alloc_string): Use plain "", don't optimize single digit
8248 strings. Use ggc_alloc_atomic.
8250 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
8252 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
8253 comparison set and one other set, use the cost of the non-comparison
8256 2017-06-30 Nathan Sidwell <nathan@acm.org>
8258 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
8261 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
8263 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
8264 loops. Remove now unneeded calls to gimple_switch_set_label() that
8265 just set removed labels to NULL_TREE.
8267 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
8269 * tree-ssanames.c (set_range_info_raw): Abstract from ...
8270 (set_range_info): ...here. Only call set_range_info_raw if domain
8272 (set_nonzero_bits): Call set_range_info_raw.
8273 * tree-ssanames.h (set_range_info_raw): New.
8275 2017-06-30 Jakub Jelinek <jakub@redhat.com>
8278 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
8279 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
8280 of nonimmediate_operand and <store_mask_constraint> instead of m
8281 for the input operand. For V8FI iterator, always split if input
8282 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
8283 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
8284 <store_mask_predicate> instead of register_operand and
8285 <store_mask_constraint> instead of v for the input operand. Make
8286 sure both operands aren't MEMs for if not <mask_applied>.
8288 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
8290 * lto-wrapper.c (copy_file) Close both file descriptors before
8293 2017-06-30 Martin Liska <mliska@suse.cz>
8296 * multiple_target.c (create_dispatcher_calls): Make ifunc
8297 also for function that don't have calls or are not referenced.
8299 2017-06-30 Richard Biener <rguenther@suse.de>
8301 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
8302 analyze the first scalar stmt. Move vector type computation
8303 for the BB case here from ...
8304 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
8305 live operation processing in the SLP case properly.
8307 2017-06-30 Richard Biener <rguenther@suse.de>
8309 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
8311 2017-06-30 Martin Liska <mliska@suse.cz>
8314 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
8315 before BUILT_IN_UNWIND_RESUME when ASAN is used.
8317 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
8319 * doc/invoke.texi (AArch64): Add missing options and remove redundant
8322 2017-06-30 Richard Biener <rguenther@suse.de>
8324 PR tree-optimization/81249
8325 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
8326 condition reduction result to original scalar type.
8328 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8330 * profile-count.h (enum profile_quality): Fix typos and whitespace
8333 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8335 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
8336 type for branch probabilities.
8338 2017-06-29 Julian Brown <julian@codesourcery.com>
8339 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8341 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
8342 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
8343 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
8344 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
8346 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8348 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
8349 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
8350 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
8351 CC usage from generic code to here.
8352 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
8353 CC usage into the target macros.
8355 2017-06-29 Maya Rashish <coypu@sdf.org>
8357 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
8360 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8362 * arm/arm-builtins.c: Include profile-count.h
8363 * except.c (sjlj_emit_function_enter): Use
8364 profile_probability::unlikely.
8366 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8368 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
8369 and tocrel_offset be pointer args rather than implicitly using
8371 (legitimate_constant_pool_address_p, rs6000_emit_move,
8372 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
8373 tocrel_offset and use in toc_relative_expr_p call.
8374 (print_operand, print_operand_address): Use static tocrel_base_oac
8375 and tocrel_offset_oac.
8376 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
8379 2017-06-29 Maya Rashish <coypu@sdf.org>
8381 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
8383 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
8385 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
8386 objects, take into account only the alignment of 'op0' and 'mode1' if
8389 2017-06-29 Steve Ellcey <sellcey@cavium.com>
8391 * ccmp.c (ccmp_tree_comparison_p): New function.
8392 (ccmp_candidate_p): Update to use above function.
8393 (get_compare_parts): New function.
8394 (expand_ccmp_next): Update to use new functions.
8395 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
8397 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
8398 take mode as argument.
8399 * ccmp.h (expand_ccmp_expr): Add mode as argument.
8400 * expr.c (expand_expr_real_1): Pass mode as argument.
8402 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
8404 * combine.c (combine_instructions): Print insns to dump_file, together
8407 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8409 * asan.c (asan_emit_stack_protection): Update.
8410 (create_cond_insert_point): Update.
8411 * auto-profile.c (afdo_propagate_circuit): Update.
8412 * basic-block.h (struct edge_def): Turn probability to
8413 profile_probability.
8414 (EDGE_FREQUENCY): Update.
8415 * bb-reorder.c (find_traces_1_round): Update.
8416 (better_edge_p): Update.
8417 (sanitize_hot_paths): Update.
8418 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
8419 (make_single_succ_edge): Update.
8420 (check_bb_profile): Update.
8421 (dump_edge_info): Update.
8422 (update_bb_profile_for_threading): Update.
8423 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
8424 probabilitycount to 0.
8425 * cfgbuild.c (compute_outgoing_frequencies): Update.
8426 * cfgcleanup.c (try_forward_edges): Update.
8427 (outgoing_edges_match): Update.
8428 (try_crossjump_to_edge): Update.
8429 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
8430 (expand_gimple_tailcall): Update.
8431 (construct_init_block): Use make_single_succ_edge.
8432 (construct_exit_block): Use make_single_succ_edge.
8433 * cfghooks.c (verify_flow_info): Update.
8434 (redirect_edge_succ_nodup): Update.
8435 (split_edge): Update.
8436 (account_profile_record): Update.
8437 * cfgloopanal.c (single_likely_exit): Update.
8438 * cfgloopmanip.c (scale_loop_profile): Update.
8439 (set_zero_probability): Remove.
8440 (duplicate_loop_to_header_edge): Update.
8441 * cfgloopmanip.h (loop_version): Update prototype.
8442 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
8443 (force_nonfallthru_and_redirect): Update.
8444 (update_br_prob_note): Update.
8445 (rtl_verify_edges): Update.
8446 (purge_dead_edges): Update.
8447 (rtl_lv_add_condition_to_bb): Update.
8448 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
8449 * cgraphunit.c (init_lowered_empty_function): Update.
8450 (cgraph_node::expand_thunk): Update.
8451 * cilk-common.c: Include profile-count.h
8452 * dojump.c (inv): Remove.
8453 (jumpifnot): Update.
8454 (jumpifnot_1): Update.
8455 (do_jump_1): Update.
8457 (do_jump_by_parts_greater_rtx): Update.
8458 (do_compare_rtx_and_jump): Update.
8459 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
8460 do_jump_1. do_compare_rtx_and_jump): Update prototype.
8461 * dwarf2cfi.c: Include profile-count.h
8462 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
8463 (sjlj_emit_dispatch_table): Likewise.
8464 * explow.c: Include profile-count.h
8465 * expmed.c (emit_store_flag_force): Update.
8466 (do_cmp_and_jump): Update.
8467 * expr.c (compare_by_pieces_d::generate): Update.
8468 (compare_by_pieces_d::finish_mode): Update.
8469 (emit_block_move_via_loop): Update.
8470 (store_expr_with_bounds): Update.
8471 (store_constructor): Update.
8472 (expand_expr_real_2): Update.
8473 (expand_expr_real_1): Update.
8474 * expr.h (try_casesi, try_tablejump): Update prototypes.
8475 * gimple-pretty-print.c (dump_probability): Update.
8476 (dump_profile): New.
8477 (dump_gimple_label): Update.
8478 (dump_gimple_bb_header): Update.
8479 * graph.c (draw_cfg_node_succ_edges): Update.
8480 * hsa-gen.c (convert_switch_statements): Update.
8481 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
8482 (find_if_case_1): Update.
8483 (find_if_case_2): Update.
8484 * internal-fn.c (expand_arith_overflow_result_store): Update.
8485 (expand_addsub_overflow): Update.
8486 (expand_neg_overflow): Update.
8487 (expand_mul_overflow): Update.
8488 (expand_vector_ubsan_overflow): Update.
8489 * ipa-cp.c (good_cloning_opportunity_p): Update.
8490 * ipa-split.c (split_function): Use make_single_succ_edge.
8491 * ipa-utils.c (ipa_merge_profiles): Update.
8492 * loop-doloop.c (add_test): Update.
8493 (doloop_modify): Update.
8494 * loop-unroll.c (compare_and_jump_seq): Update.
8495 (unroll_loop_runtime_iterations): Update.
8496 * lra-constraints.c (lra_inheritance): Update.
8497 * lto-streamer-in.c (input_cfg): Update.
8498 * lto-streamer-out.c (output_cfg): Update.
8499 * mcf.c (adjust_cfg_counts): Update.
8500 * modulo-sched.c (sms_schedule): Update.
8501 * omp-expand.c (expand_omp_for_init_counts): Update.
8502 (extract_omp_for_update_vars): Update.
8503 (expand_omp_ordered_sink): Update.
8504 (expand_omp_for_ordered_loops): Update.
8505 (expand_omp_for_generic): Update.
8506 (expand_omp_for_static_nochunk): Update.
8507 (expand_omp_for_static_chunk): Update.
8508 (expand_cilk_for): Update.
8509 (expand_omp_simd): Update.
8510 (expand_omp_taskloop_for_outer): Update.
8511 (expand_omp_taskloop_for_inner): Update.
8512 * omp-simd-clone.c (simd_clone_adjust): Update.
8513 * optabs.c (expand_doubleword_shift): Update.
8514 (expand_abs): Update.
8515 (emit_cmp_and_jump_insn_1): Update.
8516 (expand_compare_and_swap_loop): Update.
8517 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
8518 * predict.c (predictable_edge_p): Update.
8519 (edge_probability_reliable_p): Update.
8520 (set_even_probabilities): Update.
8521 (combine_predictions_for_insn): Update.
8522 (combine_predictions_for_bb): Update.
8523 (propagate_freq): Update.
8524 (estimate_bb_frequencies): Update.
8525 (force_edge_cold): Update.
8526 * profile-count.c (profile_count::dump): Add missing space into dump.
8527 (profile_count::debug): Add newline.
8528 (profile_count::differs_from_p): Explicitly convert to unsigned.
8529 (profile_count::stream_in): Update.
8530 (profile_probability::dump): New member function.
8531 (profile_probability::debug): New member function.
8532 (profile_probability::differs_from_p): New member function.
8533 (profile_probability::differs_lot_from_p): New member function.
8534 (profile_probability::stream_in): New member function.
8535 (profile_probability::stream_out): New member function.
8536 * profile-count.h (profile_count_quality): Rename to ...
8537 (profile_quality): ... this one.
8538 (profile_probability): New.
8539 (profile_count): Update.
8540 * profile.c (compute_branch_probabilities): Update.
8541 * recog.c (peep2_attempt): Update.
8542 * sched-ebb.c (schedule_ebbs): Update.
8543 * sched-rgn.c (find_single_block_region): Update.
8544 (compute_dom_prob_ps): Update.
8545 (schedule_region): Update.
8546 * sel-sched-ir.c (compute_succs_info): Update.
8547 * stmt.c (struct case_node): Update.
8548 (do_jump_if_equal): Update.
8549 (get_outgoing_edge_probs): Update.
8550 (conditional_probability): Update.
8551 (emit_case_dispatch_table): Update.
8552 (expand_case): Update.
8553 (expand_sjlj_dispatch_table): Update.
8554 (emit_case_nodes): Update.
8555 * targhooks.c: Update.
8556 * tracer.c (better_p): Update.
8557 (find_best_successor): Update.
8558 * trans-mem.c (expand_transaction): Update.
8559 * tree-call-cdce.c: Update.
8560 * tree-cfg.c (gimple_split_edge): Upate.
8561 (move_sese_region_to_fn): Upate.
8562 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
8563 * tree-eh.c (lower_resx): Upate.
8564 (cleanup_empty_eh_move_lp): Upate.
8565 * tree-if-conv.c (version_loop_for_if_conversion): Update.
8566 * tree-inline.c (copy_edges_for_bb): Update.
8567 (copy_cfg_body): Update.
8568 * tree-parloops.c (gen_parallel_loop): Update.
8569 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
8570 (gimple_gen_time_profiler): Update.
8571 * tree-ssa-dce.c (remove_dead_stmt): Update.
8572 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
8573 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
8574 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
8575 (unloop_loops): Update.
8576 (try_peel_loop): Update.
8577 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8578 * tree-ssa-loop-split.c (connect_loops): Update.
8579 (split_loop): Update.
8580 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
8581 (hoist_guard): Update.
8582 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
8583 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
8584 (value_replacement): Update.
8585 * tree-ssa-reassoc.c (branch_fixup): Update.
8586 * tree-ssa-tail-merge.c (replace_block_by): Update.
8587 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
8588 (create_edge_and_update_destination_phis): Update.
8589 (compute_path_counts): Update.
8590 (recompute_probabilities): Update.
8591 (update_joiner_offpath_counts): Update.
8592 (freqs_to_counts_path): Update.
8593 (duplicate_thread_path): Update.
8594 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
8595 (struct switch_conv_info): Update.
8596 (gen_inbound_check): Update.
8597 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
8598 (vect_do_peeling): Update.
8599 (vect_loop_versioning): Update.
8600 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8601 (optimize_mask_stores): Update.
8602 * ubsan.c (ubsan_expand_null_ifn): Update.
8603 * value-prof.c (gimple_divmod_fixed_value): Update.
8604 (gimple_divmod_fixed_value_transform): Update.
8605 (gimple_mod_pow2): Update.
8606 (gimple_mod_pow2_value_transform): Update.
8607 (gimple_mod_subtract): Update.
8608 (gimple_mod_subtract_transform): Update.
8609 (gimple_ic): Update.
8610 (gimple_stringop_fixed_value): Update.
8611 (gimple_stringops_transform): Update.
8612 * value-prof.h: Update.
8614 2017-06-29 Carl Love <cel@us.ibm.com>
8616 * config/rs6000/rs6000-c.c: Add support for built-in functions
8617 vector signed int vec_signed (vector float);
8618 vector signed long long vec_signed (vector double);
8619 vector signed int vec_signed2 (vector double, vector double);
8620 vector signed int vec_signede (vector double);
8621 vector signed int vec_signedo (vector double);
8622 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
8623 instruction generator.
8624 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
8625 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
8626 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
8628 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
8629 vunsignede_v2df): Add define_expands.
8630 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
8631 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
8632 VEC_UNSIGNEDO): Add definitions.
8633 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
8634 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
8635 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
8636 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
8637 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
8638 * config/rs6000/altivec.h (vec_signed, vec_signed2,
8639 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
8640 vec_unsignede, vec_unsignedo): Add builtin defines.
8641 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
8643 * doc/extend.texi: Update the built-in documentation file for the
8644 new built-in functions.
8646 2017-06-29 Richard Biener <rguenther@suse.de>
8648 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
8649 reduction chains to LOOP_VINFO_REDUCTIONS.
8650 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
8651 SLP reductions after processing reduction chains.
8653 2017-06-29 Nathan Sidwell <nathan@acm.org>
8655 * builtins.c (fold_builtin_FUNCTION): Use
8656 lang_hooks.decl_printable_name.
8658 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
8661 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
8662 with only one label.
8663 * stmt.c (expand_case): Assert NCASES is greater than one.
8665 2017-06-29 Richard Biener <rguenther@suse.de>
8667 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
8669 (group_case_labels): Likewise.
8670 (find_taken_edge): Push sanity checking on val to workers...
8671 (find_taken_edge_cond_expr): ... here
8672 (find_taken_edge_switch_expr): ... and here, handle cases
8673 with just a default label.
8674 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
8675 (group_case_labels): Likewise.
8676 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
8677 group_case_labels does anything cleanup the CFG again.
8679 2017-06-29 Bin Cheng <bin.cheng@arm.com>
8681 PR tree-optimization/81196
8682 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
8683 exit condition comparing two IVs.
8685 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
8687 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
8688 profile to the dummy entry at the end of the list of architectures.
8689 * config/arm/arm-cpu-cdata.h: Regenerated.
8691 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8692 Michael Collison <michael.collison@arm.com>
8695 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
8697 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
8698 (*aarch64_reg_<mode>3_minus_mask): New pattern.
8699 (*aarch64_<optab>_reg_di3_mask2): New pattern.
8700 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
8701 of shift when the shift amount is masked with constant equal to
8702 the size of the mode.
8703 * config/aarch64/predicates.md (subreg_lowpart_operator): New
8706 2017-06-29 Martin Liska <mliska@suse.cz>
8708 * config/i386/i386.opt: Change range from [1,5] to [0,5].
8710 2017-06-29 Yury Gribov <tetra2005@gmail.com>
8713 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
8716 (edge_growth_cache_entry::edge_growth_cache_entry): New
8718 (reset_edge_growth_cache): Update to use constructor.
8720 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8722 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
8723 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
8724 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
8726 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
8728 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
8729 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
8731 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
8733 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
8734 (*-linux-uclibc*): Add t-uclibc tmake_file.
8735 * config/t-musl: New.
8736 * config/t-uclibc: New.
8738 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
8740 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
8742 (gen_comm_data): Emit architectural setting of arch_prof.
8743 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
8745 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
8746 (armv8-m.base, armv8-m.main): Likewise.
8747 * arm-protos.h (arm_build_target): Add profile field.
8748 (arch_option): Likewise.
8749 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
8751 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
8752 arm_active_target.profile.
8754 2017-06-28 Richard Biener <rguenther@suse.de>
8757 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
8758 TYPE_OVERFLOW_WRAPS.
8759 * match.pd (negate_expr_p): Likewise.
8760 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
8761 fold_build2, not fold_binary.
8763 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
8765 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
8766 Convert memory address to Pmode.
8767 (aarch64_print_operand): Assert MEM operands are always Pmode.
8769 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
8772 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
8773 Remove redundant if.
8774 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
8775 * config/arm/aarch-common-protos.h
8776 (aarch_forward_to_shift_is_not_shifted_re): Remove.
8777 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
8779 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
8782 * multiple_target.c (create_dispatcher_calls): Set the default
8783 clone to be static, not public.
8785 2017-06-28 Richard Biener <rguenther@suse.de>
8787 * tree-vect-loop.c (vectorizable_reduction): Move special
8788 cond reduction IV var creation ...
8789 (vect_create_epilog_for_reduction): ... here. Remove induction_index
8790 parameter. Use STMT_VINFO_VECTYPE.
8791 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
8794 2017-06-28 Martin Liska <mliska@suse.cz>
8797 * ipa-visibility.c (non_local_p): Handle visibility.
8799 2017-06-28 Martin Liska <mliska@suse.cz>
8802 * common.opt: Add IntegerRange to various options.
8803 * opt-functions.awk (integer_range_info): New function.
8804 * optc-gen.awk: Add integer_range_info to cl_options struct.
8805 * opts-common.c (decode_cmdline_option): Handle
8806 CL_ERR_INT_RANGE_ARG.
8807 (cmdline_handle_error): Likewise.
8808 * opts.c (print_filtered_help): Show valid interval in
8809 when --help is provided.
8810 * opts.h (struct cl_option): Add range_min and range_max fields.
8811 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
8813 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
8815 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
8816 (x * C EQ/NE y * C): New transformation.
8818 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
8820 * genmultilib (combination_space): Accept '+' in option names.
8822 2017-06-28 Martin Liska <mliska@suse.cz>
8825 * asan.c (instrument_derefs): Bail out inner references
8826 that are hard register variables.
8828 2017-06-28 Jakub Jelinek <jakub@redhat.com>
8831 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
8832 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
8834 2017-06-28 Richard Biener <rguenther@suse.de>
8836 * tree-vectorizer.h (vect_get_vec_defs): Remove.
8837 (vect_get_slp_defs): Adjust.
8838 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
8840 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
8842 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
8843 get_initial_defs_for_reduction instead of vect_get_vec_defs.
8844 (vectorizable_reduction): Adjust.
8845 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
8847 (vect_get_slp_defs): Likewise.
8848 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
8849 (vectorizable_bswap): Adjust.
8850 (vectorizable_call): Likewise.
8851 (vectorizable_conversion): Likewise.
8852 (vectorizable_assignment): Likewise.
8853 (vectorizable_shift): Likewise.
8854 (vectorizable_operation): Likewise.
8855 (vectorizable_store): Likewise.
8856 (vectorizable_condition): Likewise.
8857 (vectorizable_comparison): Likewise.
8859 2017-06-28 Michael Collison <michael.collison@arm.com>
8862 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
8864 (arm_gen_movmemqi): Removed unused variable 'i'.
8865 Convert 'for' loop into 'while' loop.
8866 (arm_expand_prologue): Remove last unnecessary set of insn.
8867 (thumb_pop): Remove unused variable 'pushed_words'.
8868 (thumb_exit): Remove last unnecessary set of regs_to_pop.
8870 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8872 * config/s390/predicates.md: Use s390_rel_address_ok_p.
8873 * config/s390/s390-protos.h: Add prototype of
8874 s390_rel_address_ok_p.
8875 * config/s390/s390.c (s390_got_symbol): New function.
8876 (s390_rel_address_ok_p): New function.
8877 (legitimize_pic_address): Use s390_rel_address_ok_p.
8878 (s390_load_got): Use s390_got_symbol.
8879 (s390_option_override): Issue error if
8880 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
8881 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
8883 * config/s390/s390.opt: New option mpic-data-is-text-relative.
8885 2017-06-27 Andrew Pinski <apinski@cavium.com>
8887 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
8888 (X * copysign (1.0, X)): New pattern.
8889 (X * copysign (1.0, -X)): New pattern.
8890 (copysign (-1.0, CST)): New pattern.
8892 2017-06-27 Joseph Myers <joseph@codesourcery.com>
8894 * genmultilib (combination_space): Remove variable.
8895 Validate reuse rules against regular expression for any sequence
8896 of multilib options in any order.
8898 2017-06-27 Michael Collison <michael.collison@arm.com>
8900 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
8901 call aarch64_split_simd_combine.
8902 * (aarch64_combine_internal<mode>): Delete pattern.
8903 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
8904 Allow register and subreg operands.
8906 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8908 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
8909 specific need, just fallback on defaults.
8910 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
8912 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8913 Olivier Hainque <hainque@adacore.com>
8915 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
8917 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
8918 targets. Pick a default if no particular attempt applied.
8919 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
8922 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8924 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
8925 (x86_64-wrs-vxworks7): Likewise.
8927 2017-06-27 Marek Polacek <polacek@redhat.com>
8930 * ubsan.c (instrument_null): Check get_base_address's result for null.
8932 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
8934 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
8936 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
8938 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
8939 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
8940 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
8941 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
8943 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
8944 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
8945 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
8946 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
8947 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
8948 BUILT_IN_FEUPDATEENV): New builtins.
8949 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
8950 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
8951 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
8952 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
8954 (builtin_structptr_types): Adjust size.
8955 * tree.c (builtin_structptr_types): Add four entries.
8957 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8958 Olivier Hainque <hainque@adacore.com>
8960 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
8961 (TLS_SYM): New local macro, forcing reference to __tls__ on
8962 link command lines for VxWorks 7 RTPs, triggering initialization
8964 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
8965 OS features TLS support, true for RTPs on VxWorks 7.
8966 * config/vxworks.c (vxworks_override_options): Setup emutls
8969 2017-06-27 Jakub Jelinek <jakub@redhat.com>
8971 * predict.c (test_prediction_value_range): Use -1U instead of -1
8972 to avoid narrowing conversion warning.
8973 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
8974 to avoid narrowing conversion warning.
8975 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
8977 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
8979 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8981 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
8982 64bit configurations.
8983 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
8984 (SIZE_TYPE): Likewise.
8985 * config/vxworks.c (vxworks_emutls_var_fields): Use
8986 long_unsigned_type_node instead of unsigned_type_node as the offset
8987 field type, which is "pointer" mode in emutls.c.
8989 2017-06-27 Jakub Jelinek <jakub@redhat.com>
8992 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
8995 * gimple-fold.c (replace_call_with_call_and_fold): Handle
8996 gimple_vuse copying separately from gimple_vdef copying.
8998 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9000 * value-prof.c (free_hist): Remove call to memset and the enclosing if
9003 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
9004 Olivier Hainque <hainque@adacore.com>
9006 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
9007 for all vxworks7 targets.
9008 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
9009 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
9010 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
9011 variations for VX6/VX7 and 32/64bits later on in ...
9012 (VXWORKS_LIB_SPEC): Leverage new macros.
9013 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
9014 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
9016 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
9018 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
9019 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
9021 2017-06-26 Carl Love <cel@us.ibm.com>
9023 * config/rs6000/rs6000-c.c: Add support for built-in functions
9024 vector bool char vec_reve (vector bool char);
9025 vector signed char vec_reve (vector signed char);
9026 vector unsigned char vec_reve (vector unsigned char);
9027 vector bool int vec_reve (vector bool int);
9028 vector signed int vec_reve (vector signed int);
9029 vector unsigned int vec_reve (vector unsigned int);
9030 vector bool long long vec_reve (vector bool long long);
9031 vector signed long long vec_reve (vector signed long long);
9032 vector unsigned long long vec_reve (vector unsigned long long);
9033 vector bool short vec_reve (vector bool short);
9034 vector signed short vec_reve (vector signed short);
9035 vector double vec_reve (vector double);
9036 vector float vec_reve (vector float);
9037 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
9038 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
9039 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
9040 (altivec_vreve): New pattern.
9041 * config/rs6000/altivec.h (vec_reve): New define.
9042 * doc/extend.texi (vec_rev): Update the built-in documentation file
9043 for the new built-in functions.
9045 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9047 PR tree-optimization/71815
9048 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
9050 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
9052 (slsr_process_phi): Likewise.
9053 (replace_uncond_cands_and_profitable_phis): Don't replace a
9054 multiply candidate with a stride of 1 (copy or cast).
9055 (phi_incr_cost): Call uses_consumed_by_stmt rather than
9057 (lowest_cost_path): Likewise.
9058 (total_savings): Likewise.
9060 2017-06-26 Richard Biener <rguenther@suse.de>
9063 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
9064 Use def_builtin_pure for all gather builtins.
9066 2017-06-26 Richard Biener <rguenther@suse.de>
9068 PR tree-optimization/81203
9069 * tree-tailcall.c (find_tail_calls): Do not move stmts into
9072 2017-06-26 Marek Polacek <polacek@redhat.com>
9075 * doc/invoke.texi: Document -Wmultistatement-macros.
9077 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
9079 * doc/sourcebuild.texi (ARM-specific attributes): Document new
9080 arm_neon_ok_no_float_abi effective target.
9082 2017-06-26 Richard Biener <rguenther@suse.de>
9084 PR tree-optimization/80928
9085 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
9086 (copy_bbs): Set BB_DUPLICATED flag early.
9087 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
9089 (execute_on_shrinking_pred): Likewise.
9090 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
9091 BB_DUPLICATED blocks.
9092 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
9093 iterate over all PHIs considering removal of *gsi.
9095 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
9097 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
9100 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9102 * config/rs6000/rs6000-string.c: (expand_block_clear,
9103 do_load_for_compare, select_block_compare_mode,
9104 compute_current_alignment, expand_block_compare,
9105 expand_strncmp_align_check, expand_strn_compare,
9106 expand_block_move, rs6000_output_load_multiple)
9107 Move functions related to string/block move/compare
9109 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
9110 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
9111 for this function which is now used in two files.
9112 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
9113 * config.gcc: Add rs6000-string.o to extra_objs for
9114 targets powerpc*-*-* and rs6000*-*-*.
9116 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
9119 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
9120 32-bit, since indexed is not valid for DImode.
9121 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
9122 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
9123 (define_peephole2 for Altivec d-form load): Add 32-bit support.
9124 (define_peephole2 for Altivec d-form store): Likewise.
9127 * multiple_target.c (create_dispatcher_calls): Only create the
9128 dispatcher call if the function is the default clone of a
9131 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
9134 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
9135 a call, force the call to not be a tail call.
9137 2017-06-23 Jeff Law <law@redhat.com>
9139 * doc/contrib.texi: Add entry for Steven Pemberton's work on
9142 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
9144 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
9145 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
9146 handling for early expansion of vector shifts (sl,sr,sra,rl).
9147 (builtin_function_type): Add vector shift right instructions
9148 to the unsigned argument list.
9150 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
9152 rtl-optimizatoin/79286
9153 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
9154 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
9155 trap. PIC register plus a const unspec without offset can never trap.
9157 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
9159 * tree.h (builtin_structptr_type): New type.
9160 (builtin_structptr_types): Declare new array.
9161 * tree.c (builtin_structptr_types): New array.
9162 (free_lang_data, build_common_tree_nodes): Use it.
9164 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
9167 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
9170 2017-06-22 Matt Turner <mattst88@gmail.com>
9172 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
9173 Lake models to skylake case. Assume skylake for unknown
9174 models with clflushopt.
9176 2017-06-22 Jeff Law <law@redhat.com>
9178 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
9179 frame sizes that do not satisfy aarch64_uimm12_shift.
9181 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
9183 * profile-count.h (apply_probability,
9184 apply_scale, probability_in): Fix checks for zero.
9186 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9188 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
9189 * doc/cppdiropts.texi (-I @var{dir}): Document it.
9191 2016-06-22 Richard Biener <rguenther@suse.de>
9193 * tree-vect-loop.c (vect_model_reduction_cost): Handle
9194 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
9195 REDUC_MAX_EXPR support.
9196 (vectorizable_reduction): Likewise.
9197 (vect_create_epilog_for_reduction): Likewise.
9199 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
9201 * match.pd (A / (1 << B) -> A >> B): New.
9202 * generic-match-head.c: Include optabs-tree.h.
9203 * gimple-match-head.c: Likewise.
9204 * optabs-tree.h (target_supports_op_p): New.
9205 * optabs-tree.c (target_supports_op_p): New.
9207 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9209 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
9210 $gcc_cv_ld --help output.
9211 (gcc_cv_ld_demangle): Likewise.
9212 (gcc_cv_ld_eh_frame_hdr): Likewise.
9213 (gcc_cv_ld_pie): Likewise.
9214 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
9215 (gcc_cv_ld_buildid): Likewise.
9216 (gcc_cv_ld_sysroot): Likewise.
9217 (ld_bndplt_support): Likewise.
9218 (ld_pushpopstate_support): Likewise.
9219 * configure: Regenerate.
9220 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
9222 2017-06-21 Jakub Jelinek <jakub@redhat.com>
9225 * config/i386/sse.md (round<mode>2): Renumber match_dup and
9226 operands indexes to avoid gap between operands and match_dups.
9228 2017-06-21 Andrew Pinski <apinski@cavium.com>
9230 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
9231 Increment Arith_shift and Arith_shift_reg by 1.
9232 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
9234 * config/aarch64/aarch64.c (thunderx_tunings): Enable
9235 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
9236 (aarch64_strip_extend): Add new argument and test for it.
9237 (aarch64_cheap_mult_shift_p): New function.
9238 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
9239 add a cost if it is true.
9240 Update calls to aarch64_strip_extend.
9241 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
9243 2017-06-21 Andrew Pinski <apinski@cavium.com>
9245 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
9247 (thunderxt88): Likewise.
9248 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
9249 (thunderx_prefetch_tune): New variable.
9250 (thunderx2t99_prefetch_tune): Update for the correct values.
9251 (thunderxt88_tunings): New variable.
9252 (thunderx_tunings): Use thunderx_prefetch_tune instead of
9253 generic_prefetch_tune.
9254 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
9256 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9258 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
9259 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
9260 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
9261 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
9262 (aarch64_atomic_cas<mode>, GPI): Likewise.
9264 2017-06-21 Martin Liska <mliska@suse.cz>
9266 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
9267 statements on cold and hot labels.
9268 * predict.c (tree_estimate_probability_bb): Remove the
9269 prediction from this place.
9271 2017-06-21 Martin Liska <mliska@suse.cz>
9273 PR tree-optimization/79489
9274 * gimplify.c (maybe_add_early_return_predict_stmt): New
9276 (gimplify_return_expr): Call the function.
9277 * predict.c (tree_estimate_probability_bb): Remove handling
9279 * predict.def: Update comment about early return predictor.
9280 * gimple-predict.h (is_gimple_predict): New function.
9281 * predict.def: Change default value of early return to 66.
9282 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
9284 * passes.def: Put pass_strip_predict_hints to the beginning of
9287 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
9289 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
9290 FUNCTION_DECL declarations.
9291 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
9293 (dwaf2out_decl): Likewise.
9294 * godump.c (go_early_global_decl): Skip call to the real debug hook
9295 for FUNCTION_DECL declarations.
9296 * passes.c (rest_of_decl_compilation): Skip call to the
9297 early_global_decl debug hook for FUNCTION_DECL declarations, unless
9298 -fdump-go-spec is passed.
9300 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
9302 * config/i386/i386.c (struct builtin_isa): New field pure_p.
9303 Reorder for compactness.
9304 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
9305 (def_builtin_pure, def_builtin_pure2): New functions.
9306 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
9308 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
9310 * match.pd (nop_convert): New predicate.
9311 ((A +- CST1) +- CST2): Allow some NOP conversions.
9313 2017-06-21 Jakub Jelinek <jakub@redhat.com>
9316 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
9317 with ctors/dtors if GOVD_SHARED is set.
9319 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
9321 * config/aarch64/aarch64.md (movti_aarch64):
9322 Emit mov rather than orr.
9323 (movtf_aarch64): Likewise.
9324 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
9325 Emit mov rather than orr.
9327 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
9329 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
9330 Swap alternatives, make integer dup more expensive.
9332 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
9334 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
9335 Return true for non-tls symbols.
9337 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
9339 * config/aarch64/aarch64-cores.def (cortex-a55): New.
9340 (cortex-a75): Likewise.
9341 (cortex-a75.cortex-a55): Likewise.
9342 * config/aarch64/aarch64-tune.md: Regenerate.
9343 * doc/invoke.texi (-mtune): Document new values for -mtune.
9345 2017-06-21 Tom de Vries <tom@codesourcery.com>
9347 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
9349 (Effective-Target Keywords, Other attributes): Suggest using
9350 dg-add-options stack_size feature to get stack limit in stack_size
9351 effective target documentation.
9353 2017-06-21 Julian Brown <julian@codesourcery.com>
9354 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9356 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
9357 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
9358 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
9360 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
9361 attribute type list for neon_multiply.
9362 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
9363 attribute type list for neon_multiply.
9364 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
9365 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
9366 attribute type list for neon_multiply.
9367 * config/arm/types.md (crypto_pmull): Add.
9368 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
9369 attribute type list.
9371 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
9373 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
9376 2017-06-20 Jakub Jelinek <jakub@redhat.com>
9378 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
9379 to make sure not to dereference a NULL cost_classes_ptr pointer.
9381 2017-06-20 Carl Love <cel@us.ibm.com>
9383 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9384 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9385 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9386 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9387 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9388 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
9389 VMULOSW): New enum "unspec" values.
9390 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
9391 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
9392 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
9393 altivec_vmulosw): New patterns.
9394 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9395 VMULOSW): Add definitions.
9397 2017-06-20 Julia Koval <julia.koval@intel.com>
9399 * config/i386/i386.c: Fix rounding expand for new pattern.
9400 * config/i386/subst.md: Fix pattern (parallel -> unspec).
9402 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
9404 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
9405 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
9407 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
9409 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
9412 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
9414 * config/aarch64/aarch64-cores.def: Rearrange to sort by
9415 architecture, then by implementer ID.
9416 * config/aarch64/aarch64-tune.md: Regenerate.
9418 2017-06-20 Richard Biener <rguenther@suse.de>
9421 * fold-const.c (split_tree): Fold to type before negating.
9423 2017-06-20 David Malcolm <dmalcolm@redhat.com>
9425 * diagnostic-show-locus.c
9426 (selftest::test_fixit_deletion_affecting_newline): New function.
9427 (selftest::diagnostic_show_locus_c_tests): Call it.
9429 2017-06-20 Andreas Schwab <schwab@suse.de>
9432 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
9435 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9437 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
9438 __ARM_FEATURE_COPROC according to support.
9440 2017-06-20 Jakub Jelinek <jakub@redhat.com>
9442 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
9443 Rewritten to avoid overflow for > 32-bit pointers.
9446 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
9447 by removing enum keyword.
9448 (ubsan_type_descriptor): Likewise. Formatting fix.
9451 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
9452 splitter): Require TARGET_SSE2 in the condition.
9454 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
9457 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
9458 for doing vector set of SFmode on ISA 3.0.
9459 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
9460 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
9462 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
9463 SFmode value into a V4SF variable that was extracted from another
9464 V4SF variable without converting the element to double precision
9465 and back to single precision vector format.
9466 (vsx_insert_extract_v4sf_p9_2): Likewise.
9468 2017-06-19 Jakub Jelinek <jakub@redhat.com>
9470 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
9471 in UWHI to avoid undefined overflow.
9474 * ubsan.h (enum ubsan_encode_value_phase): New.
9475 (ubsan_encode_value): Change second argument to
9476 enum ubsan_encode_value_phase with default value of
9477 UBSAN_ENCODE_VALUE_GENERIC.
9478 * ubsan.c (ubsan_encode_value): Change second argument to
9479 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
9480 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
9481 create_tmp_var_raw instead of create_tmp_var and use a
9483 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
9484 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
9485 ubsan_encode_value callers.
9488 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
9489 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
9490 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
9492 2017-06-19 Richard Biener <rguenther@suse.de>
9495 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
9496 estimates if we changed anything.
9498 2017-06-19 Richard Biener <rguenther@suse.de>
9500 PR tree-optimization/80887
9501 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
9502 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
9503 simplified lookups, then reset mprts_hook.
9504 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
9506 (try_to_simplify): Likewise.
9508 2017-06-19 Martin Liska <mliska@suse.cz>
9511 * gimplify.c (gimplify_switch_expr):
9512 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
9514 2017-06-19 Martin Liska <mliska@suse.cz>
9516 * doc/install.texi: Document that PGO runs in 4 stages.
9518 2017-06-19 Martin Liska <mliska@suse.cz>
9521 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
9522 to dispatcher function name.
9523 * multiple_target.c (replace_function_decl): New function.
9524 (create_dispatcher_calls): Redirect both edges and references.
9526 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
9528 * profile-count.c (profile_count::dump): Dump quality.
9529 (profile_count::differs_from_p): Update for unsigned val.
9530 * profile-count.h (profile_count_quality): New enum.
9531 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
9533 2017-06-19 Richard Biener <rguenther@suse.de>
9535 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
9536 struct function as arg.
9537 (estimate_numbers_of_iterations): Export overload with loop arg.
9538 (free_numbers_of_iterations_estimates_loop): Use an overload of
9539 free_numbers_of_iterations_estimates instead.
9540 * tree-cfg.c (remove_bb): Adjust.
9541 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
9542 * tree-parloops.c (gen_parallel_loop): Likewise.
9543 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
9545 (tree_unroll_loops_completely): Likewise.
9546 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
9547 Use an overload instead and export.
9548 (estimated_loop_iterations): Adjust.
9549 (max_loop_iterations): Likewise.
9550 (likely_max_loop_iterations): Likewise.
9551 (estimate_numbers_of_iterations): Take struct function as arg
9553 (loop_exits_before_overflow): Adjust.
9554 (free_numbers_of_iterations_estimates_loop): Use an overload.
9555 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
9556 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
9558 2017-06-19 Richard Biener <rguenther@suse.de>
9561 * ipa-prop.c (find_constructor_constant_at_offset): Handle
9562 RANGE_EXPR conservatively.
9564 2017-06-16 Carl Love <cel@us.ibm.com>
9566 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9567 definitions for vec_float, vec_float2, vec_floato,
9568 vec_floate built-ins.
9569 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
9570 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
9572 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
9573 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
9574 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
9575 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
9576 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
9577 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
9578 vec_floato): Add builtin defines.
9579 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
9580 Update the built-in documentation file for the new built-in
9583 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9585 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
9586 (mthumb): Mark as the negative of -marm.
9588 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9590 * doc/invoke.texi (ARM Options, -mcpu): Document supported
9592 (ARM Options, -mtune): Document that this accepts the same
9593 extension options as -mcpu.
9594 (ARM Options, -mfpu): Document addition of -mfpu=auto.
9596 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9598 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
9599 permitted extensions.
9601 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9603 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
9604 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
9605 (armv8-m.main): Add option +nodsp.
9606 * config/arm/arm-cpu-cdata.h: Regenerated.
9608 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9610 * config/arm/t-fuchsia: New file.
9611 * config.gcc (arm*-*-fuchsia*): Use it.
9613 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9615 * config/arm/t-symbian: Rewrite for new option infrastructure.
9617 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9619 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
9620 (MULTILIB_REQUIRED): Likewise.
9622 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9624 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
9625 (MULTILIB_RESUE): Likewise.
9626 (MULTILIB_MATCHES): Likewise.
9627 (MULTLIB_REQUIRED): Likewise.
9629 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9631 * config/arm/t-rtems: Rewrite for new option framework.
9633 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9635 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
9636 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
9637 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
9638 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
9639 * config/arm/t-multilib: ... here.
9640 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
9641 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
9642 armv7-a and armv8*-a when A-profile libraries have not been built.
9643 * config/arm/t-rmprofile: Rewrite.
9645 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9647 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
9648 with a backslash. Remove the backslash after substituting unescaped
9650 * doc/fragments.texi (MULTILIB_REUSE): Document it.
9652 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9654 * config.gcc: (arm*-*-*): When building a-profile libraries, force
9655 the driver to pass through the default setting of -mfloat-abi.
9656 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
9658 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
9659 (all_feat_combs): New rule.
9660 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
9662 * config/arm/t-aprofile: Rewrite.
9664 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9666 * config/arm/arm.h (FPUTYPE_AUTO): Define.
9667 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
9668 fpu is not specified by the user/command-line.
9669 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
9670 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
9671 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
9672 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
9673 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
9674 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
9676 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9678 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
9679 * config/arm/t-arm-elf: Rewritten.
9681 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9683 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
9684 have some floating-point instructions.
9685 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
9686 (TARGET_MAYBE_HARD_FLOAT): New macro.
9687 * config/arm/arm-builtins.c (arm_init_builtins): Use
9688 TARGET_MAYBE_HARD_FLOAT.
9689 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
9691 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9693 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
9694 (configargs.h): Include it.
9695 (arm_print_hint_for_fpu_option): New function.
9696 (arm_parse_fpu_option): New function.
9697 (candidate_extension): New class.
9698 (arm_canon_for_multilib): New function.
9699 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
9700 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
9701 (ARCH_CANONICAL_SPECS): New macro.
9702 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
9704 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9706 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
9707 are set after handling multilib fragments. Set target_cpu_default2
9710 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9712 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
9714 (arm*-*-*): Set target_cpu_default2 to a quoted string.
9715 * config/arm/parsecpu.awk (check_cpu): Validate any extension
9717 (check_arch): Likewise.
9718 * config/arm/arm.c (arm_configure_build_target): Handle
9719 TARGET_CPU_DEFAULT being a string constant. Scan any feature
9720 options in the default.
9722 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9724 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
9725 when an option is an alias of another.
9726 * config/arm/parsecpu.awk (optalias): New parser token.
9727 (gen_comm_data): Mark non-alias options as such. Emit entries
9728 for extension aliases.
9729 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
9730 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
9731 (armv6kz, armv6zk, armv6t2): Likewise.
9732 (armv7): Make vfpv3-d16 an alias.
9733 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
9735 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
9736 Sort in canonical order.
9737 (armv8-a): Sort in canonical order.
9738 (armv8.1-a, armv8.2-a): Likewise.
9739 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
9741 (cortex-a9): Sort in canonical order.
9742 * config/arm/arm.c (selftests.h): Include it.
9743 (arm_test_cpu_arch_data): New function.
9744 (arm_run_self_tests): New function.
9745 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
9746 (targetm): Move declaration to the end of the file.
9747 * arm-cpu-cdata.h: Regenerated.
9749 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9751 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
9752 call to target_mode_check describing the type of option passed.
9753 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
9754 (arm_target_thumb_only): Use arm_parse_arch_option_name or
9755 arm_parse_cpu_option_name to match parameters against list of
9757 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
9758 arm_arch_core_flags data structure.
9759 * config/arm/arm-cpu_cdata.h: Regenerated.
9761 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9763 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
9765 (arm_print_hint_for_cpu_option): Likewise.
9766 (arm_print_hint_for_arch_option): Likewise.
9767 (arm_parse_cpu_option_name): Likewise.
9768 (arm_parse_arch_option_name): Likewise.
9769 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
9770 of entries in the all_fpus list.
9771 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
9772 (arm_parse_cpu_option_name): Declare.
9773 (arm_parse_arch_option_name): Declare.
9774 (arm_parse_option_features): Declare.
9775 (arm_intialize_isa): Declare.
9776 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
9778 (gen_comm_data): ... here. Make definitions non-static.
9779 * config/arm/arm-cpu-data.h: Regenerated.
9780 * config/arm/arm-cpu-cdata.h: Regenerated.
9782 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9784 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
9785 (cpu_arch_extension): New structure.
9786 (cpu_arch_option, arch_option, cpu_option): New structures.
9787 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
9789 (gen_data): Generate new format data tables.
9790 * config/arm/arm.c (cpu_tune): New structure.
9791 (cpu_option, processors): Delete.
9792 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
9793 (arm_print_hint_for_cpu_option): ... this and ...
9794 (arm_print_hint_for_arch_option): ... this.
9795 (arm_parse_arch_cpu_name): Delete. Replace with ...
9796 (arm_parse_cpu_option_name): ... this and ...
9797 (arm_parse_arch_option_name): ... this.
9798 (arm_unrecognized_feature): Change type of target parameter to
9800 (arm_parse_arch_cpu_features): Delete. Replace with ...
9801 (arm_parse_option_features): ... this.
9802 (arm_configure_build_target): Rework to use new configuration data
9804 (arm_print_tune_info): Rework for new configuration data tables.
9805 * config/arm/arm-cpu-data.h: Regenerated.
9806 * config/arm/arm-cpu.h: Regenerated.
9808 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9810 * Makefile.in (OBJS): Move sbitmap.o from here ...
9811 (OBJS-libcommon): ... to here.
9813 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9815 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
9816 (ISA_ALL_CRYPTO): New macro.
9817 (ISA_ALL_SIMD): New macro
9818 (ISA_ALL_FP): New macro.
9819 * config/arm/arm.c (fpu_bitlist): Update initializer.
9820 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
9822 (arm9e): Add fpu. Add option for nofp
9823 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
9824 (arm926ej-s, arm1026ej-s): Likewise.
9825 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
9826 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
9827 neon-fp16, neon-vfpv4, nofp and nosimd.
9828 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
9829 (cortex-a8): Add fpu. Add option for nofp.
9830 (cortex-a9): Add fpu. Add options for nosimd and nofp.
9831 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
9832 (cortex-r4f): Add fpu.
9833 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
9834 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
9836 (cortex-r8): Likewise.
9837 (cortex-m4): Add fpu. Add option for nofp.
9838 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
9839 (cortex-a17.cortex-a7): Likewise.
9840 (cortex-a32): Add fpu. Add options for crypto and nofp.
9841 (cortex-a35, cortex-a53): Likewise.
9842 (cortex-a57): Add fpu. Add option for crypto.
9843 (cortex-a72, cortex-a73): Likewise.
9844 (exynos-m1): Likewise.
9845 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
9846 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
9847 (cortex-m33): Add fpu. Add option for nofp.
9848 * config/arm/arm-cpu-cdata.h: Regenerated
9849 * config/arm/arm-cpu-data.h: Regenerated.
9851 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9853 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
9854 (armv5te, armv5tej): Likewise.
9855 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
9856 (armv7): Add options fp and vfpv3-d16.
9857 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
9858 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
9860 (armv7ve): Likewise.
9861 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
9862 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
9863 (armv8-a): Add nocrypto option.
9864 (armv8.1-a, armv8.2-a): Likewise.
9865 (armv8-m.main): add options fp, fp.dp and nofp.
9867 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9869 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
9871 (armv8-a+crc): Delete.
9872 (armv8.1-a): Add options simd, crypto and nofp.
9873 (armv8.2-a): Add options fp16, simd, crypto and nofp.
9874 (armv8.2-a+fp16): Delete.
9875 (armv8-m.main): Add option dsp.
9876 (armv8-m.main+dsp): Delete.
9877 (cortex-a8): Add fpu. Add nofp option.
9878 (cortex-a9): Add fpu. Add nofp and nosimd options.
9879 * config/arm/parsecpu.awk (gen_data): Generate option tables and
9880 link to main cpu and architecture data structures.
9881 (gen_comm_data): Only put isa attributes from the main architecture
9883 (option): New statement for architecture and CPU entries.
9884 * arm.c (struct cpu_option): New structure.
9885 (struct processors): Add entry for options.
9886 (arm_unrecognized_feature): New function.
9887 (arm_parse_arch_cpu_name): Ignore any characters after the first
9889 (arm_parse_arch_cpu_feature): New function.
9890 (arm_configure_build_target): Separate out any CPU and architecture
9891 features and parse separately. Don't error out if -mfpu=auto is
9892 used with only an architecture string.
9893 (arm_print_asm_arch_directives): New function.
9894 (arm_file_start): Call it.
9895 * config/arm/arm-cpu-cdata.h: Regenerated.
9896 * config/arm/arm-cpu-data.h: Likewise.
9897 * config/arm/arm-tables.opt: Likewise.
9899 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9901 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
9902 assembler when it is not -mfpu=auto.
9904 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9906 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
9907 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
9908 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
9909 (ASM_CPU_SPEC): Rewrite.
9910 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
9911 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
9912 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
9913 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
9914 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
9915 copied string is NUL-terminated. Also strip any characters prefixed
9917 (arm_rewrite_selected_arch): New function.
9918 (arm_rewrite_march): New function.
9920 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9922 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
9923 (x_arm_cpu_string, x_arm_tune_string): Likewise.
9924 (march, mcpu, mtune): Convert to string-based options.
9925 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
9926 (arm_parse_arch_cpu_name): New function.
9927 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
9928 identify selected architecture or CPU.
9929 (arm_option_save): New function.
9930 (TARGET_OPTION_SAVE): Redefine.
9931 (arm_option_restore): Restore string options.
9932 (arm_option_print): Print string options.
9934 2017-06-16 Martin Sebor <msebor@redhat.com>
9936 PR tree-optimization/80933
9937 PR tree-optimization/80934
9938 * builtins.c (fold_builtin_3): Do not handle bcmp here.
9939 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
9940 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
9941 (gimple_fold_builtin): Call them.
9943 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9945 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
9946 as unlikely; update profile.
9948 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9950 * predict.c (force_edge_cold): Handle declaring edges impossible
9953 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9955 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
9957 (try_unroll_loop_completely): Fix reporting.
9959 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9961 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
9963 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
9966 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
9967 if given a non-constant argument for an intrinsic which requires a
9970 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9972 * profile.c (compare_freqs): New function.
9973 (branch_prob): Sort edge list.
9974 (find_spanning_tree): Assume that the list is priority sorted.
9976 2017-06-16 Richard Biener <rguenther@suse.de>
9978 PR tree-optimization/81090
9979 * passes.def (pass_record_bounds): Remove.
9980 * tree-pass.h (make_pass_record_bounds): Likewise.
9981 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
9982 make_pass_record_bounds): Likewise.
9983 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
9984 not free niter estimates at the beginning but at the end.
9985 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
9987 2017-06-16 Richard Biener <rguenther@suse.de>
9989 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
9990 initializer to workaround ICE in host GCC 4.8.
9992 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9994 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
9996 (clone_inlined_nodes): Update.
9998 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10000 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
10001 prefetch settings, and enable prefetching by default at -O3.
10003 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10005 * config/aarch64/aarch64.c (aarch64_override_options_internal):
10006 Set flag_prefetch_loop_arrays according to tuning data.
10008 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10010 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
10011 New tune structure.
10012 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
10013 [Unrelated to main purpose of the patch] Place the pointer field last
10014 to enable type checking errors when tune structure are wrongly merged.
10015 * config/aarch64/aarch64.c (generic_prefetch_tune,)
10016 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
10017 (thunderx2t99_prefetch_tune): New tune constants.
10018 (tune_params *_tunings): Update all tunings (no functional change).
10019 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
10020 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
10021 from tunings structures.
10023 2017-06-16 Jakub Jelinek <jakub@redhat.com>
10026 * ubsan.c (instrument_null): Add T argument, use it instead
10027 of computing it based on IS_LHS.
10028 (instrument_object_size): Likewise.
10029 (pass_ubsan::execute): Adjust instrument_null and
10030 instrument_object_size callers to pass gimple_get_lhs or
10031 gimple_assign_rhs1 result to it. Use instrument_null instead of
10032 calling get_base_address and instrument_mem_ref. Handle
10033 aggregate call arguments for object-size sanitization.
10035 2017-06-16 Yury Gribov <tetra2005@gmail.com>
10037 PR tree-optimization/81089
10038 * tree-vrp.c (is_masked_range_test): Validate operands of
10041 2017-06-15 Martin Sebor <msebor@redhat.com>
10044 * dumpfile.c (dump_register): Avoid calling memset to initialize
10045 a class with a default ctor.
10046 * gcc.c (struct compiler): Remove const qualification.
10047 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
10048 * hash-table.h: Ditto.
10049 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
10051 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
10052 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
10054 * params.h (struct param_info): Make struct members non-const.
10055 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
10056 with default initialization.
10057 * vec.h (vec_copy_construct, vec_default_construct): New helper
10059 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
10060 with vec_copy_construct.
10061 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
10062 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
10063 * doc/invoke.texi (-Wclass-memaccess): Document.
10065 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10067 * emit-rtl.h (is_leaf): Update comment about local
10068 register allocator.
10070 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
10073 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
10074 for a variable to have a section before checking if the section has a
10076 Set section to.persistent if persistent attribute is set.
10077 Warn if .persistent attribute is used on an automatic variable.
10079 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
10081 PR rtl-optimization/80474
10082 * reorg.c (update_block): Do not ignore instructions in a delay slot.
10084 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
10086 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
10089 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
10091 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
10092 (casesi): Emit bounds checking as RTL.
10093 (casesi_internal_mips16_<mode>): Remove bounds checking.
10095 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
10097 * config/xtensa/xtensa.c (xtensa_option_override): Append
10098 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
10099 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
10100 xtensa_doloop_hooks): Define unconditionally.
10101 (xtensa_reorg_loops): Only call reorg_loops in the presence of
10103 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
10104 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
10105 for it in xtensa_option_override.
10106 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
10107 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
10109 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
10111 * doc/cppopts.texi: Document '-' special value to -MF.
10113 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
10115 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
10116 (cortex_a53_fconst): Likewise.
10117 (cortex_a53_fpmul): Likewise.
10118 (cortex_a53_f_load_64): Likewise.
10119 (cortex_a53_f_load_many): Likewise.
10120 (cortex_a53_advsimd_alu): Likewise.
10121 (cortex_a53_advsimd_alu_q): Likewise.
10122 (cortex_a53_advsimd_mul): Likewise.
10123 (cortex_a53_advsimd_mul_q): Likewise.
10124 (fpmac bypass): Add new bypass for fpmac-fpmac case.
10125 Add missing fmul, r2f_cvt and fconst cases.
10127 2017-06-14 Richard Biener <rguenther@suse.de>
10129 PR middle-end/81088
10130 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
10132 (fold_binary_loc): When associating do not treat pre-existing
10133 TREE_OVERFLOW on literal constants as a reason to allow
10134 TREE_OVERFLOW on associated literal constants.
10136 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
10138 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
10139 (MASK_FEATURES): New macro.
10140 * config/sparc/sparc.c (sparc_option_override): Remove the special
10141 handling of -mfpu and generalize it to all MASK_FEATURES switches.
10143 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
10145 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
10146 a division of 0 if non-call exceptions are enabled.
10148 2017-06-14 Andrew Pinski <apinski@cavium.com>
10149 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10152 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
10153 Improve vector initialization code gen for only variable case.
10155 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
10157 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
10159 2017-06-14 Richard Biener <rguenther@suse.de>
10161 PR tree-optimization/81083
10162 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
10165 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10167 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
10168 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
10169 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
10170 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
10171 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
10172 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
10174 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10176 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
10177 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
10179 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10181 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
10183 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10185 * config/rs6000/t-rtems: Don't handle SPE.
10187 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10189 * config/rs6000/t-linux: Don't handle SPE.
10191 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10193 * config/rs6000/eabispe.h: Delete file.
10195 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10197 * config/rs6000/t-spe: Delete file.
10199 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10201 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
10202 (rs6000_legitimate_offset_address_p): Return false for anything in
10203 V2SImode or V2SFmode.
10205 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10207 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
10208 except V2SF and V2SI. Rearrange the vector modes, and add comments.
10209 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
10211 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
10212 (rs6000_legitimate_offset_address_p): Ditto.
10213 (rs6000_emit_move): Ditto.
10214 (rs6000_init_builtins): Remove V4HI_type_node.
10216 2017-06-13 Martin Liska <mliska@suse.cz>
10219 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
10220 (gate_asan): Likewise.
10221 * asan.h (asan_no_sanitize_address_p): Remove the function.
10222 (sanitize_flags_p): New function.
10223 * builtins.def: Fix coding style.
10224 * common.opt: Use renamed enum value.
10225 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
10226 * doc/extend.texi: Document no_sanitize attribute.
10227 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
10228 to SANITIZE_UNDEFINED_NONDEFAULT.
10229 * gcc.c (sanitize_spec_function): Use the renamed enum value.
10230 * gimple-fold.c (optimize_atomic_compare_exchange_p):
10231 Use sanitize_flags_p.
10232 * gimplify.c (gimplify_function_tree): Likewise.
10233 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
10234 * opts.c (parse_no_sanitize_attribute): New function.
10235 (common_handle_option): Use renamed enum value.
10236 * opts.h (parse_no_sanitize_attribute): Declare.
10237 * tree.c (sanitize_flags_p): New function.
10238 * tree.h: Declared here.
10239 * tsan.c: Use sanitize_flags_p.
10240 * ubsan.c (ubsan_expand_null_ifn): Likewise.
10241 (instrument_mem_ref): Likewise.
10242 (instrument_bool_enum_load): Likewise.
10243 (do_ubsan_in_current_function): Remove the function.
10244 (pass_ubsan::execute): Use sanitize_flags_p.
10245 * ubsan.h: Remove do_ubsan_in_current_function
10246 * tree-cfg.c (print_no_sanitize_attr_value): New function.
10247 (dump_function_to_file): Use it here.
10249 2017-06-13 Martin Jambor <mjambor@suse.cz>
10251 PR tree-optimization/80803
10252 PR tree-optimization/81063
10253 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
10254 (propagate_subaccesses_across_link): Enqueue subtree whenever
10255 necessary instead of relying on the caller.
10257 2017-06-13 Martin Jambor <mjambor@suse.cz>
10259 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
10260 that have a first_link.
10261 (sort_and_splice_var_accesses): Do not check first_link before
10263 (subtree_mark_written_and_enqueue): Likewise.
10264 (propagate_all_subaccesses): Likewise and do not stop at first
10265 parent with a first_link.
10267 2017-06-13 Martin Jambor <mjambor@suse.cz>
10269 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
10272 2017-06-13 Yury Gribov <tetra2005@gmail.com>
10274 * match.pd: New pattern.
10276 2017-06-13 Yury Gribov <tetra2005@gmail.com>
10278 * tree-vrp.c (is_masked_range_test): New function.
10279 (register_edge_assert_for): Determine ranges for
10282 2017-06-13 Yury Gribov <tetra2005@gmail.com>
10284 PR tree-optimization/67328
10285 * fold-const.c (maskable_range_p): New function.
10286 (build_range_check): Generate bittests if possible.
10288 2017-06-13 Martin Liska <mliska@suse.cz>
10290 * gimple-pretty-print.c (dump_probability): Add new argument.
10291 (dump_edge_probability): Dump both probability and count.
10292 (dump_gimple_label): Likewise.
10293 (dump_gimple_bb_header): Likewise.
10295 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
10298 * config/avr/avr-devices.c: Fix indentation.
10299 * config/avr/gen-avr-mmcu-specs.c: Dito.
10301 2017-06-13 Richard Biener <rguenther@suse.de>
10303 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
10304 instead get vector type from stmt_info.
10305 (vectorizable_reduction): Adjust. Remove dead code.
10307 2017-06-13 Richard Biener <rguenther@suse.de>
10309 PR middle-end/81065
10310 * fold-const.c (extract_muldiv_1): Remove bogus distribution
10311 case of C * (x * C2 + C3).
10312 (fold_addr_of_array_ref_difference): Properly fold index difference.
10314 2017-06-12 David S. Miller <davem@davemloft.net>
10317 * config/sparc/sparc.md (return expander): Emit frame blockage if
10318 function uses alloca.
10320 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
10322 * combine.c (make_field_assignment): Check len rather than the mode
10323 precision when calling force_to_mode.
10325 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
10327 Support multilibs and devices that see flash in RAM address range.
10330 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
10331 (avr_mcu_t) <flash_pm_offset>: New field.
10332 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
10333 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
10334 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
10335 (AVR_TINY_PM_OFFSET): Remove macro.
10336 * config/avr/avr.opt (-mshort-calls): New option.
10337 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
10338 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
10339 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
10340 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
10341 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
10342 instead of avr_arch->have_jmp_call.
10343 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
10344 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
10345 avr_arch->flash_pm_offset to define.
10346 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
10347 new field flash_pm_offset. Add entry for avrxmega3.
10348 (avr_texinfo): Add entry for avrxmega3.
10349 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
10350 attiny212, attiny214,
10351 attiny412, attiny414, attiny416, attiny417,
10352 attiny814, attiny816, attiny817,
10353 attiny1614, attiny1616, attiny1617,
10354 attiny3214, attiny3216, attiny3217.
10355 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
10356 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
10357 (avr_print_operand_address) [AVR_TINY]: Same.
10358 (avr_asm_init_sections) <readonly_data_section>: Only patch
10359 callback if avr_arch->flash_pm_offset = 0.
10360 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
10361 for rodata if avr_arch->flash_pm_offset != 0.
10362 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
10363 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
10364 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
10365 (m_options): Append opt_rcall.
10366 (m_dirnames): Append dir_rcall.
10367 * config/avr/t-multilib: Regenerate.
10369 * configure.ac [target=avr]: Check whether avrxmega3 default
10370 linker description file works as needed.
10371 * configure: Regenerate.
10372 * doc/avr-mmcu.texi: Regenerate.
10373 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
10374 <__AVR_ARCH__>: Document avrxmega3 and 103.
10375 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
10376 <__AVR_SHORT_CALLS__>: Document it.
10377 <__AVR_PM_BASE_ADDRESS__>: Document it.
10378 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
10379 (AVR Variable Attributes) <progmem>: Document this is
10380 not needed for avrxmega3.
10381 (AVR Named Address Spaces) <__flash>: Dito.
10383 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
10385 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
10387 2017-06-12 Doug Rupp <rupp@adacore.com>
10389 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
10390 Append vxworks-stdint.h to the tm_file list.
10391 * config/vxworks-stdint.h: New file.
10393 2017-06-12 Martin Liska <mliska@suse.cz>
10395 PR tree-optimization/81041
10396 * tree-profile.c (gimple_gen_ic_func_profiler):
10397 Create an extra BB in profile-generate
10398 (gimple_gen_time_profiler): Likewise.
10400 2017-06-12 Jakub Jelinek <jakub@redhat.com>
10402 PR tree-optimization/81003
10403 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
10404 (update_range_test): Use it instead of force_gimple_operand_gsi.
10406 2017-06-12 Richard Biener <rguenther@suse.de>
10408 PR tree-optimization/81053
10409 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
10410 with backedge value not defined in loop. Simplify def stmt
10413 2017-06-11 Tom de Vries <tom@codesourcery.com>
10416 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
10418 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
10419 nvptx_cannot_force_const_mem.
10421 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10423 * opts.c (finish_options): Move test for flag_split_stack after
10424 it has been initialized.
10426 2017-06-11 Jason Merrill <jason@redhat.com>
10428 * tree.h (id_equal): New.
10429 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
10430 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
10431 instead of strcmp of IDENTIFIER_POINTER.
10433 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10435 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
10436 (mark_all_inlined_calls_cdtor): Fix formating.
10437 (inline_transform): Rescale profile before inlining.
10439 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10441 * cgraph.h (cgraph_edge::clone): Update prototype.
10442 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
10443 (cgraph_node::create_clone): Update.
10444 (cgraph_node::create_version_clone): Update.
10445 * tree-inline.c (copy_bb): Update.
10446 (expand_call_inline): Update.
10448 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
10450 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
10451 factored out from ...
10452 (rs6000_emit_prologue): ... here.
10454 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
10456 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
10457 factored out from ...
10458 (rs6000_emit_prologue): ... here.
10460 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10462 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
10464 (handle_missing_profiles): Fix computation of tp_first_run.
10465 (counts_to_freqs): Do not touch freqs when count is 0.
10467 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10469 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
10472 2017-06-10 Tom de Vries <tom@codesourcery.com>
10474 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
10475 attributes): Document signal effective target.
10477 2017-06-10 Tom de Vries <tom@codesourcery.com>
10479 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
10480 Document effective target stack_size.
10482 2017-06-09 David Malcolm <dmalcolm@redhat.com>
10484 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
10485 to the edit_context if they can be auto-applied.
10487 2017-06-9 Ian Lance Taylor <iant@golang.org>
10489 * opts.c (finish_options): If -fsplit-stack, disable implicit
10490 -forder-blocks-and-partition.
10491 * doc/invoke.texi (Optimize Options): Document that when using
10492 -fsplit-stack -forder-blocks-and-partition is not implicitly
10495 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10497 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
10498 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
10499 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
10500 * builtins.def (abort, trap, unreachable): Declare cold.
10501 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
10502 * tree-core.h (ECF_COLD): New.
10503 * tree.c (set_call_expr_flags): Handle ECF_COLD.
10504 (build_common_builtin_nodes): Mark unreachable and abort as cold.
10506 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10508 * predict.c (unlikely_executed_stmt_p): Cleanup.
10510 2017-06-09 Richard Biener <rguenther@suse.de>
10512 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
10513 model if the ref is always written to.
10515 2017-06-09 Tamar Christina <tamar.christina@arm.com>
10517 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
10519 2017-06-09 Tamar Christina <tamar.christina@arm.com>
10521 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
10524 2017-06-09 Tom de Vries <tom@codesourcery.com>
10527 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
10528 "target cannot support label values" when encountering LABEL_REF.
10530 2017-06-09 Martin Liska <mliska@suse.cz>
10532 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
10533 (gimple_gen_ic_func_profiler): Emit direct comparison
10534 of __gcov_indirect_call_callee with NULL.
10535 (gimple_gen_time_profiler): Change probability from
10536 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
10538 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10540 * profile.c (edge_gcov_counts): Turn to pointer.
10541 (compute_branch_probabilities, compute_branch_probabilities): Update.
10542 (branch_prob): Do not clear edge_gcov_count.
10543 * profile.h (edge_gcov_counts): Turn to pointer.
10544 (edge_gcov_count): Update.
10546 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10548 * gimple.h (gimple_check_failed): Mark cold.
10550 2017-06-09 Richard Biener <rguenther@suse.de>
10552 PR tree-optimization/66623
10553 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
10554 refactor check_reduction into two parts, properly computing
10555 whether we have to check reduction validity for outer loop
10558 2017-06-09 Richard Biener <rguenther@suse.de>
10560 PR tree-optimization/79483
10561 * graphite-scop-detection.c (order): New global.
10562 (get_order): Compute bb to order mapping that satisfies code
10563 generation constraints.
10564 (cmp_pbbs): New helper.
10565 (build_scops): Start domwalk at entry block, sort generated
10568 2017-06-09 Richard Biener <rguenther@suse.de>
10570 PR middle-end/81007
10571 * ipa-polymorphic-call.c
10572 (ipa_polymorphic_call_context::restrict_to_inner_class):
10573 Skip FIELD_DECLs with error_mark_node type.
10574 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
10577 2017-06-09 Martin Liska <mliska@suse.cz>
10579 * predict.c (struct branch_predictor): New struct.
10580 (test_prediction_value_range): New test.
10581 (predict_c_tests): New function.
10582 * selftest-run-tests.c (selftest::run_tests): Run the function.
10583 * selftest.h: Declare new tests.
10585 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
10588 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
10589 gen_add3_insn did not fail.
10590 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
10591 r0, construct that number in a temporary reg and add that reg to r0.
10592 If asked to put the result in r0 as well, fail.
10594 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
10596 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10597 for early expansion of vec_eqv.
10599 2017-06-08 Jakub Jelinek <jakub@redhat.com>
10601 PR middle-end/81005
10602 * ubsan.c (instrument_null): Avoid pointless code temporary.
10603 (pass_ubsan::execute): Instrument aggregate arguments of calls.
10605 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
10609 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
10612 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
10615 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10617 * predict.c (unlikely_executed_edge_p): Move ahead.
10618 (probably_never_executed_edge_p): Use it.
10620 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
10622 PR middle-end/79988
10623 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
10624 gimple_call_builtin_p call.
10626 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10628 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
10629 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
10630 rtl_check_failed_type2, rtl_check_failed_code1,
10631 rtl_check_failed_code2, rtl_check_failed_code_mode,
10632 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
10633 rtvec_check_failed_bounds, rtl_check_failed_flag,
10634 _fatal_insn_not_found, _fatal_insn): Likewise.
10635 * tree.h (tree_contains_struct_check_failed,
10636 tree_check_failed, tree_not_check_failed,
10637 tree_class_check_failed, tree_range_check_failed,
10638 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
10639 tree_vec_elt_check_failed, phi_node_elt_check_failed,
10640 tree_operand_check_failed, omp_clause_check_failed,
10641 omp_clause_operand_check_failed, omp_clause_range_check_failed):
10644 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10646 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
10647 flag_branch_probabilities.
10648 * ipa-inline.c (edge_badness): Likewise.
10649 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
10650 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
10651 * predict.c (maybe_hot_frequency_p): Likewise.
10652 (probably_never_executed): Likewise.
10653 * sched-ebb.c (schedule_ebbs): Likewise.
10654 * sched-rgn.c (find_single_block_region): Likewise.
10655 * tracer.c (tail_duplicate): Likewise.
10657 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10659 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
10660 longer requires x_flag_profile_use.
10662 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10664 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
10665 instead of flag_reorder_blocks_and_partition.
10666 * dbxout.c (dbxout_function_end): Likewise.
10667 * dwarf2out.c (gen_subprogram_die): Likewise.
10668 * haifa-sched.c (sched_create_recovery_edges): Likewise.
10669 * hw-doloop.c (reorg_loops): Likewise.
10670 * varasm.c (assemble_start_function,
10671 assemble_end_function): Likewise.
10672 (decide_function_section): Do not check for
10673 flag_reorder_blocks_and_partition.
10675 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
10677 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
10679 (chkp_get_hard_register_fake_addr_expr): Ditto.
10680 (chkp_build_addr_expr): Add check for hard reg case.
10681 (chkp_parse_array_and_component_ref): Ditto.
10682 (chkp_find_bounds_1): Ditto.
10683 (chkp_process_stmt): Don't generate bounds store for
10686 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10688 * predict.c (maybe_hot_bb_p): Do not check profile status.
10689 (maybe_hot_edge_p): Likewise.
10690 (probably_never_executed): Check for zero counts even if profile
10692 (unlikely_executed_edge_p): New function.
10693 (unlikely_executed_stmt_p): New function.
10694 (unlikely_executed_bb_p): New function.
10695 (set_even_probabilities): Use unlikely predicates.
10696 (combine_predictions_for_bb): Likewise.
10697 (predict_paths_for_bb): Likewise.
10698 (predict_paths_leading_to_edge): Likewise.
10699 (determine_unlikely_bbs): New function.
10700 (estimate_bb_frequencies): Use it.
10701 (compute_function_frequency): Use zero counts even if profile is
10703 * profile-count.h: Fix typo.
10705 2017-08-08 Julia Koval <julia.koval@intel.com>
10707 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
10708 _mm512_mask_cvtsepi16_storeu_epi8,
10709 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
10710 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
10711 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
10712 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
10713 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
10714 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
10715 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
10716 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
10717 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
10718 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
10719 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
10720 __builtin_ia32_pmovuswb256mem_mask,
10721 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
10722 __builtin_ia32_pmovwb512mem_mask): New builtins.
10724 2017-08-08 Julia Koval <julia.koval@intel.com>
10726 PR target/73350,80862
10727 * config/i386/subst.md (round): Fix round pattern.
10728 * config/i386/i386.c (ix86_erase_embedded_rounding):
10729 Fix erasing rounding for the fixed pattern.
10731 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10733 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
10735 2017-06-08 Martin Liska <mliska@suse.cz>
10737 PR gcov-profile/80911
10738 * gcov.c (block_info::block_info): New constructor.
10740 2017-06-07 Carl Love <cel@us.ibm.com>
10742 * config/rs6000/rs6000-c: The return type of the following
10743 built-in functions was implemented as int not long long. Fix sign
10744 of return value for the unsigned version of vec_mulo and vec_mule.
10745 vector unsigned long long vec_bperm (vector unsigned long long,
10746 vector unsigned char)
10747 vector signed long long vec_mule (vector signed int,
10749 vector unsigned long long vec_mule (vector unsigned int,
10750 vector unsigned int)
10751 vector signed long long vec_mulo (vector signed int,
10753 vector unsigned long long vec_mulo (vector unsigned int,
10754 vector unsigned int)
10755 * doc/extend.texi: Fix the documentation for the built-in
10758 2017-06-07 Carl Love <cel@us.ibm.com>
10761 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
10764 2017-06-07 Carl Love <cel@us.ibm.com>
10766 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
10767 support, Generate doublehv for signed int/float for BE case only.
10769 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
10771 * doc/invoke.texi (mcx16): Rewrite.
10773 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10775 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
10776 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
10777 *mov<mode>_softfloat, and an anonymous splitter): Use
10778 nonimmediate_operand instead of rs6000_nonimmediate_operand.
10780 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10782 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
10784 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
10785 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
10786 (rs6000_debug_reg_global): Adjust.
10787 (rs6000_init_hard_regno_mode_ok): Adjust.
10788 (rs6000_dbx_register_number): Adjust.
10789 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
10790 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
10791 Remove SPE_ACC and SPEFSCR.
10792 (REG_ALLOC_ORDER): Ditto.
10793 (FRAME_POINTER_REGNUM): Change to 111.
10794 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
10795 (REG_CLASS_NAMES): Ditto.
10796 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
10797 (REGISTER_NAMES): Ditto.
10798 (ADDITIONAL_REG_NAMES): Ditto.
10799 (rs6000_reg_names): Ditto.
10800 * config/rs6000/rs6000.md: Renumber some register number
10803 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10805 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
10807 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
10808 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
10810 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
10811 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
10812 Delete the SPE high registers.
10813 (REG_ALLOC_ORDER): Ditto.
10814 (enum reg_class): Remove SPE_HIGH_REGS.
10815 (REG_CLASS_NAMES): Ditto.
10816 (REG_CLASS_CONTENTS): Delete the SPE high registers.
10817 (REGISTER_NAMES): Ditto.
10818 (rs6000_reg_names): Ditto.
10819 * doc/tm.texi.in: Remove SPE as example.
10820 * doc/tm.texi: Regenerate.
10822 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10824 * config/rs6000/8540.md (ppc8540_brinc): Delete.
10825 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
10826 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
10827 * config/rs6000/rs6000.md (type): Remove "brinc".
10829 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10831 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
10832 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
10833 * config/rs6000/linuxspe.h: Delete file.
10834 * config/rs6000/rs6000.md: Don't include spe.md.
10835 * config/rs6000/spe.h: Delete file.
10836 * config/rs6000/spe.md: Delete file.
10837 * config/rs6000/t-rs6000: Remove spe.md.
10839 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10841 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
10842 (reg_or_none500mem_operand): Delete.
10843 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
10844 instead of reg_or_none500mem_operand.
10846 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10848 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
10849 handling of SPE flags.
10850 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
10852 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10854 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
10856 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
10857 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
10858 paired_divv2sf3): Similar.
10859 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
10860 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
10861 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
10863 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
10864 Rename the paired_* instruction patterns.
10865 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
10867 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
10868 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
10869 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
10870 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
10871 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
10872 PAIRED_VECTOR_MODE.
10873 (struct machine_function): Delete field spe_insn_chain_scanned_p.
10874 (spe_func_has_64bit_regs_p): Delete.
10875 (spe_expand_predicate_builtin): Delete.
10876 (spe_expand_evsel_builtin): Delete.
10877 (TARGET_DWARF_REGISTER_SPAN): Do not define.
10878 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
10879 (invalid_e500_subreg): Delete.
10880 (rs6000_legitimize_address): Always force_reg op2 as well, for
10881 paired single memory accesses.
10882 (rs6000_member_type_forces_blk): Delete.
10883 (rs6000_spe_function_arg): Delete.
10884 (rs6000_expand_unop_builtin): Delete SPE handling.
10885 (rs6000_expand_binop_builtin): Ditto.
10886 (spe_expand_stv_builtin): Delete.
10887 (bdesc_2arg_spe): Delete.
10888 (spe_expand_builtin): Delete.
10889 (spe_expand_predicate_builtin): Delete.
10890 (spe_expand_evsel_builtin): Delete.
10891 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
10892 (spe_init_builtins): Delete.
10893 (spe_func_has_64bit_regs_p): Delete.
10894 (savres_routine_name): Delete "info" parameter. Adjust callers.
10895 (rs6000_emit_stack_reset): Ditto.
10896 (rs6000_dwarf_register_span): Delete.
10897 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
10898 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
10899 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
10901 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
10903 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
10904 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
10905 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
10906 mulv2sf3, divv2sf3): Delete expanders.
10908 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10910 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
10912 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10914 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
10915 * config/rs6000/rs6000.c: Ditto.
10917 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10919 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
10920 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
10921 comparison_operator.
10923 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10925 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
10926 * config/rs6000/rs6000.opt: Ditto.
10927 * config/rs6000/t-rtems: Ditto.
10929 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10931 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
10932 TARGET_E500_SINGLE by 0, simplify.
10933 * config/rs6000/rs6000.c: Ditto.
10934 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
10935 (spe_build_register_parallel): Delete.
10936 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
10937 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
10938 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
10939 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
10940 (E500_CONVERT): Delete.
10941 * config/rs6000/spe.md: Remove many patterns and all define_constants.
10943 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10945 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
10946 * config/rs6000/dfp.md: Ditto.
10947 (negdd2, *negdd2_fpr): Merge.
10948 (absdd2, *absdd2_fpr): Merge.
10949 (negtd2, *negtd2_fpr): Merge.
10950 (abstd2, *abstd2_fpr): Merge.
10951 * config/rs6000/e500.h: Delete file.
10952 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
10953 TARGET_FPRS by 1 and simplify.
10954 * config/rs6000/rs6000-c.c: Ditto.
10955 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
10956 TARGET_DF_SPE by 0.
10957 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
10959 * config/rs6000/rs6000.md: Ditto.
10960 (floatdidf2, *floatdidf2_fpr): Merge.
10961 (move_from_CR_gt_bit): Delete.
10962 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
10963 (E500_CR_IOR_COMPARE): Delete.
10964 (All patterns that require !TARGET_FPRS): Delete.
10965 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
10967 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10969 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
10971 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10973 * graphds.c (add_edge): Intitialize edge's attached data.
10974 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
10975 pointer parameter. Call pointed function on each edge during
10976 graph traversing. Skip traversing the edge when the function
10978 (graphds_dfs, graphds_scc): Ditto.
10979 (for_each_edge): New parameter. Pass the new parameter to callback
10981 * graphds.h (skip_edge_callback): New function pointer type.
10982 (graphds_dfs, graphds_scc): New function pointer parameter.
10983 (graphds_edge_callback, for_each_edge): New parameter.
10985 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10987 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
10988 out code checking if runtime alias check is possible to below ...
10989 Call the new function.
10990 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
10991 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
10993 2017-06-07 Marek Polacek <polacek@redhat.com>
10996 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
10997 TYPE_OVERFLOW_WRAPS check.
10999 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11001 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
11002 if versioning is required.
11003 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
11004 peeling with the check for versioning.
11006 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11008 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
11009 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
11010 Set true to new parameter if new ssa variable is defined.
11011 (vect_gen_vector_loop_niters): Refactor. Set range information
11012 for the new vector loop bound variable.
11013 (vect_do_peeling): Ditto.
11015 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11017 * tree-affine.c (ssa.h): Include header file.
11018 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
11019 has wrapping overflow behavior.
11021 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11023 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
11025 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11027 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
11028 (tree_to_aff_combination): ... here.
11030 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11032 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
11033 reg_pressure model function.
11034 (ivopts_global_cost_for_size): Delete.
11035 (determine_set_costs, iv_ca_recount_cost): Call new model function
11036 ivopts_estimate_reg_pressure.
11038 2017-06-07 Tamar Christina <tamar.christina@arm.com>
11040 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
11041 expensive than udiv. Remove floating point cases from mod.
11043 2017-06-07 Tamar Christina <tamar.christina@arm.com>
11045 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
11046 Increase idiv cost.
11048 2017-06-07 Tamar Christina <tamar.christina@arm.com>
11050 * config/aarch64/aarch64.md
11051 (copysignsf3): Fix mask generation.
11053 2017-06-07 Jakub Jelinek <jakub@redhat.com>
11055 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
11057 (class dump_manager): Add register_dumps method.
11058 * dumpfile.c: Include langhooks.h.
11059 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
11060 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
11061 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
11062 (dump_manager::dump_register): Start with 512 entries instead of 32.
11063 (dump_manager::register_dumps): New method.
11064 * toplev.c (general_init): Instead of invoking register_dumps
11065 langhook, invoke register_dumps method on the dump manager.
11066 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
11069 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
11071 * doc/md.texi: Clarify the restrictions on a define_insn condition.
11072 Say that # requires an associated define_split to exist, and that
11073 the define_split must be suitable for use after register allocation.
11075 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11077 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
11078 (compute_outgoing_frequencies): Also initialize zero counts.
11079 (find_many_sub_basic_blocks): Do not produce uninitialized profile
11080 around loops; preserve more of profile when nothing changes.
11082 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
11084 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
11086 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
11087 * config/arm/arm-cpu-cdata.h: Regenerate.
11088 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
11089 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
11090 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
11091 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
11092 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
11094 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
11096 * config/arm/t-rmprofile: Likewise.
11097 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
11099 2017-06-06 David S. Miller <davem@davemloft.net>
11102 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
11103 blockage if function uses alloca.
11105 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11107 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
11108 New "uid" fields to hold pretty-print IDs of group and ref.
11109 Memory references are now identified as <group_id>:<ref_id>
11110 instead of using [random] addresses.
11111 (dump_mem_details): Simplify, no functional change.
11112 (dump_mem_ref): Simplify and make output more concise.
11113 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
11114 (find_or_create_group): Initialize group uid.
11115 (record_ref): Initialize ref uid. Improve debug output.
11116 (prune_group_by_reuse, should_issue_prefetch_p,)
11117 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
11118 (mark_nontemporal_store, determine_loop_nest_reuse):
11119 Improve debug output.
11121 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11123 * dbgcnt.def (prefetch): New debug counter.
11124 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
11125 (schedule_prefetches): Stop issueing prefetches if debug counter
11128 2017-06-06 Tom de Vries <tom@codesourcery.com>
11130 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
11131 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
11133 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11135 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
11136 Use aarch64_reg_or_zero predicate for operand 4.
11137 (aarch64_compare_and_swap<mode> define_insn_and_split):
11138 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
11139 (aarch64_store_exclusive<mode>): Likewise for operand 2.
11141 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
11143 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
11144 (arm_compute_save_core_reg_mask): This.
11145 (thumb1_compute_save_reg_mask): Rename into ...
11146 (thumb1_compute_save_core_reg_mask): This.
11147 (arm_compute_save_reg0_reg12_mask): Adapt comment.
11148 (arm_compute_frame_layout): Likewise.
11150 2017-06-06 Richard Biener <rguenther@suse.de>
11152 PR tree-optimization/80974
11153 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
11154 keep or clear leaders SSA info.
11156 2017-06-06 Tom de Vries <tom@codesourcery.com>
11158 * config/nvptx/nvptx.c (split_mode_p): New function.
11159 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
11161 2017-06-06 Tom de Vries <tom@codesourcery.com>
11163 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
11165 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11168 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
11171 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11173 * shrink-wrap.c (handle_simple_exit): Update profile.
11174 (try_shrink_wrapping): Upate profile.
11176 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11178 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
11179 (tree_guess_outgoing_edge_probabilities): New.
11180 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
11181 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
11183 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11185 * ipa-split.c (split_function): Initialize return bb profile.
11187 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11189 * profile.c (compute_branch_probabilities): Also initialize
11190 EXIT_BLOCK profile.
11192 2017-06-06 Richard Biener <rguenther@suse.de>
11194 PR tree-optimization/80928
11195 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
11196 (vect_analyze_loop_operations): Properly guard analysis for
11198 (vect_transform_loop): Likewise.
11199 (vect_analyze_loop_2): Also reset SLP type on PHIs.
11200 (vect_model_induction_cost): Do not cost for pure SLP.
11201 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
11202 of induction in inner loop vectorization.
11203 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
11204 (vect_get_and_check_slp_defs): Handle vect_induction_def.
11205 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
11207 (vect_analyze_slp_cost_1): Cost induction.
11208 (vect_detect_hybrid_slp_stmts): Handle PHIs.
11209 (vect_get_slp_vect_defs): Likewise.
11210 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
11211 (vect_transform_stmt): Handle SLP reductions.
11212 * tree-vectorizer.h (vectorizable_induction): Adjust.
11214 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
11216 * config/rs6000/rs6000.c (make_resolver_func): Update
11217 init_lowered_empty_function call.
11219 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
11221 * doc/invoke.texi: Document the -fprofile-abs-path option.
11222 * common.opt (fprofile-abs-path): New option.
11223 * gcov-io.h (gcov_write_filename): Declare.
11224 * gcov-io.c (gcov_write_filename): New function.
11225 * coverage.c (coverage_begin_function): Use gcov_write_filename.
11226 * profile.c (output_location): Likewise.
11228 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
11230 * shring-wrap.c: Revert accidental commit.
11232 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
11234 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
11236 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
11238 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
11240 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
11241 profile in callgraph edge.
11242 * profile-count.h (apply_probability): If THIS is 0, then result is 0
11243 (apply_scale): Likewise.
11244 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
11245 Also scale profile when inlining function with zero profile.
11246 (initialize_cfun): Update exit block profile even when it is zero.
11247 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
11248 when profile is read.
11250 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
11252 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
11253 (CLONE_*): New constants to define the processors we can generate
11254 code for with the target_clone attribute.
11255 (rs6000_clone_map): New array to identify which clone processors
11256 the current program is running on.
11257 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
11258 target_clone attribute.
11259 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
11260 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
11261 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
11262 (cpu_expand_builtin): Add support for target_clone attribute.
11263 (rs6000_valid_attribute_p): Allow "default" attribute.
11264 (get_decl_name): New debug function to simplify printing the
11265 current function name in debugging statements.
11266 (rs6000_clone_priority): New functions to support the target_clone
11267 attribute, and be able to generate code to switch between ISA 2.05
11268 through ISA 3.0 (power6 through power9).
11269 (rs6000_compare_version_priority): Likewise.
11270 (rs6000_get_function_versions_dispatcher): Likewise.
11271 (make_resolver_func): Likewise.
11272 (add_condition_to_bb): Likewise.
11273 (dispatch_function_versions): Likewise.
11274 (rs6000_generate_version_dispatcher_body): Likewise.
11275 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
11276 (fusion_gpr_load_p): Fix a spacing issue.
11277 * doc/extend.texi (Common Function Attributes): Document that the
11278 PowerPC supports the target_clone attribute.
11280 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
11282 * config/arm/arm.h: explain F symbol found in description of ARM
11283 register allocation in its legend.
11285 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
11287 * config/mips/frame-header-opt.c: Include profile-count.h.
11288 * config/riscv/riscv.c: Include profile-count.h
11290 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
11292 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
11294 (sm_set_flag_if_changed): Add bbs field.
11295 (execute_sm_if_changed_flag_set): Pass BBS.
11296 (execute_sm): Update.
11298 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11300 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
11303 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11305 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
11306 (peephole2): New peephole2 to emit the above.
11307 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
11309 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11311 * config/aarch64/aarch64.c (define_peephole2 above
11312 *sub_<shift>_<mode>): New peephole.
11314 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11316 * config/i386/i386.c (make_resolver_func): Update.
11317 * Makefile.in: Add profile-count.h and profile-count.o
11318 * auto-profile.c (afdo_indirect_call): Update to new API.
11319 (afdo_set_bb_count): Update.
11320 (afdo_propagate_edge): Update.
11321 (afdo_propagate_circuit): Update.
11322 (afdo_calculate_branch_prob): Update.
11323 (afdo_annotate_cfg): Update.
11324 * basic-block.h: Include profile-count.h
11325 (struct edge_def): Turn count to profile_count.
11326 (struct basic_block_def): Likewie.
11327 (REG_BR_PROB_BASE): Move to profile-count.h
11328 (RDIV): Move to profile-count.h
11329 * bb-reorder.c (max_entry_count): Turn to profile_count.
11330 (find_traces): Update.
11331 (rotate_loop):Update.
11332 (connect_traces):Update.
11333 (sanitize_hot_paths):Update.
11334 * bt-load.c (migrate_btr_defs): Update.
11335 * cfg.c (RDIV): Remove.
11336 (init_flow): Use alloc_block.
11337 (alloc_block): Uninitialize count.
11338 (unchecked_make_edge): Uninitialize count.
11339 (check_bb_profile): Update.
11340 (dump_edge_info): Update.
11341 (dump_bb_info): Update.
11342 (update_bb_profile_for_threading): Update.
11343 (scale_bbs_frequencies_int): Update.
11344 (scale_bbs_frequencies_gcov_type): Update.
11345 (scale_bbs_frequencies_profile_count): New.
11346 * cfg.h (update_bb_profile_for_threading): Update.
11347 (scale_bbs_frequencies_profile_count): Declare.
11348 * cfgbuild.c (compute_outgoing_frequencies): Update.
11349 (find_many_sub_basic_blocks): Update.
11350 * cfgcleanup.c (try_forward_edges): Update.
11351 (try_crossjump_to_edge): Update.
11352 * cfgexpand.c (expand_gimple_tailcall): Update.
11353 (construct_exit_block): Update.
11354 * cfghooks.c (verify_flow_info): Update.
11355 (dump_bb_for_graph): Update.
11356 (split_edge): Update.
11357 (make_forwarder_block): Update.
11358 (duplicate_block): Update.
11359 (account_profile_record): Update.
11360 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
11361 (get_estimated_loop_iterations): Update.
11362 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
11363 (single_likely_exit): Update.
11364 * cfgloopmanip.c (scale_loop_profile): Update.
11366 (set_zero_probability): Update.
11367 (lv_adjust_loop_entry_edge): Update.
11368 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
11369 (purge_dead_edges): Update.
11370 (rtl_account_profile_record): Update.
11371 * cgraph.c (cgraph_node::create): Uninitialize count.
11372 (symbol_table::create_edge): Uninitialize count.
11373 (cgraph_update_edges_for_call_stmt_node): Update.
11374 (cgraph_edge::dump_edge_flags): Update.
11375 (cgraph_node::dump): Update.
11376 (cgraph_edge::maybe_hot_p): Update.
11377 * cgraph.h: Include profile-count.h
11378 (create_clone), create_edge, create_indirect_edge): Update.
11379 (cgraph_node): Turn count to profile_count.
11380 (cgraph_edge0: Likewise.
11381 (make_speculative, clone): Update.
11382 (create_edge): Update.
11383 (init_lowered_empty_function): Update.
11384 * cgraphclones.c (cgraph_edge::clone): Update.
11385 (duplicate_thunk_for_node): Update.
11386 (cgraph_node::create_clone): Update.
11387 * cgraphunit.c (cgraph_node::analyze): Update.
11388 (cgraph_node::expand_thunk): Update.
11389 * final.c (dump_basic_block_info): Update.
11390 * gimple-streamer-in.c (input_bb): Update.
11391 * gimple-streamer-out.c (output_bb): Update.
11392 * graphite.c (print_global_statistics): Update.
11393 (print_graphite_scop_statistics): Update.
11394 * hsa-brig.c: Include basic-block.h.
11395 * hsa-dump.c: Include basic-block.h.
11396 * hsa-gen.c (T sum_slice): Update.
11397 (convert_switch_statements):Update.
11398 * hsa-regalloc.c: Include basic-block.h.
11399 * ipa-chkp.c (chkp_produce_thunks): Update.
11400 * ipa-cp.c (struct caller_statistics): Update.
11401 (init_caller_stats): Update.
11402 (gather_caller_stats): Update.
11403 (ipcp_cloning_candidate_p): Update.
11404 (good_cloning_opportunity_p): Update.
11405 (get_info_about_necessary_edges): Update.
11406 (dump_profile_updates): Update.
11407 (update_profiling_info): Update.
11408 (update_specialized_profile): Update.
11409 (perhaps_add_new_callers): Update.
11410 (decide_about_value): Update.
11411 (ipa_cp_c_finalize): Update.
11412 * ipa-devirt.c (struct odr_type_warn_count): Update.
11413 (struct decl_warn_count): Update.
11414 (struct final_warning_record): Update.
11415 (possible_polymorphic_call_targets): Update.
11416 (ipa_devirt): Update.
11417 * ipa-fnsummary.c (redirect_to_unreachable): Update.
11418 * ipa-icf.c (sem_function::merge): Update.
11419 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
11420 * ipa-inline.c (compute_uninlined_call_time): Update.
11421 (compute_inlined_call_time): Update.
11422 (want_inline_small_function_p): Update.
11423 (want_inline_self_recursive_call_p): Update.
11424 (edge_badness): Update.
11425 (lookup_recursive_calls): Update.
11426 (recursive_inlining): Update.
11427 (inline_small_functions): Update.
11428 (dump_overall_stats): Update.
11429 (dump_inline_stats): Update.
11430 * ipa-profile.c (ipa_profile_generate_summary): Update.
11431 (ipa_propagate_frequency): Update.
11432 (ipa_profile): Update.
11433 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
11434 * ipa-utils.c (ipa_merge_profiles): Update.
11435 * loop-doloop.c (doloop_modify): Update.
11436 * loop-unroll.c (report_unroll): Update.
11437 (unroll_loop_runtime_iterations): Update.
11438 * lto-cgraph.c (lto_output_edge): Update.
11439 (lto_output_node): Update.
11440 (input_node): Update.
11441 (input_edge): Update.
11442 (merge_profile_summaries): Update.
11443 * lto-streamer-in.c (input_cfg): Update.
11444 * lto-streamer-out.c (output_cfg): Update.
11445 * mcf.c (create_fixup_graph): Update.
11446 (adjust_cfg_counts): Update.
11447 (sum_edge_counts): Update.
11448 * modulo-sched.c (sms_schedule): Update.
11449 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
11450 * predict.c (maybe_hot_count_p): Update.
11451 (probably_never_executed): Update.
11452 (dump_prediction): Update.
11453 (combine_predictions_for_bb): Update.
11454 (propagate_freq): Update.
11455 (handle_missing_profiles): Update.
11456 (counts_to_freqs): Update.
11457 (rebuild_frequencies): Update.
11458 (force_edge_cold): Update.
11459 * predict.h: Include profile-count.h
11460 (maybe_hot_count_p, counts_to_freqs): UPdate.
11461 * print-rtl-function.c: Do not include cfg.h
11462 * print-rtl.c: Include basic-block.h
11463 * profile-count.c: New file.
11464 * profile-count.h: New file.
11465 * profile.c (is_edge_inconsistent): Update.
11466 (correct_negative_edge_counts): Update.
11467 (is_inconsistent): Update.
11468 (set_bb_counts): Update.
11469 (read_profile_edge_counts): Update.
11470 (compute_frequency_overlap): Update.
11471 (compute_branch_probabilities): Update; Initialize and deinitialize
11473 (branch_prob): Update.
11474 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
11475 (edge_gcov_count): New.
11476 (bb_gcov_count): New.
11477 * shrink-wrap.c (try_shrink_wrapping): Update.
11478 * tracer.c (better_p): Update.
11479 * trans-mem.c (expand_transaction): Update.
11480 (ipa_tm_insert_irr_call): Update.
11481 (ipa_tm_insert_gettmclone_call): Update.
11482 * tree-call-cdce.c: Update.
11483 * tree-cfg.c (gimple_duplicate_sese_region): Update.
11484 (gimple_duplicate_sese_tail): Update.
11485 (gimple_account_profile_record): Update.
11486 (execute_fixup_cfg): Update.
11487 * tree-inline.c (copy_bb): Update.
11488 (copy_edges_for_bb): Update.
11489 (initialize_cfun): Update.
11490 (freqs_to_counts): Update.
11491 (copy_cfg_body): Update.
11492 (expand_call_inline): Update.
11493 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11494 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
11495 (try_unroll_loop_completely): Update.
11496 (try_peel_loop): Update.
11497 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11498 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
11499 * tree-ssa-loop-split.c (connect_loops): Update.
11500 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
11501 * tree-ssa-reassoc.c (branch_fixup): Update.
11502 * tree-ssa-tail-merge.c (replace_block_by): Update.
11503 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
11504 (compute_path_counts): Update.
11505 (update_profile): Update.
11506 (recompute_probabilities): Update.
11507 (update_joiner_offpath_counts): Update.
11508 (estimated_freqs_path): Update.
11509 (freqs_to_counts_path): Update.
11510 (clear_counts_path): Update.
11511 (ssa_fix_duplicate_block_edges): Update.
11512 (duplicate_thread_path): Update.
11513 * tree-switch-conversion.c (case_bit_test_cmp): Update.
11514 (struct switch_conv_info): Update.
11515 * tree-tailcall.c (decrease_profile): Update.
11516 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11517 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11518 * value-prof.c (check_counter): Update.
11519 (gimple_divmod_fixed_value): Update.
11520 (gimple_mod_pow2): Update.
11521 (gimple_mod_subtract): Update.
11522 (gimple_ic_transform): Update.
11523 (gimple_stringop_fixed_value): Update.
11524 * value-prof.h (gimple_ic): Update.
11526 2017-06-02 Carl Love <cel@us.ibm.com>
11528 * config/rs6000/rs6000-c: Add support for built-in functions
11529 vector double vec_doublee (vector signed int);
11530 vector double vec_doublee (vector unsigned int);
11531 vector double vec_doublee (vector float);
11532 vector double vec_doubleh (vector signed int);
11533 vector double vec_doubleh (vector unsigned int);
11534 vector double vec_doubleh (vector float);
11535 vector double vec_doublel (vector signed int);
11536 vector double vec_doublel (vector unsigned int);
11537 vector double vec_doublel (vector float);
11538 vector double vec_doubleo (vector signed int);
11539 vector double vec_doubleo (vector unsigned int);
11540 vector double vec_doubleo (vector float);.
11541 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
11542 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
11544 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
11545 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
11546 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
11548 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
11549 vec_doublel, vec_doubleh.
11550 * doc/extend.texi: Update the built-in documentation file for the
11551 new built-in functions.
11553 2017-06-02 David Malcolm <dmalcolm@redhat.com>
11556 * ipa-inline-analysis.c (free_growth_caches): Set
11557 edge_removal_hook_holder to NULL after removing it.
11559 2017-06-02 Sudakshina Das <sudi.das@arm.com>
11561 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
11562 comparision with zero.
11564 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
11565 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11566 for early expansion of vec_min and vec_max builtins.
11567 (builtin_function_type): Add min/max unsigned variants to those
11568 identified as having unsigned arguments.
11570 2017-06-02 Olivier Hainque <hainque@adacore.com>
11572 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
11574 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11576 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
11577 Use VALL_F16 iterator rather than VALL.
11579 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11581 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
11582 Emit CBNZ inside loop when doing a strong exchange and comparing
11583 against zero. Generate the CC flags after the loop.
11585 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
11587 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
11588 (dl_section_ref): New.
11589 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
11590 On AIX, append an expression to subtract the size of the
11591 section length to dl_section_ref.
11593 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
11595 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11596 for early expansion of vector absolute builtins.
11598 2017-06-02 Richard Biener <rguenther@suse.de>
11600 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
11601 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
11603 2017-06-02 Richard Biener <rguenther@suse.de>
11605 PR tree-optimization/80948
11606 * tree-tailcall.c (find_tail_calls): Track stmts to move in
11607 stmt order as well.
11609 2017-06-02 Richard Biener <rguenther@suse.de>
11611 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
11613 * tree-vect-stmts.c (process_use): Do not mark backedge defs
11614 for inductions as relevant.
11616 2017-06-02 Richard Biener <rguenther@suse.de>
11618 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
11619 (vectorizable_induction): ... this. Remove dead code.
11621 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
11623 * builtins. (expand_builtin_alloca): Remove second parameter and
11624 infer its value from the first parameter instead.
11625 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
11627 2017-06-02 Jakub Jelinek <jakub@redhat.com>
11629 PR rtl-optimization/80903
11630 * loop-doloop.c (add_test): Unshare sequence.
11632 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
11634 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
11636 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
11638 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
11640 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
11641 xlogue_layout::get_instance, logue_layout::xlogue_layout,
11642 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
11643 (xlogue_layout::get_stub_rtx): Make static.
11644 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
11645 (xlogue_layout::compute_stub_managed_regs): Rename to...
11646 (xlogue_layout::count_stub_managed_regs): ...this.
11647 (xlogue_layout::is_stub_managed_reg): New function.
11648 (xlogue_layout::m_stub_names): Rename to...
11649 (xlogue_layout::s_stub_names): ...this, make static.
11650 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
11651 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
11652 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
11653 xlogue_layout::s_stub_names): Instantiate statics.
11654 (stub_managed_regs): Remove.
11655 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
11656 (disable_call_ms2sysv_xlogues): Rename to...
11657 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
11658 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
11660 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
11661 change after reload_completed.
11662 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
11664 (ix86_expand_prologue): Likewise.
11665 (ix86_expand_epilogue): Likewise.
11666 (ix86_expand_split_stack_prologue): Likewise.
11667 (ix86_compute_frame_layout): Remove frame parameter ...
11668 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
11669 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
11671 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
11672 (ix86_frame): Move from here ...
11673 * config/i386/i386.h (ix86_frame): ... to here.
11674 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
11675 complete ix86_frame data structure instead. Remove some_ld_name.
11677 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
11679 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
11680 symbols that hold a DECL_VALUE_EXPR.
11682 2017-06-01 Martin Jambor <mjambor@suse.cz>
11684 PR tree-optimization/80898
11685 * tree-sra.c (process_subtree_disqualification): Removed.
11686 (disqualify_candidate): Do not acll
11687 process_subtree_disqualification.
11688 (subtree_mark_written_and_enqueue): New function.
11689 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
11690 RHS has been disqualified and re-queue LHS if necessary. Apart
11691 from that, ignore disqualified RHS.
11693 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11695 * config/s390/s390.c (s390_emit_epilogue): Disable early return
11696 address fetch for z10 or later.
11698 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11700 * config/arc/arc.md (tst_movb): Add guard when splitting.
11702 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11704 * config/arc/arc.c (arc_can_eliminate): Test against
11705 arc_frame_pointer_needed.
11707 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11709 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
11710 to prevent store reordering.
11711 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
11712 (type): Add block type.
11713 (stack_tie): Define special instruction to be used in
11716 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11718 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
11719 constraint. It is not valid for the pattern.
11720 (noncommutative_binary_comparison): Likewise.
11722 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11724 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
11727 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11729 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
11730 be used by the reg-alloc.
11732 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11734 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
11735 reg-alloc when having mul64 or mul32x16 instructions.
11736 (mulsidi3): Likewise.
11737 (umulsidi3): Likewise.
11738 (mulsi32x16): New pattern.
11739 (mulsi64): Likewise.
11740 (mulsidi64): Likewise.
11741 (umulsidi64): Likewise.
11742 (MUL32x16_REG): Define.
11743 (mul64_600): Use MUL32x16_REG.
11744 (mac64_600): Likewise.
11745 (umul64_600): Likewise.
11746 (umac64_600): Likewise.
11748 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11750 * config/arc/arc.md (mulsi3_700): Make it commutative.
11752 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
11754 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
11756 (*sign_extendsidi2_insn): Likewise for movstosw.
11758 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
11760 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
11761 the type of the input discriminant value. Convert the
11762 discriminant value of signedness vary.
11764 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
11766 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
11767 Add to -Wall section.
11769 2017-06-01 Richard Biener <rguenther@suse.de>
11771 PR middle-end/66313
11772 * fold-const.c (fold_plusminus_mult_expr): If the factored
11773 factor may be zero use a wrapping type for the inner operation.
11774 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
11775 and handle moved defs.
11776 (process_assignment): Properly guard the unary op case. Return a
11777 tri-state indicating that moving the stmt before the call may allow
11778 to continue. Pass through to_move.
11779 (find_tail_calls): Handle moving unrelated defs before
11782 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
11785 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
11786 splitter result in the canonical way.
11788 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
11790 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
11791 also for 32bit target. Update insn attributes.
11792 (zero-extendsidi2 splitter): Allow all registers for operand 1.
11794 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
11796 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
11797 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
11798 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
11799 (_mm_maskz_min_ss): New intrinsics.
11801 2017-05-31 Martin Liska <mliska@suse.cz>
11803 * tree-vect-loop.c (vect_create_epilog_for_reduction):
11804 Change comment style to one we normally use.
11805 (vectorizable_reduction): Likewise.
11806 (vectorizable_induction): Likewise.
11807 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11808 (vectorizable_call): Likewise.
11809 (vectorizable_simd_clone_call): Likewise.
11810 (vectorizable_conversion): Likewise.
11811 (vectorizable_assignment): Likewise.
11812 (vectorizable_shift): Likewise.
11813 (vectorizable_operation): Likewise.
11814 (vectorizable_store): Likewise.
11815 (vectorizable_load): Likewise.
11816 * tree-vectorizer.h: Likewise.
11818 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
11820 * passes.c (emergency_dump_function): New.
11821 * tree-pass.h (emergency_dump_function): Declare.
11822 * plugin.c (plugins_internal_error_function): Remove.
11823 * plugin.h (plugins_internal_error_function): Remove declaration.
11824 * toplev.c (internal_error_function): New static function. Use it...
11825 (general_init): ...here.
11827 2017-05-31 Graham Markall <graham.markall@embecosm.com>
11829 * config/arc/arc.c (arc_print_operand): Handle constant operands.
11830 (arc_rtx_costs): Add costs for new patterns.
11831 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
11832 * config/arc/predicates.md: Add _1_2_3_operand predicate.
11834 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
11836 * tree-ssa-strlen.c (get_next_strinfo): New function.
11837 (get_stridx_plus_constant): Use it.
11838 (zero_length_string): Likewise.
11839 (adjust_related_strinfos): Likewise.
11840 (adjust_last_stmt): Likewise.
11842 2017-05-31 Richard Biener <rguenther@suse.de>
11845 * config/i386/i386.c (ix86_expand_builtin): Remove assert
11846 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
11848 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
11850 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
11851 loop_vinfo argument and use of dependence distance vectors.
11852 Check instead whether the two references differ only in their
11853 initial value and assume that they have the same alignment if the
11854 difference is a multiple of the vector alignment.
11855 (vect_analyze_data_refs_alignment): Update call accordingly.
11857 2017-05-31 Martin Liska <mliska@suse.cz>
11860 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
11862 2017-05-31 Bin Cheng <bin.cheng@arm.com>
11864 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
11865 (create_intersect_range_checks): Move from ...
11866 * tree-data-ref.c (create_intersect_range_checks_index)
11867 (create_intersect_range_checks): ... to here.
11868 (create_runtime_alias_checks): New function factored from ...
11869 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
11870 here. Call above function.
11871 * tree-data-ref.h (create_runtime_alias_checks): New function.
11873 2017-05-31 Bin Cheng <bin.cheng@arm.com>
11875 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
11876 segment length for dr_b and compute it in wide_int.
11878 2017-05-31 Richard Biener <rguenther@suse.de>
11880 PR tree-optimization/80906
11881 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
11882 and pass through iv_map.
11883 (copy_bb_and_scalar_dependences): Adjust.
11884 (translate_pending_phi_nodes): Likewise.
11885 (copy_loop_close_phi_args): Handle code-generating IVs instead
11888 2017-05-30 David Malcolm <dmalcolm@redhat.com>
11890 * diagnostic-color.c (color_dict): Add "type-diff".
11891 (parse_gcc_colors): Update comment.
11892 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
11893 -fdiagnostics-show-template-tree and -fno-elide-type.
11894 (GCC_COLORS): Add type-diff to example.
11896 (-fdiagnostics-show-template-tree): New.
11897 (-fno-elide-type): New.
11898 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
11899 the pp_format_decoder callback. Call any m_format_postprocessor's
11901 (pretty_printer::pretty_printer): Initialize
11902 m_format_postprocessor.
11903 (pretty_printer::~pretty_printer): Delete any
11904 m_format_postprocessor.
11905 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
11906 (class format_postprocessor): New class.
11907 (struct pretty_printer::format_decoder): Document the new parameters.
11908 (struct pretty_printer::m_format_postprocessor): New field.
11909 * tree-diagnostic.c (default_tree_printer): Update for new
11910 bool and const char ** params.
11911 * tree-diagnostic.h (default_tree_printer): Likewise.
11913 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
11915 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
11916 (lwa_operand): Delete rs6000_gen_cell_microcode test.
11917 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
11918 rs6000_gen_cell_microcode code.
11919 (rs6000_final_prescan_insn): Delete.
11920 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
11921 "warn-cell-microcode" entries.
11922 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
11923 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
11924 throughout. Change cc_reg_not_micro_cr0_operand to
11925 cc_reg_not_cr0_operand throughout.
11926 (*extendhi<mode>2_noload): Delete.
11927 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
11928 (mwarn-cell-microcode): Delete.
11929 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
11930 -mgen-cell-microcode and -mwarn-cell-microcode.
11932 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
11935 * config/i386/constraints.md (Yd): New constraint.
11937 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
11938 and (?Yd, r) alternatives. Update insn attributes.
11939 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
11940 and (?*Yd, r) alternatives. Update insn attributes.
11941 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
11943 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
11945 * gimplify.c (gimplify_modify_expr): Don't create a
11946 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
11949 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
11951 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
11953 2017-05-30 Richard Biener <rguenther@suse.de>
11955 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
11956 and reduc_def fields.
11957 (STMT_VINFO_REDUC_TYPE): New define.
11958 (STMT_VINFO_REDUC_DEF): Likewise.
11959 (vect_force_simple_reduction): Adjust prototype.
11960 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
11961 (vect_is_simple_reduction): Remove check_reduction argument.
11962 (vect_force_simple_reduction): Adjust and set
11963 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
11964 (vectorizable_reduction): Do not re-do reduction analysis
11965 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
11966 * tree-parloops.c (gather_scalar_reductions): Adjust.
11968 2017-05-30 Richard Biener <rguenther@suse.de>
11970 PR middle-end/80901
11971 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
11974 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
11976 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
11977 Introduce unknown_misalignment parameter and remove vf.
11978 (vect_peeling_hash_get_lowest_cost):
11979 Pass unknown_misalignment parameter.
11980 (vect_enhance_data_refs_alignment):
11981 Fix unsupportable data ref treatment.
11983 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11985 * tree-vect-data-refs.c (vect_get_data_access_cost):
11986 Workaround for SLP handling.
11987 (vect_enhance_data_refs_alignment):
11988 Compute costs for doing no peeling at all, compare to the best
11989 peeling costs so far and avoid peeling if cheaper.
11991 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11993 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
11994 Return peeling info and set costs to zero for unlimited cost
11996 (vect_enhance_data_refs_alignment): Also inspect all datarefs
11997 with unknown misalignment. Compute and costs for unknown
11998 misalignment, compare them to the costs for known misalignment
11999 and choose the cheapest for peeling.
12001 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
12003 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
12004 (vect_get_peeling_costs_all_drs): Create function.
12005 (vect_peeling_hash_get_lowest_cost):
12006 Use vect_get_peeling_costs_all_drs.
12007 (vect_peeling_supportable): Create function.
12008 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
12010 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
12012 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
12013 DR_HAS_NEGATIVE_STEP.
12014 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
12015 (vect_enhance_data_refs_alignment): Use.
12016 (vect_duplicate_ssa_name_ptr_info): Use.
12017 * tree-vectorizer.h (dr_misalignment): Use.
12018 (known_alignment_for_access_p): Use.
12020 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12023 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
12025 (has_section_name): New function.
12027 2017-05-30 Martin Liska <mliska@suse.cz>
12030 * auto-profile.c (get_function_decl_from_block): Fix
12033 2017-05-30 Richard Biener <rguenther@suse.de>
12035 PR middle-end/80876
12036 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
12038 2017-05-30 Martin Liska <mliska@suse.cz>
12040 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
12041 * dumpfile.h (struct dump_file_info): Remove ctors.
12043 2017-05-30 Martin Liska <mliska@suse.cz>
12045 * predict.def: Fix GNU coding style.
12047 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
12049 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
12050 Mark 'to' argument with ATTRIBUTE_UNUSED.
12052 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
12054 * config/xtensa/xtensa.c (xtensa_emit_call): Use
12055 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
12056 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
12059 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
12061 * doc/install.texi (Options specification): Restore entry of
12062 --enable-sjlj-exceptions.
12064 2017-05-27 Michael Eager <eager@eagercon.com>
12067 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
12069 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
12071 * config/microblaze/microblaze.h
12072 (FIXED_REGISTERS): Update in macro.
12073 (CALL_USED_REGISTERS): Update in macro.
12075 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
12077 * doc/install.texi: Add links to macOS binary distributions.
12079 2017-05-27 Jakub Jelinek <jakub@redhat.com>
12083 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
12085 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
12087 2017-05-26 Martin Liska <mliska@suse.cz>
12089 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
12091 2017-05-26 Martin Liska <mliska@suse.cz>
12093 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
12094 always leading ';; '.
12095 (dump_bb_info): Likewise.
12096 (brief_dump_cfg): Likewise.
12097 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
12098 * dumpfile.c: Remove usage of TDF_VERBOSE.
12099 * dumpfile.h (enum dump_kind): Likewise.
12100 (dump_gimple_bb_header): Do not use TDF_COMMENT.
12101 * print-tree.c (debug_verbose): Remove.
12102 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
12103 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
12104 * tree-diagnostic.c (default_tree_printer): Replace
12105 TDF_DIAGNOSTIC with TDF_SLIM.
12107 2017-05-26 Bin Cheng <bin.cheng@arm.com>
12109 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
12110 in parameter loop, rather than loop_vinfo.
12111 (create_intersect_range_checks): Ditto.
12112 (vect_create_cond_for_alias_checks): Update call to above functions.
12114 2017-05-26 Bin Cheng <bin.cheng@arm.com>
12116 PR tree-optimization/80815
12117 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
12118 for merging runtime alias checks. Handle negative DR_STEPs.
12120 2017-05-26 Bin Cheng <bin.cheng@arm.com>
12122 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
12124 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
12125 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
12126 out code pruning runtime alias checks.
12127 * tree-data-ref.c (prune_runtime_alias_test_list): New function
12128 factored out from above.
12129 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
12131 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
12133 (prune_runtime_alias_test_list): New decalaration.
12135 2017-05-26 Bin Cheng <bin.cheng@arm.com>
12137 * tree-vect-data-refs.c (compare_tree): Rename and move ...
12138 * tree-data-ref.c (data_ref_compare_tree): ... to here.
12139 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
12140 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
12141 (operator==, comp_dr_with_seg_len_pair): Ditto.
12142 (vect_prune_runtime_alias_test_list): Ditto.
12144 2017-05-26 Martin Liska <mliska@suse.cz>
12147 * params.def: Bound partial-inlining-entry-probability param.
12149 2017-05-26 Marek Polacek <polacek@redhat.com>
12152 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
12155 2017-05-26 Richard Biener <rguenther@suse.de>
12157 PR tree-optimization/80842
12158 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
12161 2017-05-26 Richard Biener <rguenther@suse.de>
12163 PR tree-optimization/80844
12164 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
12166 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
12168 * doc/md.texi (Machine Constraints): Update x86 family
12169 machine constraints section to match 'config/i386/constraints.md'.
12171 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
12173 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
12175 2017-05-25 Nathan Sidwell <nathan@acm.org>
12177 * doc/invoke.texi (--enable-languages): Update documentation.
12179 2017-05-25 Martin Liska <mliska@suse.cz>
12181 * dumpfile.c: Add TDF_FOLDING.
12182 * dumpfile.h (enum dump_kind): Likewise.
12183 * genmatch.c (dt_simplify::gen_1): Use it.
12185 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
12187 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
12189 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
12191 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
12192 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
12194 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
12196 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
12197 * match.pd (X == C): Rewrite it here.
12198 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
12199 with_certain_nonzero_bits2): New predicates.
12200 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
12202 2017-05-24 Nathan Sidwell <nathan@acm.org>
12204 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
12207 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
12210 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
12212 * config/powerpcspe: New port. Files are copied from the rs6000
12213 port, with "rs6000" in filenames replaced by "powerpcspe".
12215 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
12217 PR rtl-optimization/80754
12218 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
12220 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
12222 * config/sparc/sparc.md (length): Return the correct value for -mflat
12223 sibcalls to match output_sibcall.
12225 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
12229 * config/rs6000/rs6000.c (struct machine_function): Add new field
12231 (rs6000_get_separate_components): Init that field, use it.
12232 (rs6000_components_for_bb): Use the field.
12234 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
12236 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
12238 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
12240 PR middle-end/80823
12241 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
12243 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12246 * config/s390/s390.c (s390_check_qrst_address): Check incoming
12247 address against address_operand predicate.
12248 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
12250 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
12252 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
12253 parameters passed indirectly.
12255 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
12257 * config/i386/i386.md (*movdi_internal): Remove SSE4
12258 alternative 18 (?r, *v). Update insn attributes.
12259 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
12260 Update insn attributes.
12261 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
12262 Update insn attributes.
12263 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
12264 alternative 1 (r, v). Remove isa attribute.
12265 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
12266 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
12267 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
12269 2017-05-23 Tom de Vries <tom@codesourcery.com>
12271 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
12274 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12276 * cgraphunit.c (symbol_table::process_new_functions): Update.
12277 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
12278 (inline_generate_summary): Rename to ...
12279 (ipa_fn_summary_generate): ... this one.
12280 (inline_read_summary): Rename to ...
12281 (ipa_fn_summary_read): ... this one.
12282 (inline_write_summary): Rename to ...
12283 (ipa_fn_summary_write): ... this one.
12284 (inline_free_summary): Rename to ...
12285 (ipa_free_fn_summary): ... this one.
12286 (pass_data_local_fn_summary, pass_local_fn_summary,
12287 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
12288 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
12289 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
12290 make_pass_ipa_fn_summary): New.
12291 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
12292 inline_write_summary, inline_free_summary): Remove.
12293 (ipa_free_fn_summary) : New.
12294 * ipa-inline.c (ipa_inline): Update.
12295 (pass_ipa_inline): Do not generate summaries.
12296 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
12298 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
12299 and add pass_ipa_fn_summary.
12300 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
12302 (make_pass_inline_parameters): Remove.
12304 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
12306 * omp-low.c (struct omp_context): Remove "default_kind" member.
12309 * omp-offload.c (execute_oacc_device_lower): Remove the
12310 parallelism dimensions function attributes for unparallelized
12311 OpenACC kernels constructs.
12313 2017-05-23 Martin Liska <mliska@suse.cz>
12315 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
12317 (cgraph_edge::make_speculative): Likewise.
12318 (cgraph_edge::resolve_speculation): Likewise.
12319 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
12320 (cgraph_node::dump): Likewise.
12321 * cgraph.h: Likewise.
12322 * cgraphunit.c (analyze_functions): Likewise.
12323 (symbol_table::compile): Likewise.
12324 * ipa-cp.c (print_all_lattices): Likewise.
12325 (determine_versionability): Likewise.
12326 (initialize_node_lattices): Likewise.
12327 (ipcp_verify_propagated_values): Likewise.
12328 (estimate_local_effects): Likewise.
12329 (update_profiling_info): Likewise.
12330 (create_specialized_node): Likewise.
12331 (perhaps_add_new_callers): Likewise.
12332 (decide_about_value): Likewise.
12333 (decide_whether_version_node): Likewise.
12334 (identify_dead_nodes): Likewise.
12335 (ipcp_store_bits_results): Likewise.
12336 * ipa-devirt.c (dump_targets): Likewise.
12337 (ipa_devirt): Likewise.
12338 * ipa-icf.c (sem_item::dump): Likewise.
12339 (sem_function::equals): Likewise.
12340 (sem_variable::equals): Likewise.
12341 (sem_item_optimizer::read_section): Likewise.
12342 (sem_item_optimizer::execute): Likewise.
12343 (congruence_class::dump): Likewise.
12344 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
12345 (dump_inline_summary): Likewise.
12346 (estimate_node_size_and_time): Likewise.
12347 (inline_analyze_function): Likewise.
12348 * ipa-inline-transform.c (inline_call): Likewise.
12349 * ipa-inline.c (report_inline_failed_reason): Likewise.
12350 (want_early_inline_function_p): Likewise.
12351 (edge_badness): Likewise.
12352 (update_edge_key): Likewise.
12353 (inline_small_functions): Likewise.
12354 * ipa-profile.c (ipa_profile): Likewise.
12355 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
12356 (ipa_make_edge_direct_to_target): Likewise.
12357 (remove_described_reference): Likewise.
12358 (ipa_impossible_devirt_target): Likewise.
12359 (propagate_controlled_uses): Likewise.
12360 (ipa_print_node_params): Likewise.
12361 (ipcp_transform_function): Likewise.
12362 * ipa-pure-const.c (pure_const_read_summary): Likewise.
12363 (propagate_pure_const): Likewise.
12364 * ipa-reference.c (generate_summary): Likewise.
12365 (read_write_all_from_decl): Likewise.
12366 (propagate): Likewise.
12367 (ipa_reference_read_optimization_summary): Likewise.
12368 * ipa-utils.c (ipa_merge_profiles): Likewise.
12369 * ipa.c (walk_polymorphic_call_targets): Likewise.
12370 (symbol_table::remove_unreachable_nodes): Likewise.
12371 (ipa_single_use): Likewise.
12372 * passes.c (execute_todo): Likewise.
12373 * predict.c (drop_profile): Likewise.
12374 * symtab.c (symtab_node::get_dump_name): New function.
12375 (symtab_node::dump_name): Likewise.
12376 (symtab_node::dump_asm_name): Likewise.
12377 (symtab_node::dump_references): Likewise.
12378 (symtab_node::dump_referring): Likewise.
12379 (symtab_node::dump_base): Likewise.
12380 (symtab_node::debug_symtab): Likewise.
12381 * tree-sra.c (convert_callers_for_node): Likewise.
12382 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
12383 * value-prof.c (init_node_map): Likewise.
12385 2017-05-23 Martin Liska <mliska@suse.cz>
12387 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
12388 and symtab_node::debug_symtab to symbol_table::debug.
12389 * cgraphunit.c (analyze_functions): Use the renamed function.
12390 (symbol_table::compile): Likewise.
12391 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
12392 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
12393 * passes.c (execute_todo): Likewise.
12394 * symtab.c (symbol_table::dump): New function.
12395 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
12397 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12399 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
12400 that nonconst implies exec.
12402 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12404 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
12405 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
12406 (inline_edge_summary_vec): Turn into ...
12407 (ipa_call_summaries): ... this one.
12408 (redirect_to_unreachable, edge_set_predicate,
12409 evaluate_properties_for_edge, inline_summary_alloc,
12410 reset_ipa_call_summary, reset_inline_summary,
12411 inline_summary_t::duplicate): Update.
12412 (inline_edge_duplication_hook): Turn to ...
12413 (ipa_call_summary_t::duplicate): ... this one.
12414 (inline_edge_removal_hook): Turn to ...
12415 (ipa_call_summary_t::remove): ... this one.
12416 (dump_inline_edge_summary): Turn to ...
12417 (dump_ipa_call_summary): ... this one.
12418 (estimate_function_body_sizes): Update.
12419 (inline_update_callee_summaries): Update.
12420 (remap_edge_change_prob): Update.
12421 (remap_edge_summaries): Update.
12422 (inline_merge_summary): Update.
12423 (do_estimate_edge_time): Update.
12424 (inline_generate_summary): Update.
12425 (inline_read_section): Update.
12426 (inline_read_summary): Update.
12427 (inline_free_summary): Update.
12428 * ipa-inline.c (can_inline_edge_p): Update.
12429 (compute_inlined_call_time): Update.
12430 (want_inline_small_function_p): Update.
12431 (edge_badness): Update.
12432 (early_inliner): Update.
12433 * ipa-inline.h (inline_edge_summary): Turn to ...
12434 (ipa_call_summary): ... this one.
12435 (ipa_call_summary_t): New class.
12436 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
12437 (ipa_call_summaries): New.
12438 (inline_edge_summary): Remove.
12439 (estimate_edge_growth): Update.
12440 * ipa-profile.c (ipa_propagate_frequency_1): Update.
12441 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
12442 * ipa-split.c (execute_split_functions): Update.
12443 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
12445 2017-05-23 Tom de Vries <tom@codesourcery.com>
12447 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
12448 attributes): Document rdrand effective target.
12450 2017-05-23 Tom de Vries <tom@codesourcery.com>
12452 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
12453 attributes): Sort alphabetically.
12455 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
12457 * config/avr/genmultilib.awk: Use gsub instead of gensub.
12459 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
12462 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
12463 V2DF/V2DI splat into two separate patterns, one that handles
12464 registers, and the other that only handles memory. Drop support
12465 for splatting from a GPR on ISA 2.07 and then splitting the
12466 splat into direct move and splat.
12467 (vsx_splat_<mode>_reg): Likewise.
12468 (vsx_splat_<mode>_mem): Likewise.
12470 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
12472 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
12474 2017-05-22 Jakub Jelinek <jakub@redhat.com>
12476 PR middle-end/80809
12477 * omp-low.c (finish_taskreg_remap): New function.
12478 (finish_taskreg_scan): If unit size of ctx->record_type
12479 is non-constant, unshare the size expression and replace
12480 decls in it with possible outer var refs.
12482 PR middle-end/80809
12483 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
12484 GOVD_SHARED rather than GOVD_PRIVATE with it.
12485 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
12486 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
12488 PR middle-end/80853
12489 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
12490 as last argument to build_outer_var_ref for pointer bases of array
12491 section reductions.
12493 2017-05-19 Martin Sebor <msebor@redhat.com>
12495 * print-tree.c (print_node): Print DECL_READ_P flag.
12497 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12499 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
12500 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
12501 * cgraph.c: Likewise.
12502 * cgraphunit.c: Likewise.
12503 * gengtype.c: Likewise.
12504 * ipa-cp.c: Likewise.
12505 * ipa-devirt.c: Likewise.
12506 * ipa-icf.c: Likewise.
12507 * ipa-predicate.c: Likewise.
12508 * ipa-profile.c: Likewise.
12509 * ipa-prop.c: Likewise.
12510 * ipa-split.c: Likewise.
12512 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
12513 edge_predicate_pool, dump_inline_hints,
12514 inline_summary::account_size_time, redirect_to_unreachable,
12515 edge_set_predicate, set_hint_predicate,
12516 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
12517 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
12518 inline_summary_t::remove, remap_hint_predicate_after_duplication,
12519 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
12520 ipa_call_summary_t::remove, initialize_growth_caches,
12521 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
12522 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
12523 mark_modified, unmodified_parm_1, unmodified_parm,
12524 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
12525 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
12526 compute_bb_predicates, will_be_nonconstant_expr_predicate,
12527 will_be_nonconstant_predicate, record_modified_bb_info,
12528 get_minimal_bb, record_modified, param_change_prob,
12529 phi_result_unknown_predicate, predicate_for_phi_result,
12530 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
12531 estimate_function_body_sizes, compute_inline_parameters,
12532 compute_inline_parameters_for_curren, pass_data_inline_parameters,
12533 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
12534 inline_update_callee_summaries, remap_edge_change_prob,
12535 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
12536 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
12537 inline_analyze_function, inline_summary_t::insert,
12538 inline_generate_summary, read_ipa_call_summary, inline_read_section,
12539 inline_read_summary, write_ipa_call_summary, inline_write_summary,
12540 inline_free_summary): Move to ipa-fnsummary.h
12541 (predicate_t): Remove.
12542 * ipa-fnsummary.c: New file.
12543 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
12544 (enum inline_hints_vals, inline_hints, agg_position_info,
12545 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
12546 inline_summaries, ipa_call_summary, ipa_call_summary_t,
12547 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
12548 dump_inline_summary, dump_inline_hints, inline_generate_summary,
12549 inline_read_summary, inline_write_summary, inline_free_summary,
12550 inline_analyze_function, initialize_inline_failed,
12551 inline_merge_summary, inline_update_overall_summary,
12552 compute_inline_parameters): Move to ipa-fnsummary.h
12553 * ipa-fnsummary.h: New file.
12554 * ipa-inline-transform.h: Include ipa-inline.h.
12555 * ipa-inline.c: LIkewise.
12557 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12559 * ipa-inline.c (edge_badness): Use inlined_time instead of
12560 inline_summaries->get.
12562 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12564 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
12566 2017-05-22 Nathan Sidwell <nathan@acm.org>
12568 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
12569 (fdump-lang): Document 'raw' option.
12570 * dumpfile.h (TDI_tu): Delete.
12571 * dumpfile.c (dump_files): Remove translation-unit.
12572 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
12574 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
12576 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
12577 command option from $(AWK) call.
12578 * config/avr/genmultilib.awk: Simplify and rewrite so that it
12579 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
12580 [FORMAT]: Remove handling of variable.
12581 * config/avr/t-multilib: Regenerate.
12583 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12585 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
12587 (dump_inline_summary): Do not print self_time.
12588 (estimate_function_body_sizes): Do not set self_time.
12589 (compute_inline_parameters): Likewise.
12590 (inline_read_section, inline_write_summary): Do not stream self_time.
12591 * ipa-inline.h (inline_summary): Drop self_time.
12593 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12595 * ipa-inline-analysis.c (account_size_time): Rename to ...
12596 (inline_summary::account_size_time): ... this one.
12597 (reset_ipa_call_summary): Turn to ...
12598 (ipa_call_summary::reset): ... this one.
12599 (reset_inline_summary): Turn to ...
12600 (inline_summary::reset): ... this one.
12601 (inline_summary_t::remove): Update.
12602 (inline_summary_t::duplicate): Update.
12603 (ipa_call_summary_t::remove): Update.
12604 (dump_inline_summary): Update.
12605 (estimate_function_body_sizes): Update.
12606 (compute_inline_parameters): Update.
12607 (estimate_node_size_and_time): Update.
12608 (inline_merge_summary): Update.
12609 (inline_update_overall_summary): Update.
12610 (inline_read_section): Update.
12611 (inline_write_summary): Update.
12612 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
12613 add account_size_time and reset member functions.
12614 (ipa_call_summary): Add reset function.
12615 * ipa-predicate.h (predicate::operator &): Constify.
12617 2017-05-22 Richard Biener <rguenther@suse.de>
12619 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
12621 2017-05-19 Jason Merrill <jason@redhat.com>
12623 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
12625 2017-05-19 Marek Polacek <polacek@redhat.com>
12628 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
12629 TYPE_OVERFLOW_WRAPS checks.
12631 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
12633 * tree-core.h (enum omp_clause_default_kind): Add
12634 "OMP_CLAUSE_DEFAULT_PRESENT".
12635 * tree-pretty-print.c (dump_omp_clause): Handle it.
12636 * gimplify.c (enum gimplify_omp_var_data): Add
12637 "GOVD_MAP_FORCE_PRESENT".
12638 (gimplify_adjust_omp_clauses_1): Map it to
12639 "GOMP_MAP_FORCE_PRESENT".
12640 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
12642 * gimplify.c (oacc_default_clause): Clarify.
12644 2017-05-19 Nathan Sidwell <nathan@acm.org>
12646 LANG_HOOK_REGISTER_DUMPS
12647 * toplev.c (general_init): Call register dump lang hook.
12648 * doc/invoke.texi: Document -fdump-lang option family.
12649 * dumpfile.c (dump_files): Remove class dump here.
12650 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
12651 * dumpfile.h (tree_dump_index): Remove TDI_class.
12652 * langhooks-def.h (lhd_register_dumps): Declare.
12653 (LANG_HOOKS_REGISTER_DUMPS): Define.
12654 (LANG_HOOKS_INITIALIZER): Add it.
12655 * langhooks.c (lhd_register_dumps): Define.
12656 * langhooks.h (struct lang_hooks): Add register_dumps.
12658 2017-05-19 Nathan Sidwell <nathan@acm.org>
12660 * context.h (context::set_passes): New.
12661 * context.c (context::context): Do not create pass manager.
12662 * toplev.c (general_init): Create pass manager here.
12664 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
12666 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
12667 use this splitter if two add or or instructions would also work for
12668 the constant we want to generate.
12670 2017-05-19 Richard Biener <rguenther@suse.de>
12673 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
12674 predicate evaluation.
12676 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
12678 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
12680 * ipa-inline.c (want_inline_small_function_p): Do not cast to
12683 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
12685 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
12686 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
12687 (inline_edge_summary_vec): Turn into ...
12688 (ipa_call_summaries): ... this one.
12689 (redirect_to_unreachable, edge_set_predicate,
12690 evaluate_properties_for_edge, inline_summary_alloc,
12691 reset_ipa_call_summary, reset_inline_summary,
12692 inline_summary_t::duplicate): Update.
12693 (inline_edge_duplication_hook): Turn to ...
12694 (ipa_call_summary_t::duplicate): ... this one.
12695 (inline_edge_removal_hook): Turn to ...
12696 (ipa_call_summary_t::remove): ... this one.
12697 (dump_inline_edge_summary): Turn to ...
12698 (dump_ipa_call_summary): ... this one.
12699 (estimate_function_body_sizes): Update.
12700 (inline_update_callee_summaries): Update.
12701 (remap_edge_change_prob): Update.
12702 (remap_edge_summaries): Update.
12703 (inline_merge_summary): Update.
12704 (do_estimate_edge_time): Update.
12705 (inline_generate_summary): Update.
12706 (inline_read_section): Update.
12707 (inline_read_summary): Update.
12708 (inline_free_summary): Update.
12709 * ipa-inline.c (can_inline_edge_p): Update.
12710 (compute_inlined_call_time): Update.
12711 (want_inline_small_function_p): Update.
12712 (edge_badness): Update.
12713 (early_inliner): Update.
12714 * ipa-inline.h (inline_edge_summary): Turn to ...
12715 (ipa_call_summary): ... this one.
12716 (ipa_call_summary_t): New class.
12717 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
12718 (ipa_call_summaries): New.
12719 (inline_edge_summary): Remove.
12720 (estimate_edge_growth): Update.
12721 * ipa-profile.c (ipa_propagate_frequency_1): Update.
12722 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
12723 * ipa-split.c (execute_split_functions): Update.
12724 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
12726 2017-05-19 Richard Biener <rguenther@suse.de>
12728 PR middle-end/80764
12729 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
12731 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
12733 * config/rs6000/rs6000.c (struct machine_function): Add field
12734 fpr_is_wrapped_separately.
12735 (rs6000_get_separate_components): Use 64 components. Handle the
12736 new FPR components.
12737 (rs6000_components_for_bb): Handle the FPR components.
12738 (rs6000_emit_prologue_components): Handle the FPR components.
12739 (rs6000_emit_epilogue_components): Handle the FPR components.
12740 (rs6000_set_handled_components): Handle the FPR components.
12741 (rs6000_emit_prologue): Don't output prologue code for those FPRs
12742 that are already separately shrink-wrapped.
12743 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
12744 that are already separately shrink-wrapped.
12746 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
12749 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
12752 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
12753 (define_peephole2 for Altivec d-form load): Add peepholes to catch
12754 cases where the register allocator uses a move and an offsettable
12755 memory operation to/from a FPR register on ISA 2.06/2.07.
12756 (define_peephole2 for Altivec d-form store): Likewise.
12758 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
12761 * config/i386/mmx.md (*mov<mode>_internal): Enable
12762 alternatives 11, 12, 13 and 14 also for 32bit targets.
12763 Remove alternatives 15, 16, 17 and 18.
12764 * config/i386/sse.md (vec_concatv2di): Change
12765 alternative (!x, *y) to (x, ?!*Yn).
12767 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
12769 * dumpfile.h (enum dump_kind): Remove stray comma.
12771 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
12773 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
12774 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
12775 predicate::num_conditions
12776 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
12777 (CHANGED): turn into predicate::changed.
12778 (agg_position_info): Move to ipa-predicate.h
12779 (add_condition, predicate::add_clause, predicate::operator &=,
12780 predicate::or_with, predicate::evaluate, predicate::probability,
12781 dump_condition, dump_clause, predicate::dump,
12782 predicate::remap_after_duplication, predicate::remap_after_inlining,
12783 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
12784 (evaluate_conditions_for_known_args): Update.
12785 (set_cond_stmt_execution_predicate): Update.
12786 * ipa-inline.h: Include ipa-predicate.h
12787 (condition, inline_param_summary, conditions, agg_position_info,
12788 predicate): Move to ipa-predicate.h
12789 * ipa-predicate.c: New file.
12790 * ipa-predicate.h: New file.
12792 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
12794 * final.c (leaf_function_p): Check we are not in a sequence.
12796 2017-05-18 Martin Liska <mliska@suse.cz>
12798 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
12799 * dumpfile.c (dump_register): Use new enum dump_kind.
12800 (get_dump_file_name): Likewise.
12801 (dump_enable_all): Likewise.
12802 (dump_switch_p_1): Likewise.
12803 (enable_rtl_dump_file): Remove usage of TDF_RTL.
12804 * dumpfile.h (enum dump_kind): New enum type.
12805 (struct dump_file_info): Create constructor and
12806 format fields and comments.
12807 * passes.c (pass_manager::register_one_dump_file):
12809 * statistics.c (statistics_early_init): Likewise.
12810 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
12811 TDF_TREE with TDF_SLIM.
12812 (gather_memory_references_ref): Likewise.
12814 2017-05-18 Martin Liska <mliska@suse.cz>
12816 * vec.h (struct vnull): Use it.
12818 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
12820 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
12821 (true_predicate, false_predicate, true_predicate_p,
12822 false_predicate_p): Remove.
12823 (single_cond_predicate, not_inlined_predicate): Turn to member function
12825 (add_condition): Update.
12826 (add_clause): Turn to...
12827 (predicate::add_clause): ... this one; update; allow passing NULL
12829 (and_predicates): Turn to ...
12830 (predicate::operator &=): ... this one.
12831 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
12832 (or_predicates): Turn to ...
12833 (predicate::or_with): ... this one.
12834 (evaluate_predicate): Turn to ...
12835 (predicate::evaluate): ... this one.
12836 (predicate_probability): Turn to ...
12837 (predicate::probability): ... this one.
12838 (dump_condition): Update.
12839 (dump_predicate): Turn to ...
12840 (predicate::dump): ... this one.
12841 (account_size_time): Update.
12842 (edge_set_predicate): Update.
12843 (set_hint_predicate): UPdate.
12844 (evaluate_conditions_for_known_args): Update.
12845 (evaluate_properties_for_edge): Update.
12846 (remap_predicate_after_duplication): Turn to...
12847 (predicate::remap_after_duplication): ... this one.
12848 (remap_hint_predicate_after_duplication): Update.
12849 (inline_summary_t::duplicate): UPdate.
12850 (dump_inline_edge_summary): Update.
12851 (dump_inline_summary): Update.
12852 (set_cond_stmt_execution_predicate): Update.
12853 (set_switch_stmt_execution_predicate): Update.
12854 (compute_bb_predicates): Update.
12855 (will_be_nonconstant_expr_predicate): Update.
12856 (will_be_nonconstant_predicate): Update.
12857 (phi_result_unknown_predicate): Update.
12858 (predicate_for_phi_result): Update.
12859 (array_index_predicate): Update.
12860 (estimate_function_body_sizes): Update.
12861 (estimate_node_size_and_time): Update.
12862 (estimate_ipcp_clone_size_and_time): Update.
12863 (remap_predicate): Rename to ...
12864 (predicate::remap_after_inlining): ... this one.
12865 (remap_hint_predicate): Update.
12866 (inline_merge_summary): Update.
12867 (inline_update_overall_summary): Update.
12868 (estimate_size_after_inlining): Update.
12869 (read_predicate): Rename to ...
12870 (predicate::stream_in): ... this one.
12871 (read_inline_edge_summary): Update.
12872 (write_predicate): Rename to ...
12873 (predicate::stream_out): ... this one.
12874 (write_inline_edge_summary): Update.
12875 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
12876 (clause_t): Turn to uint32_t
12877 (predicate): Turn to class; implement constructor and operators
12879 (size_time_entry): Update.
12880 (inline_summary): Update.
12881 (inline_edge_summary): Update.
12883 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
12885 * fold-const.c (fold_binary_loc): Move transformation...
12886 * match.pd (C - X CMP X): ... here.
12888 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
12890 * config/sparc/sparc.c (sparc_option_override): Set function
12891 alignment for -mcpu=niagara7 to 64 to match the I$ line.
12892 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
12894 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
12896 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
12898 2017-05-18 Marek Polacek <polacek@redhat.com>
12901 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
12902 (pass_ubsan::execute): Call gimple_assign_single_p instead of
12903 gimple_assign_load_p.
12905 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
12907 PR middle-end/80692
12908 * real.c (do_compare): Give decimal_do_compare preference over
12909 comparing just the signs.
12911 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
12913 * doc/md.texi (Canonicalization of Instructions): Describe the
12914 canonical form of instructions that inherently set a condition
12917 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
12919 PR middle-end/80775
12920 * tree-cfg.c: Move deletion of unreachable case statements to after
12921 the merging of consecutive case labels.
12923 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
12925 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
12926 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
12927 restoring of callee-saved registers.
12929 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
12931 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
12932 * config/visium/visium.c (single_set_and_flags): Likewise.
12933 * config/visium/visium.md (Substitutions): Likewise.
12935 2017-05-17 Martin Liska <mliska@suse.cz>
12937 * cfg.c: Introduce dump_flags_t type and
12938 use it instead of int type.
12940 * cfghooks.c: Likewise.
12941 * cfghooks.h (struct cfg_hooks): Likewise.
12942 * cfgrtl.c: Likewise.
12943 * cfgrtl.h: Likewise.
12944 * cgraph.c (cgraph_node::get_body): Likewise.
12945 * coretypes.h: Likewise.
12946 * domwalk.c: Likewise.
12947 * domwalk.h: Likewise.
12948 * dumpfile.c (struct dump_option_value_info): Likewise.
12949 (dump_enable_all): Likewise.
12950 (dump_switch_p_1): Likewise.
12951 (opt_info_switch_p): Likewise.
12952 * dumpfile.h (enum tree_dump_index): Likewise.
12953 (struct dump_file_info): Likewise.
12954 * genemit.c: Likewise.
12955 * generic-match-head.c: Likewise.
12956 * gengtype.c (open_base_files): Likewise.
12957 * gimple-pretty-print.c: Likewise.
12958 * gimple-pretty-print.h: Likewise.
12959 * graph.c (print_graph_cfg): Likewise.
12960 * graphite-scop-detection.c (dot_all_sese): Likewise.
12961 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
12962 * loop-unroll.c (report_unroll): Likewise.
12963 * passes.c (pass_manager::register_one_dump_file): Likewise.
12964 * print-tree.c: Likewise.
12965 * statistics.c: Likewise.
12966 * tree-cfg.c: Likewise.
12967 * tree-cfg.h: Likewise.
12968 * tree-dfa.c: Likewise.
12969 * tree-dfa.h: Likewise.
12970 * tree-dump.c (dump_function): Likewise.
12971 * tree-dump.h (struct dump_info): Likewise.
12972 * tree-pretty-print.c: Likewise.
12973 * tree-pretty-print.h: Likewise.
12974 * tree-ssa-live.c: Likewise.
12975 * tree-ssa-live.h: Likewise.
12976 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
12977 * tree-vect-loop.c: Likewise.
12978 * tree-vect-slp.c: Likewise.
12980 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
12981 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12983 PR tree-optimization/80457
12984 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
12985 of all arguments to a statement as scalar_to_vec operations.
12986 (vectorizable_call): Adjust call to vect_model_simple_cost for
12988 (vectorizable_conversion): Likewise.
12989 (vectorizable_assignment): Likewise.
12990 (vectorizable_shift): Likewise.
12991 (vectorizable_operation): Likewise.
12992 (vectorizable_comparison): Likewise.
12993 (vect_is_simple_cond): Record the def types for operands.
12994 (vectorizable_condition): Likewise, call vect_model_simple_cost.
12995 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
12996 for statement argument count.
12998 2017-05-16 Carl Love <cel@us.ibm.com>
13000 * config/rs6000/rs6000-c: Add support for built-in functions
13001 vector unsigned long long vec_bperm (vector unsigned long long,
13002 vector unsigned char)
13003 vector signed long long vec_mule (vector signed int,
13005 vector unsigned long long vec_mule (vector unsigned int,
13006 vector unsigned int)
13007 vector signed long long vec_mulo (vector signed int,
13009 vector unsigned long long vec_mulo (vector unsigned int,
13010 vector unsigned int)
13011 vector signed char vec_sldw (vector signed char,
13012 vector signed char,
13014 vector unsigned char vec_sldw (vector unsigned char,
13015 vector unsigned char,
13017 vector signed short vec_sldw (vector signed short,
13018 vector signed short,
13020 vector unsigned short vec_sldw (vector unsigned short,
13021 vector unsigned short,
13023 vector signed int vec_sldw (vector signed int,
13026 vector unsigned int vec_sldw (vector unsigned int,
13027 vector unsigned int,
13029 vector signed long long vec_sldw (vector signed long long,
13030 vector signed long long,
13032 vector unsigned long long vec_sldw (vector unsigned long long,
13033 vector unsigned long long,
13035 * config/rs6000/rs6000-c: Add support for built-in functions
13036 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
13037 * config/rs6000/altivec.h: Add defintion for vec_sldw.
13038 * doc/extend.texi: Update the built-in documentation for the
13039 new built-in functions.
13041 2017-05-16 Marek Polacek <polacek@redhat.com>
13045 * tree.c (save_expr): Don't fold the expression.
13047 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
13049 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
13050 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
13051 and (?*y,m). Update insn attributes.
13053 2017-05-16 Martin Liska <mliska@suse.cz>
13055 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
13056 flags argument of print_gimple_stmt, print_gimple_expr,
13057 print_generic_stmt and print_generic_expr.
13058 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
13059 * coretypes.h: Likewise.
13060 * except.c (dump_eh_tree): Likewise.
13061 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
13062 * gimple-pretty-print.h: Likewise.
13063 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
13064 (backprop::push_to_worklist): Likewise.
13065 (backprop::pop_from_worklist): Likewise.
13066 (backprop::process_use): Likewise.
13067 (backprop::intersect_uses): Likewise.
13068 (note_replacement): Likewise.
13069 * gimple-ssa-store-merging.c
13070 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
13071 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
13072 (pass_store_merging::execute): Likewise.
13073 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
13074 (ssa_base_cand_dump_callback): Likewise.
13075 (dump_incr_vec): Likewise.
13076 (replace_refs): Likewise.
13077 (replace_mult_candidate): Likewise.
13078 (create_add_on_incoming_edge): Likewise.
13079 (create_phi_basis): Likewise.
13080 (insert_initializers): Likewise.
13081 (all_phi_incrs_profitable): Likewise.
13082 (introduce_cast_before_cand): Likewise.
13083 (replace_one_candidate): Likewise.
13084 * gimplify.c (gimplify_expr): Likewise.
13085 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
13086 (set_rename): Likewise.
13087 (rename_uses): Likewise.
13088 (copy_loop_phi_nodes): Likewise.
13089 (add_close_phis_to_merge_points): Likewise.
13090 (copy_loop_close_phi_args): Likewise.
13091 (copy_cond_phi_args): Likewise.
13092 (graphite_copy_stmts_from_block): Likewise.
13093 (translate_pending_phi_nodes): Likewise.
13094 * graphite-poly.c (print_pdr): Likewise.
13095 (dump_gbb_cases): Likewise.
13096 (dump_gbb_conditions): Likewise.
13097 (print_scop_params): Likewise.
13098 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
13099 (build_cross_bb_scalars_use): Likewise.
13100 (gather_bbs::before_dom_children): Likewise.
13101 * hsa-dump.c (dump_hsa_immed): Likewise.
13102 * ipa-cp.c (print_ipcp_constant_value): Likewise.
13103 (get_replacement_map): Likewise.
13104 * ipa-inline-analysis.c (dump_condition): Likewise.
13105 (estimate_function_body_sizes): Likewise.
13106 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
13107 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
13108 * ipa-prop.c (ipa_dump_param): Likewise.
13109 (ipa_print_node_jump_functions_for_edge): Likewise.
13110 (ipa_modify_call_arguments): Likewise.
13111 (ipa_modify_expr): Likewise.
13112 (ipa_dump_param_adjustments): Likewise.
13113 (ipa_dump_agg_replacement_values): Likewise.
13114 (ipcp_modif_dom_walker::before_dom_children): Likewise.
13115 * ipa-pure-const.c (check_stmt): Likewise.
13116 (pass_nothrow::execute): Likewise.
13117 * ipa-split.c (execute_split_functions): Likewise.
13118 * omp-offload.c (dump_oacc_loop_part): Likewise.
13119 (dump_oacc_loop): Likewise.
13120 * trans-mem.c (tm_log_emit): Likewise.
13121 (tm_memopt_accumulate_memops): Likewise.
13122 (dump_tm_memopt_set): Likewise.
13123 (dump_tm_memopt_transform): Likewise.
13124 * tree-cfg.c (gimple_verify_flow_info): Likewise.
13125 (print_loop): Likewise.
13126 * tree-chkp-opt.c (chkp_print_addr): Likewise.
13127 (chkp_gather_checks_info): Likewise.
13128 (chkp_get_check_result): Likewise.
13129 (chkp_remove_check_if_pass): Likewise.
13130 (chkp_use_outer_bounds_if_possible): Likewise.
13131 (chkp_reduce_bounds_lifetime): Likewise.
13132 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
13133 (chkp_mark_completed_bounds): Likewise.
13134 (chkp_register_incomplete_bounds): Likewise.
13135 (chkp_mark_invalid_bounds): Likewise.
13136 (chkp_maybe_copy_and_register_bounds): Likewise.
13137 (chkp_build_returned_bound): Likewise.
13138 (chkp_get_bound_for_parm): Likewise.
13139 (chkp_build_bndldx): Likewise.
13140 (chkp_get_bounds_by_definition): Likewise.
13141 (chkp_generate_extern_var_bounds): Likewise.
13142 (chkp_get_bounds_for_decl_addr): Likewise.
13143 * tree-chrec.c (chrec_apply): Likewise.
13144 * tree-data-ref.c (dump_data_reference): Likewise.
13145 (dump_subscript): Likewise.
13146 (dump_data_dependence_relation): Likewise.
13147 (analyze_overlapping_iterations): Likewise.
13148 * tree-inline.c (expand_call_inline): Likewise.
13149 (tree_function_versioning): Likewise.
13150 * tree-into-ssa.c (dump_defs_stack): Likewise.
13151 (dump_currdefs): Likewise.
13152 (dump_names_replaced_by): Likewise.
13153 (dump_update_ssa): Likewise.
13154 (update_ssa): Likewise.
13155 * tree-object-size.c (pass_object_sizes::execute): Likewise.
13156 * tree-parloops.c (build_new_reduction): Likewise.
13157 (try_create_reduction_list): Likewise.
13158 (ref_conflicts_with_region): Likewise.
13159 (oacc_entry_exit_ok_1): Likewise.
13160 (oacc_entry_exit_single_gang): Likewise.
13161 * tree-pretty-print.h: Likewise.
13162 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
13163 (get_scalar_evolution): Likewise.
13164 (add_to_evolution): Likewise.
13165 (get_loop_exit_condition): Likewise.
13166 (analyze_evolution_in_loop): Likewise.
13167 (analyze_initial_condition): Likewise.
13168 (analyze_scalar_evolution): Likewise.
13169 (instantiate_scev): Likewise.
13170 (number_of_latch_executions): Likewise.
13171 (gather_chrec_stats): Likewise.
13172 (final_value_replacement_loop): Likewise.
13173 (scev_const_prop): Likewise.
13174 * tree-sra.c (dump_access): Likewise.
13175 (disqualify_candidate): Likewise.
13176 (create_access): Likewise.
13177 (reject): Likewise.
13178 (maybe_add_sra_candidate): Likewise.
13179 (create_access_replacement): Likewise.
13180 (analyze_access_subtree): Likewise.
13181 (analyze_all_variable_accesses): Likewise.
13182 (sra_modify_assign): Likewise.
13183 (initialize_constant_pool_replacements): Likewise.
13184 (find_param_candidates): Likewise.
13185 (decide_one_param_reduction): Likewise.
13186 (replace_removed_params_ssa_names): Likewise.
13187 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
13188 * tree-ssa-copy.c (dump_copy_of): Likewise.
13189 (copy_prop_visit_cond_stmt): Likewise.
13190 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
13191 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
13192 (record_equivalences_from_stmt): Likewise.
13193 * tree-ssa-dse.c (compute_trims): Likewise.
13194 (delete_dead_call): Likewise.
13195 (delete_dead_assignment): Likewise.
13196 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
13197 (forward_propagate_into_cond): Likewise.
13198 (pass_forwprop::execute): Likewise.
13199 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
13200 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
13202 (move_computations_worker): Likewise.
13203 (execute_sm): Likewise.
13204 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
13205 (remove_exits_and_undefined_stmts): Likewise.
13206 (remove_redundant_iv_tests): Likewise.
13207 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
13208 (adjust_iv_update_pos): Likewise.
13209 * tree-ssa-math-opts.c (bswap_replace): Likewise.
13210 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
13211 (value_replacement): Likewise.
13212 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
13213 * tree-ssa-pre.c (print_pre_expr): Likewise.
13214 (get_representative_for): Likewise.
13215 (create_expression_by_pieces): Likewise.
13216 (insert_into_preds_of_block): Likewise.
13217 (eliminate_insert): Likewise.
13218 (eliminate_dom_walker::before_dom_children): Likewise.
13219 (eliminate): Likewise.
13220 (remove_dead_inserted_code): Likewise.
13221 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
13222 * tree-ssa-reassoc.c (get_rank): Likewise.
13223 (eliminate_duplicate_pair): Likewise.
13224 (eliminate_plus_minus_pair): Likewise.
13225 (eliminate_not_pairs): Likewise.
13226 (undistribute_ops_list): Likewise.
13227 (eliminate_redundant_comparison): Likewise.
13228 (update_range_test): Likewise.
13229 (optimize_range_tests_var_bound): Likewise.
13230 (optimize_vec_cond_expr): Likewise.
13231 (rewrite_expr_tree): Likewise.
13232 (rewrite_expr_tree_parallel): Likewise.
13233 (linearize_expr): Likewise.
13234 (break_up_subtract): Likewise.
13235 (linearize_expr_tree): Likewise.
13236 (attempt_builtin_powi): Likewise.
13237 (attempt_builtin_copysign): Likewise.
13238 (transform_stmt_to_copy): Likewise.
13239 (transform_stmt_to_multiply): Likewise.
13240 (dump_ops_vector): Likewise.
13241 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
13242 (print_scc): Likewise.
13243 (set_ssa_val_to): Likewise.
13244 (visit_reference_op_store): Likewise.
13245 (visit_use): Likewise.
13246 (sccvn_dom_walker::before_dom_children): Likewise.
13247 (run_scc_vn): Likewise.
13248 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
13250 (expr_hash_elt::print): Likewise.
13251 (const_and_copies::pop_to_marker): Likewise.
13252 (const_and_copies::record_const_or_copy_raw): Likewise.
13253 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
13254 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
13255 (dump_predicates): Likewise.
13256 (find_uninit_use): Likewise.
13257 (warn_uninitialized_phi): Likewise.
13258 (pass_late_warn_uninitialized::execute): Likewise.
13259 * tree-ssa.c (verify_vssa): Likewise.
13260 (verify_ssa): Likewise.
13261 (maybe_optimize_var): Likewise.
13262 * tree-vrp.c (dump_value_range): Likewise.
13263 (dump_all_value_ranges): Likewise.
13264 (dump_asserts_for): Likewise.
13265 (register_edge_assert_for_2): Likewise.
13266 (vrp_visit_cond_stmt): Likewise.
13267 (vrp_visit_switch_stmt): Likewise.
13268 (vrp_visit_stmt): Likewise.
13269 (vrp_visit_phi_node): Likewise.
13270 (simplify_cond_using_ranges_1): Likewise.
13271 (fold_predicate_in): Likewise.
13272 (evrp_dom_walker::before_dom_children): Likewise.
13273 (evrp_dom_walker::push_value_range): Likewise.
13274 (evrp_dom_walker::pop_value_range): Likewise.
13275 (execute_early_vrp): Likewise.
13277 2017-05-16 Richard Biener <rguenther@suse.de>
13279 * dwarf2out.c (loc_list_from_tree_1): Do not create
13280 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
13282 2017-05-16 Richard Biener <rguenther@suse.de>
13284 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
13286 (note_variable_value_in_expr): If we resolved the decl ref
13287 do not push to the stack.
13289 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
13291 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
13292 operations in fast-math mode.
13293 (vaddq_f16): Likewise.
13294 (vmul_f16): Likewise.
13295 (vmulq_f16): Likewise.
13296 (vsub_f16): Likewise.
13297 (vsubq_f16): Likewise.
13298 * config/arm/neon.md (add<mode>3): New.
13300 (fma:<VH:mode>3): New. Also remove outdated comment.
13303 2017-05-16 Martin Liska <mliska@suse.cz>
13307 * ipa-devirt.c (warn_types_mismatch): Fix typo.
13308 (odr_types_equivalent_p): Likewise.
13310 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
13312 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
13314 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
13317 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
13318 non-interunit SSE move alternatives with '?'.
13319 (zero-extendsidi peephole2): New peephole to skip intermediate
13320 general register in SSE zero-extend sequence.
13322 2017-05-15 Jeff Law <law@redhat.com>
13324 * reorg.c (relax_delay_slots): Create a new variable to hold
13325 the temporary target rather than clobbering TARGET_LABEL.
13327 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
13328 missing argument to extract_bit_field call.
13329 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
13331 2017-05-15 Martin Liska <mliska@suse.cz>
13334 * gcc.c (process_command): Do not allow empty argument of -o option.
13336 2017-05-15 Renlin Li <renlin.li@arm.com>
13338 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
13339 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
13340 * config/aarch64/constraints.md (Usf): Add long call check.
13341 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
13342 (call_value): Likewise.
13343 (sibcall): Likewise.
13344 (sibcall_value): Likewise.
13346 (call_value_insn): New.
13347 (sibcall_insn): Update rtx pattern.
13348 (sibcall_value_insn): Likewise.
13349 (call_internal): Remove.
13350 (call_value_internal): Likewise.
13351 (sibcall_internal): Likewise.
13352 (sibcall_value_internal): Likewise.
13353 (call_reg): Likewise.
13354 (call_symbol): Likewise.
13355 (call_value_reg): Likewise.
13356 (call_value_symbol): Likewise.
13358 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
13361 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
13363 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
13365 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
13366 compatible with CCGOCmode and with CCZmode.
13368 2017-05-14 Martin Sebor <msebor@redhat.com>
13370 PR middle-end/77671
13371 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
13372 (gimple_fold_builtin_snprintf): Same.
13373 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
13374 (gimple_fold_builtin_snprintf): Same.
13375 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
13376 of character types.
13377 (is_call_safe): New function.
13378 (try_substitute_return_value): Call it.
13379 (try_simplify_call): New function.
13380 (pass_sprintf_length::handle_gimple_call): Call it.
13382 2017-05-14 Martin Sebor <msebor@redhat.com>
13384 PR middle-end/80669
13385 * builtins.c (expand_builtin_stpncpy): Simplify.
13387 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
13389 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
13390 * config/i386/i386.h
13391 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
13392 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
13393 (struct machine_function): Add new members call_ms2sysv,
13394 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
13395 (struct machine_frame_state): New fields sp_realigned and
13396 sp_realigned_offset.
13397 * config/i386/i386.c
13398 (enum xlogue_stub): New enum.
13399 (enum xlogue_stub_sets): New enum.
13400 (class xlogue_layout): New class.
13401 (struct ix86_frame): New fields stack_realign_allocate_offset,
13402 stack_realign_offset and outlined_save_offset. Modify comments to
13403 detail stack layout when using out-of-line stubs.
13404 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
13405 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
13406 -mcall-ms2sysv-xlogues.
13407 (stub_managed_regs): New static variable.
13408 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
13409 registers managed by out-of-line stub.
13410 (disable_call_ms2sysv_xlogues): New function.
13411 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
13412 m->call_ms2sysv when appropriate and compute frame layout for
13414 (sp_valid_at, fp_valid_at): New inline functions.
13415 (choose_basereg): New function.
13416 (choose_baseaddr): Add align parameter, use choose_basereg and modify
13418 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
13419 Use align parameter of choose_baseaddr to generated aligned SSE movs
13421 (pro_epilogue_adjust_stack): Modify to track
13422 machine_frame_state::sp_realigned.
13423 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
13424 (ix86_nsaved_sseregs): Likewise.
13425 (ix86_emit_save_regs): Likewise.
13426 (ix86_emit_save_regs_using_mov): Likewise.
13427 (ix86_emit_save_sse_regs_using_mov): Likewise.
13428 (get_scratch_register_on_entry): Likewise.
13429 (gen_frame_set): New function.
13430 (gen_frame_load): Likewise.
13431 (gen_frame_store): Likewise.
13432 (emit_outlined_ms2sysv_save): Likewise.
13433 (emit_outlined_ms2sysv_restore): Likewise.
13434 (ix86_expand_prologue): Modify stack re-alignment code and call
13435 emit_outlined_ms2sysv_save when appropriate.
13436 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
13437 parameter rtx_insn *insn, which allows the function to be used to only
13438 generate the notes.
13439 (ix86_expand_epilogue): Modify validity checks of frame and stack
13440 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
13441 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
13442 * config/i386/predicates.md
13443 (save_multiple): New predicate.
13444 (restore_multiple): Likewise.
13445 * config/i386/sse.md
13446 (save_multiple<mode>): New pattern.
13447 (save_multiple_realign<mode>): Likewise.
13448 (restore_multiple<mode>): Likewise.
13449 (restore_multiple_and_return<mode>): Likewise.
13450 (restore_multiple_leave_return<mode>): Likewise.
13451 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
13453 2017-05-14 Julia Koval <julia.koval@intel.com>
13455 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
13456 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
13457 (__builtin_ia32_xsetbv): New builtins.
13458 * config/i386/i386.c (ix86_expand_special_args_builtin):
13460 (ix86_expand_builtin): Special expand for new intrinsics.
13461 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
13462 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
13463 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
13465 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13467 * cfganal.c (inverted_post_order_compute): Change argument type
13469 * cfganal.h (inverted_post_order_compute): Adjust prototype.
13470 * df-core.c (rest_of_handle_df_initialize): Adjust.
13471 (rest_of_handle_df_finish): Likewise.
13472 (df_analyze_1): Likewise.
13473 (df_analyze): Likewise.
13474 (loop_inverted_post_order_compute): Change argument to be a vec *.
13475 (df_analyze_loop): Adjust.
13476 (df_get_n_blocks): Likewise.
13477 (df_get_postorder): Likewise.
13478 * df.h (struct df_d): Change field to be a vec.
13479 * lcm.c (compute_laterin): Adjust.
13480 (compute_available): Likewise.
13481 * lra-lives.c (lra_create_live_ranges_1): Likewise.
13482 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
13483 * tree-ssa-pre.c (compute_antic): Likewise.
13485 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13487 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
13488 (depth_first_search::depth_first_search): Change structure init
13489 function to this constructor.
13490 (depth_first_search::add_bb): Rename function to this member.
13491 (depth_first_search::execute): Likewise.
13492 (flow_dfs_compute_reverse_finish): Adjust.
13494 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13496 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
13497 (longest_simple_path): Likewise.
13498 * shrink-wrap.c (spread_components): Likewise.
13499 (disqualify_problematic_components): Likewise.
13500 (emit_common_heads_for_components): Likewise.
13501 (emit_common_tails_for_components): Likewise.
13502 (insert_prologue_epilogue_for_components): Likewise.
13504 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13506 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
13509 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13511 * df-core.c (df_set_blocks): Start using auto_bitmap.
13512 (df_compact_blocks): Likewise.
13513 * df-problems.c (df_rd_confluence_n): Likewise.
13514 * df-scan.c (df_insn_rescan_all): Likewise.
13515 (df_process_deferred_rescans): Likewise.
13516 (df_update_entry_block_defs): Likewise.
13517 (df_update_exit_block_uses): Likewise.
13518 (df_entry_block_bitmap_verify): Likewise.
13519 (df_exit_block_bitmap_verify): Likewise.
13520 (df_scan_verify): Likewise.
13521 * lra-constraints.c (lra_constraints): Likewise.
13522 (undo_optional_reloads): Likewise.
13523 (lra_undo_inheritance): Likewise.
13524 * lra-remat.c (calculate_gen_cands): Likewise.
13525 (do_remat): Likewise.
13526 * lra-spills.c (assign_spill_hard_regs): Likewise.
13527 (spill_pseudos): Likewise.
13528 * tree-ssa-pre.c (bitmap_set_and): Likewise.
13529 (bitmap_set_subtract_values): Likewise.
13531 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13533 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
13534 management with auto_bitmap.
13535 (fix_inter_tick): Likewise.
13536 (fix_recovery_deps): Likewise.
13537 * ira.c (add_store_equivs): Likewise.
13538 (find_moveable_pseudos): Likewise.
13539 (split_live_ranges_for_shrink_wrap): Likewise.
13540 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
13541 (rtx_reuse_manager::seen_def_p): Likewise.
13542 (rtx_reuse_manager::set_seen_def): Likewise.
13543 * print-rtl.h (class rtx_reuse_manager): Likewise.
13545 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13547 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
13549 (migrate_btr_def): Likewise.
13550 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
13551 * df-core.c (loop_post_order_compute): Likewise.
13552 (loop_inverted_post_order_compute): Likewise.
13553 * hsa-common.h: Likewise.
13554 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
13555 * init-regs.c (initialize_uninitialized_regs): Likewise.
13556 * ipa-inline.c (resolve_noninline_speculation): Likewise.
13557 (inline_small_functions): Likewise.
13558 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
13559 * ira.c (combine_and_move_insns): Likewise.
13560 (build_insn_chain): Likewise.
13561 * loop-invariant.c (find_invariants): Likewise.
13562 * lower-subreg.c (propagate_pseudo_copies): Likewise.
13563 * predict.c (tree_predict_by_opcode): Likewise.
13564 (predict_paths_leading_to): Likewise.
13565 (predict_paths_leading_to_edge): Likewise.
13566 (estimate_loops_at_level): Likewise.
13567 (estimate_loops): Likewise.
13568 * shrink-wrap.c (try_shrink_wrapping): Likewise.
13569 (spread_components): Likewise.
13570 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
13571 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
13572 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
13573 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
13574 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
13575 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
13576 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
13577 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
13578 (mark_threaded_blocks): Likewise.
13579 (thread_through_all_blocks): Likewise.
13580 * tree-ssa.c (verify_ssa): Likewise.
13581 (execute_update_addresses_taken): Likewise.
13582 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
13584 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13586 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
13588 (post_order_compute): Likewise.
13589 (inverted_post_order_compute): Likewise.
13590 (pre_and_rev_post_order_compute_fn): Likewise.
13592 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13594 * genrecog.c (int_set::int_set): Explicitly construct our
13595 auto_vec base class.
13596 * vec.h (auto_vec::auto_vec): New constructor.
13598 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13600 * bitmap.h (class auto_bitmap): New constructor taking
13601 bitmap_obstack * argument.
13603 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13605 * bitmap.h (class auto_bitmap): Change type of m_bits to
13606 bitmap_head, and adjust ctor / dtor and member operators.
13608 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
13610 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
13611 when returned register mode doesn't match original mode.
13613 2017-05-12 Jeff Law <law@redhat.com>
13614 Jakub Jelinek <jakub@redhat.com>
13616 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
13617 we look for cc setter after the compare-elim changes.
13618 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
13619 within the vector to match what compare-elim now expects.
13620 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
13621 (xorsi3_flags, one_cmplsi2_flags): Likewise.
13623 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
13624 after the compare-elim changes.
13625 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
13626 the vector to match what compare-elim now expects.
13627 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
13628 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
13629 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
13630 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
13631 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
13633 * config/visium/visium.c (single_set_and_flags): Fix where
13634 we look for cc setter after the compare-elim changes.
13635 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
13636 with the vector to match what compare-elim now expects.
13637 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
13638 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
13639 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
13640 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
13641 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
13642 (neg<mode>2_insn_set_overflow): Likewise.
13644 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
13646 PR middle-end/79794
13647 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
13648 maybe_expand_insn call, set ops[0].target. If still set after call,
13649 set alt_rtl. Add extra arg to recursive calls.
13650 (extract_bit_field): Add alt_rtl argument. Pass to
13652 * expmed.h (extract_bit_field): Fix prototype.
13653 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
13654 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
13655 to extract_bit_field_calls.
13656 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
13657 Pass alt_rtl to extract_bit_field calls.
13658 * calls.c (store_unaligned_arguments_into_psuedos)
13659 load_register_parameters): Pass extra NULL to extract_bit_field calls.
13660 * optabs.c (maybe_legitimize_operand): Clear op->target when call
13662 * optabs.h (struct expand_operand): Add target bitfield.
13664 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
13666 * compare-elim.c (try_eliminate_compare): Canonicalize
13667 operation with embedded compare to
13668 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
13669 (set (reg) (operation)].
13671 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
13673 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
13676 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
13677 cost of adding a carry flag for ADC instruction.
13678 [case MINUS]: Ignore the cost of subtracting a carry flag
13679 for SBB instruction.
13681 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
13683 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
13685 * config/rs6000/bmiintrin.h: New file.
13686 * config/rs6000/bmi2intrin.h: New file.
13687 * config/rs6000/x86intrin.h: New file.
13689 2017-05-12 Jeff Law <law@redhat.com>
13691 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
13694 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
13696 PR middle-end/80707
13697 * tree-cfg.c: Remove cfg edges of unreachable case statements.
13699 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13701 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
13702 early expansion of vector divide builtins.
13703 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
13704 builtins identified as having unsigned arguments.
13706 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13708 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
13709 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
13710 expansion of vector logical operations (and, andc, or, xor,
13713 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13715 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
13716 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
13718 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13720 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
13721 early GIMPLE expansion of vector multiplies.
13723 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13725 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
13726 TARGET_HAVE_MOVT conditional.
13727 (movt splitter): Likewise.
13729 2017-05-12 Richard Biener <rguenther@suse.de>
13731 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
13732 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13733 Fold all stmts not inplace.
13735 2017-05-12 Richard Biener <rguenther@suse.de>
13737 PR tree-optimization/80713
13738 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
13739 inserted_exprs bit for not removed stmts.
13741 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
13743 PR middle-end/69921
13744 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
13745 parallelized" attribute for parallelized OpenACC kernels.
13746 * omp-offload.c (execute_oacc_device_lower): Use it.
13748 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
13749 Set "oacc kernels" attribute.
13750 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
13751 parameter. Adjust all users.
13752 (oacc_fn_attrib_kernels_p): Remove function.
13753 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
13754 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
13755 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
13756 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
13757 assert "oacc kernels" attribute is set.
13759 2017-05-11 Carl Love <cel@us.ibm.com>
13761 * config/rs6000/rs6000-c: Add support for built-in functions
13762 vector unsigned char vec_popcnt (vector signed char)
13763 vector unsigned char vec_popcnt (vector unsigned char)
13764 vector unsigned short vec_popcnt (vector signed short)
13765 vector unsigned short vec_popcnt (vector unsigned short)
13766 vector unsigned int vec_popcnt (vector signed int)
13767 vector unsigned int vec_popcnt (vector unsigned int)
13768 vector unsigned long long vec_popcnt (vector signed long long)
13769 vector unsigned long long vec_popcnt (vector unsigned long long)
13770 vector signed long long vec_slo (vector signed long long,
13771 vector signed char)
13772 vector signed long long vec_slo (vector signed long long,
13773 vector unsigned char)
13774 vector unsigned long long vec_slo (vector unsigned long long,
13775 vector signed char)
13776 vector unsigned long long vec_slo (vector unsigned long long,
13777 vector unsigned char)
13778 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
13779 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
13780 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
13781 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
13782 * doc/extend.texi: Update the built-in documentation file for the
13783 new built-in functions.
13785 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
13787 * attribs.h (sorted_attr_string): Move machine independent
13788 functions for target clone support from the i386 port to common
13789 code. Rename ix86_function_versions to common_function_versions.
13790 Rename make_name to make_unique_name.
13791 (common_function_versions): Likewise.
13792 (make_unique_name): Likewise.
13793 (make_dispatcher_decl): Likewise.
13794 (is_function_default_version): Likewise.
13795 * attribs.c (attr_strcmp): Likewise.
13796 (sorted_attr_string): Likewise.
13797 (common_function_versions): Likewise.
13798 (make_unique_name): Likewise.
13799 (make_dispatcher_decl): Likewise.
13800 (is_function_default_version): Likewise.
13801 * config/i386/i386.c (attr_strcmp): Likewise.
13802 (sorted_attr_string): Likewise.
13803 (ix86_function_versions): Likewise.
13804 (make_name): Likewise.
13805 (make_dispatcher_decl): Likewise.
13806 (is_function_default_version): Likewise.
13807 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
13809 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13812 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
13813 Account for direct move costs for vec_construct of integer
13816 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
13819 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
13820 (UNSPEC_STX_ATOMIC): Ditto.
13821 (loaddi_via_sse): New insn.
13822 (storedi_via_sse): Ditto.
13823 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
13824 Update corresponding peephole2 patterns.
13825 (atomic_storedi_fpu): Ditto.
13827 2017-05-11 Julia Koval <julia.koval@intel.com>
13829 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
13830 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
13832 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
13833 (__builtin_ia32_rsqrt14ss_mask): New builtins.
13834 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
13836 2017-05-11 Nathan Sidwell <nathan@acm.org>
13838 * graphite-poly.c: Include dumpfile.h.
13840 * dumpfle.h (dump_function): Declare here ...
13841 * tree-dump.h (dump_function): ... not here.
13842 * dumpfile.c: #include tree-cfg.h.
13843 (dump_function): Move here from ...
13844 * tree-dump.c (dump_function): ... here.
13845 * gimplify.c: #include splay-tree.h, not tree-dump.h.
13846 * graphite-poly.c: Don't include tree-dump.h.
13847 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
13848 * print-tree.c: Likewise.
13849 * stor-layout.c: Likewise.
13850 * tree-nested.c: Likewise.
13852 * dumpfile.c (dump_start): Use TDF_FLAGS.
13853 (dump_enable_all): Fix TDF_KIND check thinko.
13855 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
13857 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13858 array entries to represent two legal parameterizations of the
13859 overloaded __builtin_cmpb function, as represented by the
13860 P6_OV_BUILTIN_CMPB constant.
13861 (altivec_resolve_overloaded_builtin): Add special case handling
13862 for the __builtin_cmpb function, as represented by the
13863 P6_OV_BUILTIN_CMPB constant.
13864 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
13865 (BU_P6_64BIT_2): New macro.
13866 (BU_P6_OVERLOAD_2): New macro
13867 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
13868 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
13869 (CMPB): Add overload support to represent both 32-bit and 64-bit
13870 compare-bytes function.
13871 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
13872 support for TARGET_CMPB.
13873 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
13874 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
13875 documentation of the __builtin_cmpb overloaded built-in function.
13877 2017-05-11 Richard Biener <rguenther@suse.de>
13879 PR tree-optimization/80705
13880 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
13881 bases are not vectorizable.
13883 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13885 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
13886 when counting register pressure.
13888 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13890 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
13891 (iv_ca_more_deps): Renamed to ...
13892 (iv_ca_compare_deps): ... this.
13893 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
13895 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13897 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
13899 (determine_group_iv_costs): ... here.
13900 (find_inv_vars_cb): Record inv var if it's not recorded before.
13902 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13904 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
13905 (get_shiftadd_cost): Ditto.
13907 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13909 * tree-ssa-address.c: Include header file.
13910 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
13912 (add_to_parts): Refactor.
13913 (addr_to_parts): New parameter. Update use of move_hint_to_base.
13914 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
13917 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13919 PR tree-optimization/53090
13920 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
13922 (extract_cond_operands): Detect condition with IV on both sides
13923 and return COMP_IV_EXPR_2.
13924 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
13925 (rewrite_use_compare): Simplify by removing call to function
13926 extract_cond_operands.
13928 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13930 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
13931 (extract_cond_operands): Detect condition comparing against non-
13932 invariant bound and return appropriate enum value.
13933 (find_interesting_uses_cond): Update use of extract_cond_operands.
13934 Handle its return value accordingly.
13935 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
13937 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13939 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
13940 nonlinear iv_use computation in loop invariant sensitive way.
13942 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13944 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
13945 (find_iv_candidates): Call relate_compare_use_with_all_cands.
13947 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13949 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
13950 (dump_cand): Support iv_cand.inv_exprs.
13951 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
13953 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
13956 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13958 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
13960 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
13961 as local function. Include necessary header files.
13962 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
13964 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13966 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
13968 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13970 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
13971 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
13972 RSHIFT_EXPR and BIT_NOT_EXPR.
13974 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13976 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
13977 (adjust_setup_cost): New parameter supporting round up adjustment.
13978 (struct address_cost_data): Delete.
13979 (force_expr_to_var_cost): Don't bound cost with spill_cost.
13980 (split_address_cost, ptr_difference_cost): Delete.
13981 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
13982 (struct ainc_cost_data): New struct.
13983 (get_address_cost_ainc): New function.
13984 (get_address_cost, get_computation_cost): Reimplement.
13985 (determine_group_iv_cost_address): Record inv_expr for all uses of
13987 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
13988 (iv_ca_has_deps): Reimplemented to ...
13989 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
13991 (iv_ca_extend): Call iv_ca_more_deps.
13993 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13995 * tree-ssa-address.c (struct mem_address): Move to header file.
13996 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
13997 * tree-ssa-address.h (struct mem_address): Move from C file.
13998 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
14000 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14002 * tree-affine.h (aff_combination_type): New interface.
14003 (aff_combination_zero_p): Remove static.
14004 (aff_combination_const_p): New interface.
14005 (aff_combination_singleton_var_p): New interfaces.
14007 2017-05-11 Richard Biener <rguenther@suse.de>
14009 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
14010 Skip unreachable blocks and destinations.
14011 (eliminate): Move stmt removal and fixup ...
14012 (fini_eliminate): ... here. Skip inserted exprs.
14013 (pass_pre::execute): Move fini_pre after fini_eliminate.
14014 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
14015 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
14016 PRE to get rid of dead code that has invalid SSA form and
14017 split critical edges again.
14019 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14021 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
14023 2017-05-11 Richard Biener <rguenther@suse.de>
14025 * passes.c (execute_function_todo): Verify loops if they are
14026 said to be up-to-date.
14027 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
14028 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
14030 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
14033 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
14034 handle calling assemble_external ourself.
14037 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
14038 modes with zero size. Enhance comment.
14040 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14042 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
14043 built-ins for vec_xl and vec_xst with short and char pointer
14046 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
14048 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
14049 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
14050 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
14051 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
14052 (_mm_maskz_min_round_ss): New intrinsics.
14053 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
14054 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
14055 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
14056 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
14057 (__builtin_ia32_minss_mask_round): New builtins.
14058 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14059 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
14060 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
14062 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
14063 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
14065 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
14068 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
14070 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
14071 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
14072 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
14073 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
14074 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
14075 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
14076 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
14077 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14078 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
14079 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
14080 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
14081 (__builtin_ia32_mulss_mask_round): New builtins.
14082 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14083 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
14084 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
14086 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
14087 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
14089 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
14092 2017-05-10 Julia Koval <julia.koval@intel.com>
14094 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
14095 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
14096 (_mm256_setr_m128i): New intrinsics.
14098 2017-05-10 Julia Koval <julia.koval@intel.com>
14100 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
14101 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
14102 (_mm_maskz_rcp14_ss): New intrinsics.
14103 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
14104 (__builtin_ia32_rcp14ss_mask): New builtins.
14105 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
14107 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
14109 PR tree-optimization/51513
14110 * tree-cfg.c (gimple_seq_unreachable_p): New function.
14111 (assert_unreachable_fallthru_edge_p): Use it.
14112 (group_case_labels_stmt): Likewise.
14113 * tree-cfg.h: Prototype it.
14114 * stmt.c: Include cfghooks.h and tree-cfg.h.
14115 (emit_case_dispatch_table) <gap_label>: New local variable.
14116 Use it to fill dispatch table gaps.
14117 Test for default_label before updating probabilities.
14118 (expand_case) <default_label>: Remove unneeded initialization.
14119 Test for unreachable default case statement and remove its edge.
14120 Set default_label accordingly.
14121 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
14123 2017-05-10 Carl Love <cel@us.ibm.com>
14125 * config/rs6000/rs6000-c: Add support for built-in functions
14126 vector signed char vec_neg (vector signed char)
14127 vector signed short int vec_neg (vector short int)
14128 vector signed int vec_neg (vector signed int)
14129 vector signed long long vec_neg (vector signed long long)
14130 vector float vec_neg (vector float)
14131 vector double vec_neg (vector double)
14132 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
14134 * config/rs6000/altivec.h: Add define for vec_neg
14135 * doc/extend.texi: Update the built-in documentation for the
14136 new built-in functions.
14138 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14140 PR tree-optimization/77644
14141 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
14143 2017-05-10 Nathan Sidwell <nathan@acm.org>
14145 * dumpfile.h (TDI_lang_all): New.
14146 (TDF_KIND): New. Renumber others
14147 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
14149 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
14151 (get_dump_file_name): Adjust suffix generation.
14152 (dump_enable_all): Use TDF_KIND.
14153 * doc/invoke.texi (-fdump-lang-all): Document.
14155 * dumpfile.h: Tabify.
14157 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
14160 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
14161 Move member access before delete.
14163 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
14165 * tree-inline.c (expand_call_inline): Split block at stmt
14168 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
14171 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
14172 are now unused after splitting mov{sf,sd}_hardfloat.
14173 (f32_lr2): Likewise.
14174 (f32_lm): Likewise.
14175 (f32_lm2): Likewise.
14176 (f32_li): Likewise.
14177 (f32_li2): Likewise.
14178 (f32_lv): Likewise.
14179 (f32_sr): Likewise.
14180 (f32_sr2): Likewise.
14181 (f32_sm): Likewise.
14182 (f32_sm2): Likewise.
14183 (f32_si): Likewise.
14184 (f32_si2): Likewise.
14185 (f32_sv): Likewise.
14186 (f32_dm): Likewise.
14187 (f32_vsx): Likewise.
14188 (f32_av): Likewise.
14189 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
14190 For movsf, order stores so the VSX stores occur before the GPR
14191 store which encourages the register allocator to use a traditional
14192 FPR instead of a GPR. For movsd, order the stores so that the GPR
14193 store comes before the VSX stores to allow the power6 to work.
14194 This is due to the power6 not having a 32-bit integer store
14195 instruction from a FPR.
14196 (movsf_hardfloat): Likewise.
14197 (movsd_hardfloat): Likewise.
14199 2017-05-09 Martin Sebor <msebor@redhat.com>
14201 PR translation/80280
14202 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
14205 PR translation/80280
14206 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
14207 data member added in r247778.
14208 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
14210 2017-05-09 Nathan Sidwell <nathan@acm.org>
14212 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
14214 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
14217 2017-05-09 Marek Polacek <polacek@redhat.com>
14219 * doc/invoke.texi: Fix typo.
14221 2017-05-09 Richard Biener <rguenther@suse.de>
14223 * tree-vrp.c (vrp_val_is_max): Adjust comment.
14224 (vrp_val_is_min): Likewise.
14225 (set_value_range_to_value): Likewise.
14226 (set_value_range_to_nonnegative): Likewise.
14227 (gimple_assign_nonzero_p): Likewise.
14228 (gimple_stmt_nonzero_p): Likewise.
14229 (vrp_int_const_binop): Likewise. Remove unreachable case.
14230 (adjust_range_with_scev): Adjust comments.
14231 (compare_range_with_value): Likewise.
14232 (extract_range_from_phi_node): Likewise.
14233 (test_for_singularity): Likewise.
14235 2017-05-09 Richard Biener <rguenther@suse.de>
14237 * tree-vrp.c (get_single_symbol): Add assert that we don't
14238 get overflowed constants as invariant part.
14239 (compare_values_warnv): Add comment before the TREE_NO_WARNING
14240 checks. Use wi::cmp instead of recursing for integer constants.
14241 (compare_values): Just ignore whether we assumed undefined
14242 overflow instead of failing the compare.
14243 (extract_range_for_var_from_comparison_expr): Add comment before the
14244 TREE_NO_WARNING sets.
14245 (test_for_singularity): Likewise.
14246 (extract_range_from_comparison): Do not disable optimization
14247 when we assumed undefined overflow.
14248 (extract_range_basic): Remove init of unused var.
14250 2017-05-09 Richard Biener <rguenther@suse.de>
14252 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
14253 (extract_range_from_multiplicative_op_1): Adjust.
14254 (extract_range_from_binary_expr_1): Use int_const_binop.
14256 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
14259 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
14260 rs6000_store_data_bypass_p in seven define_bypass directives and
14261 in several comments.
14262 * config/rs6000/rs6000-protos.h: Add prototype for
14263 rs6000_store_data_bypass_p function.
14264 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
14265 function implements slightly different (rs6000-specific) semantics
14266 than store_data_bypass_p, returning false rather than aborting
14267 with assertion error when arguments do not satisfy the
14268 requirements of store data bypass.
14269 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
14270 rs6000_store_data_bypass_p.
14272 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
14274 * config/xtensa/xtensa-protos.h
14275 (xtensa_initial_elimination_offset): New declaration.
14276 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
14277 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
14278 macro definition, add case for FRAME_POINTER_REGNUM when
14279 FRAME_GROWS_DOWNWARD.
14280 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
14281 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
14282 xtensa_initial_elimination_offset.
14284 2017-05-08 Nathan Sidwell <nathan@acm.org>
14286 * doc/invoke.texi: Alphabetize -fdump options.
14288 2017-05-08 Martin Sebor <msebor@redhat.com>
14290 PR translation/80280
14291 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
14293 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
14295 * target.def (compute_frame_layout): New optional target hook.
14296 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
14297 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
14298 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
14300 * reload1.c (verify_initial_elim_offsets): Likewise.
14301 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
14302 (use_simple_return_p): Call arm_compute_frame_layout if needed.
14303 (arm_get_frame_offsets): Split up into this ...
14304 (arm_compute_frame_layout): ... and this function.
14306 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
14308 * config/aarch64/constraints.md (Usa): New constraint.
14309 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
14311 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
14313 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
14314 with_multilib_list after it has been checked.
14316 2017-05-08 Richard Biener <rguenther@suse.de>
14318 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
14319 (bitmap_set_subtract_values): Likewise.
14321 2017-05-08 Richard Biener <rguenther@suse.de>
14323 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
14324 (gimple_assign_nonzero): ... this and remove strict_overflow_p
14326 (gimple_stmt_nonzero_warnv_p): Rename to ...
14327 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
14329 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
14330 (extract_range_basic): Adjust, do not disable propagation on
14331 strict overflow sensitive simplification.
14332 (vrp_visit_cond_stmt): Likewise.
14334 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
14336 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
14337 body size unconditionally.
14339 2017-05-07 Jeff Law <law@redhat.com>
14342 2017-05-06 Jeff Law <law@redhat.com>
14343 PR tree-optimization/78496
14344 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
14347 PR tree-optimization/78496
14348 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
14349 (simplify_stmt_using_ranges): Call it.
14350 (vrp_dom_walker::before_dom_children): Extract equivalences
14351 from an ASSERT_EXPR with an equality comparison against a
14354 2017-05-06 Jeff Law <law@redhat.com>
14356 PR tree-optimization/78496
14357 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
14360 PR tree-optimization/78496
14361 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
14362 (simplify_stmt_using_ranges): Call it.
14363 (vrp_dom_walker::before_dom_children): Extract equivalences
14364 from an ASSERT_EXPR with an equality comparison against a
14367 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
14369 * lra-constraints.c (lra_copy_reg_equiv): New function.
14370 (split_reg): Use it to copy equivalence information from the
14371 original register to the spill register.
14373 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
14375 PR rtl-optimization/75964
14376 * simplify-rtx.c (simplify_const_relational_operation): Remove
14377 invalid handling of comparisons of integer ABS.
14379 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
14381 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
14382 initialize to zero.
14383 (init_regs): Remove declaration.
14384 (function_arg_advance_32): Initialize error_p as boolean variable.
14386 2017-05-05 Nathan Sidwell <nathan@acm.org>
14388 * store-motion.c (remove_reachable_equiv_notes): Reformat long
14389 lines. Use for (;;).
14391 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14393 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
14394 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
14395 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
14396 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
14397 VF=2 that require versioning.
14399 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14401 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
14404 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14406 * diagnostic.h (diagnostic_override_option_index): Convert from
14407 macro to inline function.
14409 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14411 * diagnostic.c (last_module_changed_p): New function.
14412 (set_last_module): New function.
14413 (diagnostic_report_current_module): Convert macro usage to
14414 the above functions.
14415 * diagnostic.h (diagnostic_context::last_module): Strengthen
14416 from const line_map * to const line_map_ordinary *.
14417 (diagnostic_last_module_changed): Delete macro.
14418 (diagnostic_set_last_module): Delete macro.
14420 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14422 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
14423 with diagnostic_report_diagnostic.
14424 (diagnostic_n_impl_richloc): Likewise.
14425 * diagnostic.h (report_diagnostic): Delete macro.
14426 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
14427 with diagnostic_report_diagnostic.
14428 * substring-locations.c (format_warning_va): Likewise.
14430 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14432 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
14433 save/restor of format_spec. Move option-printing code to...
14434 (print_option_information): ...this new function, and
14435 reimplement by simply printing to the pretty_printer,
14436 rather than appending to the format string.
14438 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14440 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
14441 handling logic into...
14442 (update_effective_level_from_pragmas): ...this new function.
14444 2017-05-04 Andrew Waterman <andrew@sifive.com>
14446 * config/riscv/riscv.opt (mstrict-align): New option.
14447 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
14448 (SLOW_UNALIGNED_ACCESS): Define.
14449 (riscv_slow_unaligned_access): Declare.
14450 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
14452 (riscv_slow_unaligned_access): New variable.
14453 (rocket_tune_info): Set slow_unaligned_access to true.
14454 (optimize_size_tune_info): Set slow_unaligned_access to false.
14455 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
14456 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
14457 (riscv_option_override): Set riscv_slow_unaligned_access.
14458 * doc/invoke.texi: Add -mstrict-align to RISC-V.
14460 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
14462 * config/riscv/riscv.md: Unify indentation.
14464 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14469 * config/rs6000/rs6000.md (u code attribute): Add FIX and
14471 (extendsi<mode>2): Add support for doing sign extension via
14472 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
14473 don't have ISA 3.0 instructions.
14474 (extendsi<mode>2 splitter): Likewise.
14475 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
14476 generate the normal insns since SImode can now go in vector
14477 registers. Disallow the special UNSPECs needed for previous
14478 machines to hide SImode being used. Add new insns
14479 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
14480 (fix_trunc<mode>si2_stfiwx): Likewise.
14481 (fix_trunc<mode>si2_internal): Likewise.
14482 (fixuns_trunc<mode>si2): Likewise.
14483 (fixuns_trunc<mode>si2_stfiwx): Likewise.
14484 (fctiw<u>z_<mode>_smallint): Likewise.
14485 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
14486 of floating point to 32-bit integer from doing a direct move to
14487 the GPR registers to do a store.
14488 (fctiwz_<mode>): Break long line.
14490 2017-05-05 Bin Cheng <bin.cheng@arm.com>
14492 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
14493 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
14494 (addr_list, addr_offset_valid_p): New.
14495 (split_address_groups): Check offset validity with above function.
14496 (gt-tree-ssa-loop-ivopts.h): Include header file.
14498 2017-05-05 Nathan Sidwell <nathan@acm.org>
14500 * config.gcc (arm*-*-*): Add missing 'fi'.
14502 2017-05-05 Steve Ellcey <sellcey@cavium.com>
14504 * doc/invoke.texi (-fopt-info): Explicitly say order of options
14505 included in -fopt-info does not matter.
14506 * doc/optinfo.texi (-fopt-info): Fix description of default
14507 behavour. Explicitly say order of options included in -fopt-info
14510 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
14512 * config.gcc: Allow combinations of aprofile and rmprofile values for
14513 --with-multilib-list.
14514 * config/arm/t-multilib: New file.
14515 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
14516 variables. Remove setting of ISA and floating-point ABI in
14517 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
14518 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
14519 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
14520 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
14522 * config/arm/t-rmprofile: Likewise except for the matches changes.
14523 * doc/install.texi (--with-multilib-list): Document the combination of
14524 aprofile and rmprofile values and warn about pitfalls in doing that.
14526 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
14528 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
14529 (movdi_aarch64): Likewise.
14531 2017-05-05 Jakub Jelinek <jakub@redhat.com>
14533 PR tree-optimization/80632
14534 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
14536 (build_arrays): Initialize it for virtual phis.
14537 (fix_phi_nodes): Use it for virtual phis.
14539 PR tree-optimization/80558
14540 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
14541 [x, y] op z into [x op, y op z] for op & or | if conditions
14544 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
14545 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14548 * config/arm/arm.md (use_literal_pool): Remove.
14549 (64-bit immediate split): No longer takes cost into consideration
14550 if arm_disable_literal_pool is enabled.
14551 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
14552 used when arm_disable_literal_pool is enabled.
14553 (arm_max_const_double_inline_cost): Remove use of
14554 arm_disable_literal_pool.
14555 (push_minipool_fix): Add assert.
14556 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
14557 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
14558 (no_literal_pool_sf_immediate): New.
14560 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14562 PR tree-optimization/80613
14563 * tree-ssa-dce.c (propagate_necessity): Remove cases for
14564 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
14566 2017-05-05 Richard Biener <rguenther@suse.de>
14568 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
14570 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
14572 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
14573 of this flag from insn conditions due to removal from r247495.
14575 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
14577 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
14579 (arm_early_store_addr_dep_ptr): Likewise.
14580 * config/arm/aarch-common-protos.h
14581 (arm_early_load_addr_dep_ptr): Add prototype.
14582 (arm_early_store_addr_dep_ptr): Likewise.
14583 * config/arm/cortex-a53.md: Add new bypasses.
14585 2017-05-05 Jakub Jelinek <jakub@redhat.com>
14587 * tree.c (next_type_uid): Change type to unsigned.
14588 (type_hash_canon): Decrement back next_type_uid if
14589 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
14590 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
14593 2017-05-04 Martin Sebor <msebor@redhat.com>
14595 * builtins.c: Fix a trivial typo in a comment.
14597 PR middle-end/79234
14598 * builtins.c (check_sizes): Adjust to handle reading past the end.
14599 Avoid printing excessive upper bound of ranges. Use %E to print
14600 tree nodes instead of converting them to %wu.
14601 (expand_builtin_memchr): New function.
14602 (compute_dest_size): Rename...
14603 (compute_objsize): ...to this.
14604 (expand_builtin_memcpy): Adjust.
14605 (expand_builtin_mempcpy): Adjust.
14606 (expand_builtin_strcat): Adjust.
14607 (expand_builtin_strcpy): Adjust.
14608 (check_strncat_sizes): Adjust.
14609 (expand_builtin_strncat): Adjust.
14610 (expand_builtin_strncpy): Adjust and simplify.
14611 (expand_builtin_memset): Adjust.
14612 (expand_builtin_bzero): Adjust.
14613 (expand_builtin_memcmp): Adjust.
14614 (expand_builtin): Handle memcmp.
14615 (maybe_emit_chk_warning): Check strncat just once.
14617 2017-05-04 Martin Sebor <msebor@redhat.com>
14619 PR preprocessor/79214
14620 PR middle-end/79222
14621 PR middle-end/79223
14622 * builtins.c (check_sizes): Add inlining context and issue
14623 warnings even when -Wno-system-headers is set.
14624 (check_strncat_sizes): Same.
14625 (expand_builtin_strncat): Same.
14626 (expand_builtin_memmove): New function.
14627 (expand_builtin_stpncpy): Same.
14628 (expand_builtin): Handle memmove and stpncpy.
14630 2017-05-04 Bin Cheng <bin.cheng@arm.com>
14632 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
14633 which is not used any more.
14635 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14637 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
14639 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14641 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
14642 (cortexa53_tunings): Likewise.
14643 (cortexa57_tunings): Likewise.
14644 (cortexa72_tunings): Likewise.
14645 (cortexa73_tunings): Likewise.
14647 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14649 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
14650 Set loop alignment to 8.
14652 2017-05-04 Martin Sebor <msebor@redhat.com>
14654 PR translation/80280
14655 * builtins.c (expand_builtin_object_size): Add missing quoting to
14656 %D and like directives.
14657 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
14658 (hsa_type_for_tree_type): Same.
14659 (verify_function_arguments): Same.
14660 * symtab.c (symbol_table::change_decl_assembler_name): Same.
14661 * varasm.c (get_section): Same.
14664 2017-05-04 Martin Sebor <msebor@redhat.com>
14666 PR translation/80280
14667 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
14669 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14671 * config/aarch64/aarch64.c (generic_addrcost_table):
14672 Change HI/TI mode setting.
14674 2017-05-04 Martin Jambor <mjambor@suse.cz>
14676 PR tree-optimization/80622
14677 * tree-sra.c (comes_initialized_p): New function.
14678 (build_accesses_from_assign): Only set write lazily when
14679 comes_initialized_p is false.
14680 (analyze_access_subtree): Use comes_initialized_p.
14681 (propagate_subaccesses_across_link): Assert !comes_initialized_p
14682 instead of testing for PARM_DECL.
14684 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14686 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
14687 constraint on operand 0 to allow more general addressing modes.
14688 Adjust output template.
14689 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
14691 * config/aarch64/aarch64-protos.h
14692 (aarch64_address_valid_for_prefetch_p): Declare prototype.
14693 * config/aarch64/constraints.md (Dp): New address constraint.
14694 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
14697 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
14699 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
14700 update use of estimate_ipcp_clone_size_and_time.
14701 (estimate_local_effects): Update use of
14702 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
14703 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
14704 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
14705 Return nonspecialized time.
14707 2017-05-04 Richard Biener <rguenther@suse.de>
14709 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
14710 for the last VUSE which def dominates the PHI. Directly call
14712 (get_continuation_for_phi_1): Remove.
14714 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
14716 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
14717 to explain the use of truncating division. Cap the number of
14718 iterations to the maximum given by nb_iterations_upper_bound,
14721 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
14723 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
14724 * configure: Regenerate.
14725 * config.in: Regenerate.
14726 * config/i386/driver-mingw32.c: new file.
14727 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
14728 * config.host: Link driver-mingw32.o on MinGW host.
14729 * doc/install.texi: Document new --enable-mingw-wildcard configure
14732 2017-05-04 Marek Polacek <polacek@redhat.com>
14734 PR tree-optimization/80612
14735 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
14737 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14738 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
14740 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
14741 (movt splitter): Likewise.
14742 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
14743 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
14744 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
14745 block for Thumb-1 with MOVT.
14746 (thumb2_legitimate_address_p): Move code block ...
14747 (can_avoid_literal_pool_for_label_p): ... into this new function.
14748 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
14750 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
14751 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
14752 "M-profile targets with the MOVT instruction".
14754 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14756 * config/arm/arm-builtins.c (arm_init_builtins): Rename
14757 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
14758 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
14760 2017-05-04 Martin Liska <mliska@suse.cz>
14762 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
14763 variable cond_code.
14765 2017-05-04 Richard Biener <rguenther@suse.de>
14767 * tree.c (array_at_struct_end_p): Handle arrays at struct
14768 end with flexarrays more conservatively. Refactor and treat
14769 arrays of arrays or aggregates more strict. Fix
14770 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
14771 * tree.c (array_at_struct_end_p): Adjust prototype.
14772 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
14773 * gimple-fold.c (get_range_strlen): Likewise.
14774 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
14776 2017-05-04 Richard Biener <rguenther@suse.de>
14778 PR tree-optimization/31130
14779 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
14781 (supports_overflow_infinity): Likewise.
14782 (is_negative_overflow_infinity): Likewise.
14783 (is_positive_overflow_infinity): Likewise.
14784 (is_overflow_infinity): Likewise.
14785 (stmt_overflow_infinity): Likewise.
14786 (overflow_infinity_range_p): Likewise.
14787 (usable_range_p): Remove as always returning true.
14788 (make_overflow_infinity): Remove.
14789 (negative_overflow_infinity): Likewise.
14790 (positive_overflow_infinity): Likewise.
14791 (avoid_overflow_infinity): Likewise.
14792 (set_value_range): Adjust accordingly.
14793 (set_value_range_to_nonnegative): Likewise, remove now unused
14794 overflow_infinity arg.
14795 (vrp_operand_equal_p): Adjust.
14796 (update_value_range): Likewise.
14797 (range_int_cst_singleton_p): Likewise.
14798 (operand_less_p): Likewise.
14799 (compare_values_warnv): Likewise.
14800 (extract_range_for_var_from_comparison_expr): Likewise.
14801 (vrp_int_const_binop): Likewise.
14802 (zero_nonzero_bits_from_vr): Likewise.
14803 (extract_range_from_multiplicative_op_1): Likewise.
14804 (extract_range_from_binary_expr_1): Likewise.
14805 (extract_range_from_unary_expr): Likewise.
14806 (extract_range_from_comparison): Likewise.
14807 (extract_range_basic): Likewise.
14808 (adjust_range_with_scev): Likewise.
14809 (compare_ranges): Likewise.
14810 (compare_range_with_value): Likewise.
14811 (dump_value_range): Likewise.
14812 (test_for_singularity): Likewise, remove strict_overflow_p parameter
14814 (simplify_cond_using_ranges): Adjust.
14816 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
14818 * brig-builtins.def: Added a builtin for class_f64.
14819 * builtin-types.def: Added a builtin type needed by class_f64.
14821 2017-05-03 Jason Merrill <jason@redhat.com>
14823 * timevar.def: Add TV_CONSTEXPR.
14825 2017-05-03 David Malcolm <dmalcolm@redhat.com>
14827 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
14829 2017-05-03 Martin Jambor <mjambor@suse.cz>
14831 * ipa-prop.c (ipa_update_after_lto_read): Removed.
14832 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
14833 * ipa-cp.c (ipcp_propagate_stage): Do not call
14834 ipa_update_after_lto_read.
14835 * ipa-inline.c (ipa_inline): Likewise.
14837 2017-05-03 Martin Jambor <mjambor@suse.cz>
14839 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
14840 tag. Added a default constructor and a destructor.
14841 (ipa_edge_args_sum_t): New class;
14842 (ipa_edge_args_sum): Declare.
14843 (ipa_edge_args_vector): Remove declaration.
14844 (IPA_EDGE_REF): Use ipa_edge_args_sum.
14845 (ipa_free_edge_args_substructures): Remove declaration.
14846 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
14847 (ipa_edge_args_info_available_for_edge_p): Likewise.
14848 * ipa-prop.c (ipa_edge_args_vector): Removed.
14849 (edge_removal_hook_holder): Likewise.
14850 (edge_duplication_hook_holder): Likewise.
14851 (ipa_edge_args_sum): New variable.
14852 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
14853 ipa_edge_args_vector.
14854 (ipa_free_edge_args_substructures): Likewise.
14855 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
14856 ipa_edge_args_vector.
14857 (ipa_edge_removal_hook): Turned into method
14858 ipa_edge_args_sum_t::remove.
14859 (ipa_edge_duplication_hook): Turned into method
14860 ipa_edge_args_sum_t::duplicate.
14861 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
14862 registering edge hooks.
14863 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
14864 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
14865 ipa_edge_args_sum instead of ipa_edge_args_vector.
14866 * ipa-profile.c (ipa_profile): Likewise.
14868 2017-05-03 Martin Jambor <mjambor@suse.cz>
14870 * symbol-summary.h (function_summary): New method exists.
14871 (function_summary::symtab_removal): Deallocate through release.
14872 (call_summary): New class.
14873 (gt_ggc_mx): New overload.
14874 (gt_pch_nx): Likewise.
14875 (gt_pch_nx): Likewise.
14877 2017-05-03 Jeff Law <law@redhat.com>
14879 PR tree-optimization/78496
14880 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
14881 from simplify_cond_using_ranges. Split off code to walk
14882 backwards through casts into ...
14883 (simplify_cond_using_ranges_2): New function.
14884 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
14885 (execute_vrp): After identifying jump threads, call
14886 simplify_cond_using_ranges_2.
14888 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
14891 * ipa-inline.h (inline_summary): Add ctor.
14892 (create_ggc): Do not use ggc_cleared_alloc.
14894 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
14895 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14897 * gcc.c (handle_braces): Support escaping in switch matching
14899 * doc/invoke.texi (Spec Files): Document it.
14900 Remove superfluous @code markup in items.
14902 2017-05-03 David Malcolm <dmalcolm@redhat.com>
14904 * diagnostic-show-locus.c (struct column_range): New struct.
14905 (get_affected_columns): New function.
14906 (get_printed_columns): New function.
14907 (struct correction): New struct.
14908 (correction::ensure_capacity): New function.
14909 (correction::ensure_terminated): New function.
14910 (struct line_corrections): New struct.
14911 (line_corrections::~line_corrections): New dtor.
14912 (line_corrections::add_hint): New function.
14913 (layout::print_trailing_fixits): Reimplement in terms of the new
14915 (selftest::test_overlapped_fixit_printing): New function.
14916 (selftest::diagnostic_show_locus_c_tests): Call it.
14918 2017-05-03 Nathan Sidwell <nathan@acm.org>
14920 Canonicalize canonical type hashing
14921 * tree.h (type_hash_canon_hash): Declare.
14922 * tree.c (type_hash_list, attribute_hash_list): Move into
14923 type_hash_canon_hash.
14924 (build_type_attribute_qual_variant): Break out hash code calc into
14925 type_hash_canon_hash.
14926 (type_hash_canon_hash): New. Generic type hash computation.
14927 (build_range_type_1, build_array_type_1, build_function_type,
14928 build_method_type_directly, build_offset_type, build_complex_type,
14929 make_vector_type): Call it.
14931 2017-05-03 Richard Biener <rguenther@suse.de>
14933 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
14934 When all DRs have unknown misaligned do not always peel
14935 when there is a store but apply the same costing model as if
14936 there were only loads.
14938 2017-05-03 Richard Biener <rguenther@suse.de>
14941 PR tree-optimization/80492
14942 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
14943 compare_base_decls returning dont-know properly.
14945 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
14947 * config/arm/iterators.md (CCSI): New mode iterator.
14948 (arch): New mode attribute.
14949 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
14950 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
14951 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
14952 code iterator for success result mode.
14953 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
14954 the corresponding new insn generators.
14956 2017-05-03 Bin Cheng <bin.cheng@arm.com>
14959 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14960 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
14962 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
14964 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
14965 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
14966 (DDR_A): Wrap DDR argument in brackets.
14967 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
14968 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
14969 (DDR_REVERSED_P): Likewise.
14971 2017-05-03 Jakub Jelinek <jakub@redhat.com>
14973 PR tree-optimization/79472
14974 * tree-switch-conversion.c (struct switch_conv_info): Add
14975 contiguous_range and default_case_nonstandard fields.
14976 (collect_switch_conv_info): Compute contiguous_range and
14977 default_case_nonstandard fields, don't clear final_bb if
14978 contiguous_range and only the default case doesn't have the required
14980 (check_all_empty_except_final): Set default_case_nonstandard instead
14981 of failing if contiguous_range and the default case doesn't have empty
14983 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
14984 and only the default case doesn't have the required constants. Skip
14986 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
14987 if default_case_nonstandard.
14988 (build_constructors): Build constant 1 just once. Assert that default
14989 values aren't inserted in between cases if contiguous_range. Skip
14991 (build_arrays): Skip virtual phis.
14992 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
14993 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
14994 Handle virtual phis.
14995 (gen_inbound_check): Handle default_case_nonstandard case.
14996 (process_switch): Adjust check_final_bb caller. Call
14997 gather_default_values with the first non-default case instead of
14998 default case if default_case_nonstandard.
15000 2017-05-02 Nathan Sidwell <nathan@acm.org>
15002 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
15003 check. Fix formatting.
15005 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
15007 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
15008 errors when comparing specialized and unspecialized times.
15010 2017-05-02 David Malcolm <dmalcolm@redhat.com>
15012 * diagnostic-show-locus.c
15013 (layout::should_print_annotation_line_p): Make private.
15014 (layout::print_annotation_line): Make private.
15015 (layout::annotation_line_showed_range_p): Make private.
15016 (layout::show_ruler): Make private.
15017 (layout::print_source_line): Make private. Pass in line and
15018 line_width, rather than calling location_get_source_line. Drop
15020 (layout::print_leading_fixits): New method.
15021 (layout::print_any_fixits): Rename to...
15022 (layout::print_trailing_fixits): ...this, and make private.
15023 Don't print newline fixits.
15024 (diagnostic_show_locus): Move logic for printing one row into...
15025 (layout::print_line): ...this new function. Move the
15026 location_get_source_line call and error-handling from
15027 print_source_line to here. Call print_leading_fixits, and rename
15028 print_any_fixits to print_trailing_fixits.
15029 (selftest::test_fixit_insert_containing_newline): Update now that
15030 newlines are partially supported.
15031 (selftest::test_fixit_insert_containing_newline_2): New test.
15032 (selftest::test_fixit_replace_containing_newline): Update comments.
15033 (selftest::diagnostic_show_locus_c_tests): Call the new test.
15034 * edit-context.c (class added_line): New class.
15035 (class edited_line): Describe newline handling in comment.
15036 (edited_line::actually_edited_p): New method.
15037 (edited_line::print_content): Delete redundant decl.
15038 (edited_line::m_predecessors): New field.
15039 (edited_file::print_content): Call edited_line::print_content.
15040 (edited_file::print_diff): Update to support newlines.
15041 (edited_file::print_diff_hunk): Likewise.
15042 (edited_file::print_run_of_changed_lines): New function.
15043 (edited_file::print_diff_line): Convert to...
15044 (print_diff_line): ...this.
15045 (edited_file::get_effective_line_count): New function.
15046 (edited_line::edited_line): Initialize new field m_predecessors.
15047 (edited_line::~edited_line): Clean up m_predecessors.
15048 (edited_line::apply_fixit): Handle newlines.
15049 (edited_line::get_effective_line_count): New function.
15050 (edited_line::print_content): New function.
15051 (edited_line::print_diff_lines): New function.
15052 (selftest::test_applying_fixits_insert_containing_newline): New
15054 (selftest::test_applying_fixits_replace_containing_newline): New
15056 (selftest::insert_line): New function.
15057 (selftest::test_applying_fixits_multiple_lines): Add example of
15059 (selftest::edit_context_c_tests): Call the new tests.
15061 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15063 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
15064 parameter cand. Update dump information.
15065 (get_computation_cost): Update uses.
15067 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15069 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
15070 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
15071 (get_computation_at, rewrite_use_address): Update use of
15072 get_computation_aff.
15074 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15076 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
15077 (get_computation): Delete.
15078 (get_computation_cost): Implement like get_computation_cost_at.
15079 Use get_computation_at.
15080 (get_computation_cost_at): Delete.
15081 (rewrite_use_nonlinear_expr): Use get_computation_at.
15082 (rewrite_use_compare, remove_unused_ivs): Ditto.
15084 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15086 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
15088 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15090 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
15091 (ivopts_global_cost_for_size): Rename parameter and update uses.
15092 (iv_ca_recount_cost): Update uses.
15093 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
15094 candidates seperately in n_invs and n_cands.
15095 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
15097 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15099 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
15100 (find_inv_vars_cb): New.
15101 (find_depends): Renamed to ...
15102 (find_inv_vars): ... this.
15103 (add_candidate_1, force_var_cost): Call find_inv_vars.
15104 (split_address_cost, determine_group_iv_cost_cond): Ditto.
15106 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15108 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
15109 inv_vars. Add inv_exprs.
15110 (struct iv_cand): Rename depends_on to inv_vars.
15111 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
15112 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
15113 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
15114 (dump_cand): Dump inv_vars.
15115 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
15116 (record_invariant, find_depends, add_candidate_1): Ditto.
15117 (set_group_iv_cost, force_var_cost): Ditto.
15118 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
15119 (get_computation_cost_at, get_computation_cost): Ditto.
15120 (determine_group_iv_cost_generic): Ditto.
15121 (determine_group_iv_cost_address): Ditto.
15122 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
15123 (determine_group_iv_costs): Ditto.
15124 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
15125 (iv_ca_set_remove_invariants): Renamed to ...
15126 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
15127 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
15128 (iv_ca_set_add_invariants): Renamed to ...
15129 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
15130 (iv_ca_set_cp): Use iv_ca_set_add_invs.
15131 (iv_ca_has_deps): Support inv_vars and inv_exprs.
15132 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
15133 (create_new_ivs): Remove useless dump.
15135 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15137 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
15139 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
15140 (iv_ca_set_no_cp, create_new_iv): Ditto.
15142 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15144 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
15146 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
15148 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
15149 function tree_check2.
15151 2017-05-02 Martin Liska <mliska@suse.cz>
15153 * doc/gcov.texi: Add missing preposition.
15154 * gcov.c (function_info::function_info): Properly fill up
15155 all member variables.
15157 2017-05-02 Tamar Christina <tamar.christina@arm.com>
15159 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
15161 2017-05-02 Tamar Christina <tamar.christina@arm.com>
15163 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
15165 2017-05-02 Martin Liska <mliska@suse.cz>
15168 * lto-streamer-in.c (lto_read_tree_1): Remove
15169 LTO_STREAMER_DEBUG.
15170 * lto-streamer.c (struct tree_hash_entry): Likewise.
15171 (struct tree_entry_hasher): Likewise.
15172 (tree_entry_hasher::hash): Likewise.
15173 (tree_entry_hasher::equal): Likewise.
15174 (lto_streamer_init): Likewise.
15175 (lto_orig_address_map): Likewise.
15176 (lto_orig_address_get): Likewise.
15177 (lto_orig_address_remove): Likewise.
15178 * lto-streamer.h: Likewise.
15179 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
15180 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
15182 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
15184 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
15185 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
15186 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
15187 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
15188 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
15189 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
15190 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
15191 (mm_maskz_sub_ss): New intrinsics.
15192 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15193 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
15194 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
15195 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
15196 (__builtin_ia32_subss_mask_round): New builtins.
15197 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15198 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15199 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
15201 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
15202 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
15204 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
15207 2017-05-02 Martin Jambor <mjambor@suse.cz>
15209 PR tree-optimization/78687
15210 * tree-sra.c (access): New field parent.
15211 (process_subtree_disqualification): New function.
15212 (disqualify_candidate): Call it.
15213 (build_accesses_from_assign): Reset write flag if creating an
15215 (build_access_subtree): Fill in parent field and also prpagate
15216 down grp_write flag.
15217 (create_artificial_child_access): New parameter set_grp_write, set
15218 grp_write to its value.
15219 (propagate_subaccesses_across_link): Also propagate grp_write flag
15221 (propagate_all_subaccesses): Push the closest parent back to work
15222 queue if add_access_to_work_queue returned true.
15224 2017-05-02 Richard Biener <rguenther@suse.de>
15226 * common.opt (fstrict-overflow): Alias negative to fwrapv.
15227 * doc/invoke.texi (fstrict-overflow): Remove all traces of
15228 -fstrict-overflow documentation.
15229 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
15230 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
15231 flag_strict_overflow.
15232 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
15233 * lto-opts.c (lto_write_options): Do not stream it.
15234 * lto-wrapper.c (merge_and_complain): Do not handle it.
15235 * opts.c (default_options_table): Do not set -fstrict-overflow.
15236 (finish_options): Likewise do not clear it when sanitizing.
15237 * simplify-rtx.c (simplify_const_relational_operation): Do not
15238 test flag_strict_overflow.
15240 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
15242 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
15243 using enabled attribute.
15244 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
15245 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
15246 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
15247 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
15248 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
15249 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
15250 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
15251 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
15252 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
15253 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
15255 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
15257 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
15259 2017-05-02 Richard Biener <rguenther@suse.de>
15261 PR tree-optimization/80591
15263 2017-04-10 Richard Biener <rguenther@suse.de>
15265 * tree-ssa-structalias.c (find_func_aliases): Properly handle
15268 2017-05-02 Richard Biener <rguenther@suse.de>
15270 PR tree-optimization/80549
15271 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
15272 (cleanup_tree_cfg_noloop): Create forwarders to known loop
15273 headers if they do not have a preheader.
15275 2017-05-02 Martin Liska <mliska@suse.cz>
15278 * common.opt: Fix typo.
15279 * doc/invoke.texi: Likewise.
15281 2017-05-01 Jan Beulich <jbeulich@suse.com>
15283 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
15284 swapping, add (x,x,m,x,n) alternative.
15286 2017-05-01 Nathan Sidwell <nathan@acm.org>
15288 * calls.c (combine_pending_stack_adjustment_and_call): Remove
15289 unnecessary unadjusted_alignment check.
15291 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
15294 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
15296 * gimplify.c (gimplify_cilk_detach): New function.
15297 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
15298 * tree-core.h: Document EXPR_CILK_SPAWN.
15299 * tree.h (EXPR_CILK_SPAWN): Define.
15301 2017-05-01 David Malcolm <dmalcolm@redhat.com>
15303 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
15304 to use new fixit_hint representation, using the "replace" logic.
15305 (get_line_span_for_fixit_hint): Likewise.
15306 (layout::print_any_fixits): Likewise.
15307 (selftest::test_one_liner_many_fixits): Rename to...
15308 (selftest::test_one_liner_many_fixits_1): ...this, and update
15309 comment and expected output to reflect that the multiple fix-it
15310 hints are now consolidated into one insertion.
15311 (selftest::test_one_liner_many_fixits_2): New test.
15312 (selftest::test_diagnostic_show_locus_one_liner): Update for
15314 (selftest::test_fixit_consolidation): Update for fix-it API
15316 * diagnostic.c (print_parseable_fixits): Likewise.
15317 * edit-context.c (edited_line::m_line_events): Convert from
15318 auto_vec <line_event *> to auto_vec <line_event>.
15319 (class line_event): Convert from abstract base class to a concrete
15320 class, taking over the role of replace_event.
15321 (class insert_event): Delete.
15322 (class replace_event): Rename to class line_event. Convert to
15324 (edit_context::add_fixits): Reimplement.
15325 (edit_context::apply_insert): Delete.
15326 (edit_context::apply_replace): Rename to...
15327 (edit_context::apply_fixit): ...this. Convert to half-open range.
15328 (edited_file::apply_insert): Delete.
15329 (edited_file::apply_replace): Rename to...
15330 (edited_file::apply_fixit): ...this.
15331 (edited_line::~edited_line): Drop deletion of events.
15332 (edited_line::apply_insert): Delete.
15333 (edited_line::apply_replace): Rename to...
15334 (edited_line::apply_fixit): ...this. Convert to half-open range.
15335 Update for change to type of m_line_events.
15336 * edit-context.h (edit_context::apply_insert): Delete.
15337 (edit_context::apply_replace): Rename to...
15338 (edit_context::apply_fixit): ...this.
15340 2017-05-01 Martin Sebor <msebor@redhat.com>
15342 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
15345 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
15348 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
15349 __get_cpuid_max returns 0.
15350 (__get_cpuid_count): Ditto.
15352 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
15354 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
15355 replacement expression is another instance of one of its arguments.
15357 2017-05-01 Jakub Jelinek <jakub@redhat.com>
15360 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
15361 check for stack push/pop autoinc.
15362 * config/i386/i386.c (ix86_agi_dependent): Return false
15363 if the only reason why modified_in_p returned true is that
15364 addr is SP based and set_insn is a push or pop.
15366 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
15368 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
15371 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
15374 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
15375 (account_size_time): Use two predicates - exec_pred and
15377 (evaluate_conditions_for_known_args): Compute both clause and
15379 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
15380 (inline_summary_t::duplicate): Update.
15381 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
15383 (compute_inline_parameters): Likewise.
15384 (estimate_edge_size_and_time): Update caluclation of time.
15385 (estimate_node_size_and_time): Compute both time and nonspecialized
15387 (estimate_ipcp_clone_size_and_time): Update.
15388 (inline_merge_summary): Update.
15389 (do_estimate_edge_time): Update.
15390 (do_estimate_edge_size): Update.
15391 (do_estimate_edge_hints): Update.
15392 (inline_read_section, inline_write_summary): Stream both new predicates.
15393 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
15395 (compute_inlined_call_time): Cleanup.
15396 (big_speedup_p): Update.
15397 (edge_badness): Update.
15398 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
15399 (size_time_entry): Replace predicate by exec_predicate and
15400 nonconst_predicate.
15401 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
15402 (estimate_edge_time): Return also nonspec_time.
15403 (reset_edge_growth_cache): Update.
15405 2017-04-29 Jakub Jelinek <jakub@redhat.com>
15407 PR rtl-optimization/80491
15408 * ifcvt.c (noce_process_if_block): When looking for x setter
15409 with missing else_bb, don't check only the insn right before
15410 cond_earliest, but look for the last insn that x is modified in
15411 within the same bb.
15413 PR rtl-optimization/80491
15414 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
15416 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
15418 PR tree-optimization/80487
15419 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
15421 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15423 PR tree-optimization/79697
15424 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
15425 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
15426 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
15428 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
15429 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
15431 2017-04-28 Martin Sebor <msebor@redhat.com>
15433 PR tree-optimization/80523
15434 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
15435 (init_target_to_host_charmap, target_to_host, target_strtol10): New
15437 (maybe_warn, format_directive, parse_directive): Use new functions.
15438 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
15440 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
15442 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
15444 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
15446 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
15447 target_header_dir): Set correctly.
15448 * configure: Regenerated.
15449 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
15450 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
15451 instead of SYSTEM_HEADER_DIR.
15453 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
15455 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
15456 (estimate_local_effects): Likewise.
15457 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
15458 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
15459 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
15460 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
15461 do_estimate_edge_time, estimate_edge_time): Likewise.
15462 * ipa-inline-analysis.c (estimate_node_size_and_time,
15463 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
15464 (estimate_time_after_inlining): Remove.
15466 2017-04-28 Martin Liska <mliska@suse.cz>
15468 * doc/gcov.texi: Enhance documentation of gcov.
15470 2017-04-28 Martin Liska <mliska@suse.cz>
15472 * doc/gcov.texi: Sort options in alphabetic order.
15473 * doc/gcov-dump.texi: Likewise.
15474 * doc/gcov-tool.texi: Likewise.
15475 * gcov.c (print_usage): Likewise.
15476 * gcov-dump.c (print_usage): Likewise.
15477 * gcov-tool.c (print_merge_usage_message): Likewise.
15478 (print_rewrite_usage_message): Likewise.
15479 (print_overlap_usage_message): Likewise.
15481 2017-04-28 Martin Liska <mliska@suse.cz>
15483 PR gcov-profile/53915
15484 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
15486 2017-04-28 Martin Liska <mliska@suse.cz>
15488 PR gcov-profile/79891
15489 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
15490 is marked by compiler as living on a line.
15491 (get_cycles_count): Remove usage of the union.
15492 (output_intermediate_file): Likewise.
15493 (find_source): Fix GNU coding style.
15494 (accumulate_line_counts): Remove old non-all block mode.
15495 (output_lines): Remove usage of the union.
15496 * profile.c (output_location): Include all BBs, even if
15497 belonging to a same line (and file) as a previous BB.
15499 2017-04-28 Martin Liska <mliska@suse.cz>
15501 * gcov.c (process_args): Handle new argument 'w'.
15502 (read_graph_file): Assign ID to BBs.
15503 (output_branch_count): Display BB # if verbose flag is set.
15504 (output_lines): Likewise for arcs.
15505 (print_usage): Add '--verbose' option help.
15506 * doc/gcov.texi: Document --verbose (-w) option.
15508 2017-04-28 Martin Liska <mliska@suse.cz>
15510 * gcov.c (struct block_location_info): New struct.
15511 (process_file): Fill up the new structure.
15512 (read_graph_file): Replace usage of encoding by the newly added
15514 (add_line_counts): Likewise.
15515 (accumulate_line_counts): Remove usage of the union.
15516 (function_info::function_info): New function.
15517 (function_info::~function_info): Likewise.
15518 (process_file): Call delete instead of release_function.
15519 (release_function): Release the function.
15520 (release_structures): Call delete instead of release_function.
15521 (solve_flow_graph): Replace usage of num_blocks.
15522 (find_exception_blocks): Likewise.
15523 (output_lines): Fix GNU coding style.
15525 2017-04-28 Martin Liska <mliska@suse.cz>
15528 * coverage.c (coverage_remove_note_file): New function.
15529 * coverage.h: Declare the function.
15530 * toplev.c (finalize): Clean if an error has been seen.
15532 2017-04-28 Martin Liska <mliska@suse.cz>
15534 PR gcov-profile/80031
15535 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
15536 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
15537 * gcov.c (read_graph_file): Read just number of blocks.
15538 * profile.c (branch_prob): Do not stream 0 flags per a basic
15541 2017-04-28 Martin Liska <mliska@suse.cz>
15543 * gcov-dump.c (tag_*): Add new argument to declarations.
15544 (dump_gcov_file): Likewise.
15545 (tag_blocks): Add and use new argument depth.
15546 (tag_arcs): Likewise.
15547 (tag_lines): Likewise.
15548 (tag_counters): Likewise.
15549 (tag_summary): Likewise.
15550 (dump_working_sets): Use depth to do a proper indentation.
15552 2017-04-28 Jakub Jelinek <jakub@redhat.com>
15555 * cgraph.h (symtab_node::debug_symtab): No longer inline.
15556 * symtab.c (symtab_node::debug_symtab): Move definition here.
15558 2017-04-28 Richard Biener <rguenther@suse.de>
15560 * lto-streamer.h (LTO_major_version): Bump to 7.
15562 2017-04-28 Richard Biener <rguenther@suse.de>
15564 * tree-vrp.c (assert_info): New struct.
15565 (add_assert_info): New helper.
15566 (register_edge_assert_for_2): Refactor to add asserts to a vector
15568 (register_edge_assert_for_1): Likewise.
15569 (register_edge_assert_for): Likewise.
15570 (finish_register_edge_assert_for): New helper actually registering
15571 asserts where live on edge.
15572 (find_conditional_asserts): Adjust.
15573 (find_switch_asserts): Likewise.
15574 (evrp_dom_walker::try_find_new_range): Generalize.
15575 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
15577 2017-04-27 Marek Polacek <polacek@redhat.com>
15580 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
15581 arg10 and arg11 to itype.
15583 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
15585 * doc/extend.texi (Object Size Checking): Improve grammar.
15587 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
15590 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
15591 that the logic for permitting reciprocal estimates matches that
15594 2017-04-27 Jakub Jelinek <jakub@redhat.com>
15597 * tree.c (type_cache_hasher::equal): Only compare
15598 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
15599 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
15600 non-aggregate element types.
15601 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
15602 about the flag on ARRAY_TYPEs in the comment, formatting fix.
15604 2017-04-27 Richard Biener <rguenther@suse.de>
15606 PR middle-end/80533
15607 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
15608 stripping ARRAY_REFs from MEM_EXPR make sure we're not
15609 keeping a reference to a trailing array.
15611 2017-04-27 Richard Biener <rguenther@suse.de>
15613 PR middle-end/80539
15614 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
15615 being in loop-closed SSA form conservatively.
15616 (chrec_fold_multiply_poly_poly): Likewise.
15618 2017-04-27 Tamar Christina <tamar.christina@arm.com>
15620 PR middle-end/79665
15621 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
15622 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
15624 2017-04-27 Jakub Jelinek <jakub@redhat.com>
15627 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
15628 (aarch64_function_arg_alignment): Return unsigned int again, but still
15629 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
15630 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
15631 Don't emit -Wpsabi note.
15632 (aarch64_function_arg_boundary): Likewise.
15633 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
15636 2017-04-26 Nathan Sidwell <nathan@acm.org>
15638 * tree.h (crc32_unsigned_n): Declare.
15639 (crc32_unsigned, crc32_unsigned): Make inline.
15640 * tree.c (crc32_unsigned_bits): Replace with ...
15641 (crc32_unsigned_n): ... this.
15642 (crc32_unsigned, crc32_byte): Remove.
15643 (crc32_string): Remove unnecessary braces.
15645 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
15647 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
15648 * ipa-inline-analysis.c (MAX_TIME): Remove.
15649 (account_size_time): Use sreal for time.
15650 (dump_inline_summary): Update.
15651 (estimate_function_body_sizes): Update.
15652 (estimate_edge_size_and_time): Update.
15653 (estimate_calls_size_and_time): Update.
15654 (estimate_node_size_and_time): Update.
15655 (inline_merge_summary): Update.
15656 (inline_update_overall_summary): Update.
15657 (estimate_time_after_inlining): Update.
15658 (inline_read_section): Update.
15659 (inline_write_summary): Update.
15660 * ipa-inline.c (compute_uninlined_call_time): Update.
15661 (compute_inlined_call_time): Update.
15662 (recursive_inlining): Update.
15663 (inline_small_functions): Update.
15664 (dump_overall_stats): Update.
15665 * ipa-inline.h: Include sreal.h.
15666 (size_time_entry): Turn time to sreal.
15667 (inline_summary): Turn self_time nad time to sreal.
15669 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
15671 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
15673 (sreal::stream_out, sreal::stream_in): New.
15674 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
15676 2017-04-25 Jakub Jelinek <jakub@redhat.com>
15678 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
15681 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
15684 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
15686 (dimode_scalar_chain::compute_convert_gain): Ditto.
15687 (dimode_scalar_chain::make_vector_copies): Ditto.
15688 (dimode_scalar_chain::convert_reg): Ditto.
15689 (dimode_scalar_chain::convert_insn): Ditto.
15690 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
15691 (VI248_AVX512BW_1): New mode iterator.
15692 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
15693 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
15696 2017-04-25 Martin Sebor <msebor@redhat.com>
15698 PR tree-optimization/80497
15699 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
15700 constants are representable in HOST_WIDE_INT.
15701 (parse_directive): Ditto.
15703 2017-04-25 Martin Sebor <msebor@redhat.com>
15706 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
15707 (new_zero_array): Adjust signature.
15708 (dom_info::dom_init): Used unsigned rather that size_t.
15709 (dom_info::dom_info): Same.
15711 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15712 Jakub Jelinek <jakub@redhat.com>
15715 * config/arm/arm.c: Include gimple.h.
15716 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
15717 returns negative, increment ncrn only if it returned positive.
15718 (arm_needs_doubleword_align): Return int instead of bool,
15719 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
15720 members, but if there is any such non-FIELD_DECL
15721 > PARM_BOUNDARY aligned decl, return -1 instead of false.
15722 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
15723 returns negative, increment nregs only if it returned positive.
15724 (arm_setup_incoming_varargs): Likewise.
15725 (arm_function_arg_boundary): Emit -Wpsabi note if
15726 arm_needs_doubleword_align returns negative, return
15727 DOUBLEWORD_ALIGNMENT only if it returned positive.
15729 2017-04-25 Marek Polacek <polacek@redhat.com>
15732 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
15733 first argument to type.
15735 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
15738 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
15739 type checks to test for compatibility instead of equality.
15741 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15742 Jakub Jelinek <jakub@redhat.com>
15745 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
15747 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
15748 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
15749 the alignment computation, but return their maximum in warn_alignment.
15750 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
15751 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
15753 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
15754 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
15757 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15759 * config/arc/simdext.md (dmpyh): Fix typo.
15761 2017-04-25 Richard Biener <rguenther@suse.de>
15763 PR tree-optimization/80492
15764 * alias.c (compare_base_decls): Handle registers with asm
15765 specification conservatively.
15766 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
15767 compare_base_decls returning dont-know properly.
15769 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15771 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
15772 (legitimate_offset_address_p): New function.
15773 (arc_legitimate_address_p): Use above function.
15775 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15777 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
15779 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15781 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
15782 ACCH registers whenever they are available.
15784 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15786 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
15787 double regs fix when not used.
15789 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15791 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
15793 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
15794 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
15796 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15798 * config/arc/arc.c (arc_output_addsi): Check for h-register class
15799 when emitting short ADD instructions.
15801 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15803 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
15805 (cmpsi_cc_c_insn): Likewise.
15806 (cbranchsi4_scratch): Compute proper instruction length using
15807 compact_hreg_operand.
15808 * config/arc/predicates.md (compact_hreg_operand): New predicate.
15810 2017-04-25 Richard Biener <rguenther@suse.de>
15812 PR middle-end/80509
15813 * passes.c (pass_manager::pass_manager): Initialize
15814 m_name_to_pass_map.
15816 2017-04-25 Richard Biener <rguenther@suse.de>
15818 PR tree-optimization/79201
15819 * tree-ssa-sink.c (statement_sink_location): Handle calls.
15821 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15824 * config/s390/vector.md: Split MEM->GPR vector moves for
15825 non-s_operand addresses.
15827 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15830 * config/s390/predicates.md (reload_const_wide_int_operand): New
15832 * config/s390/s390.md ("movti"): Remove d/P alternative.
15833 ("movti_bigconst"): New pattern definition.
15835 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15838 * s390-protos.h (s390_expand_cs_hqi): Removed.
15839 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
15840 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
15841 modes as well as CCZ1mode and CCZmode.
15842 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
15843 signature of s390_emit_compare_and_swap.
15844 (s390_expand_cs_hqi): Likewise, make static.
15845 (s390_expand_cs_tdsi): Generate an explicit compare before trying
15846 compare-and-swap, in some cases.
15847 (s390_expand_cs): Wrapper function.
15848 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
15850 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
15851 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
15852 patterns for small and large integers. Forbid symref memory operands.
15853 Move expander to s390.c. Require cc register.
15854 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
15855 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
15856 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
15857 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
15858 symref memory operands. Remove CC mode and call s390_match_ccmode
15860 ("atomic_exchange<mode>"): Allow and implement all integer modes.
15862 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15864 * config/s390/s390.md (define_peephole2): New peephole to help
15865 combining the load-and-test pattern with volatile memory.
15867 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15869 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
15870 with CCZmode for TARGET_Z196.
15872 2017-04-25 Jakub Jelinek <jakub@redhat.com>
15874 PR rtl-optimization/80501
15875 * combine.c (make_compound_operation_int): Set subreg_code to SET
15876 even for AND with mask of the sign bit of mode.
15878 PR rtl-optimization/80500
15879 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
15880 sum's initial value.
15882 2017-04-25 Julian Brown <julian@codesourcery.com>
15883 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15885 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
15887 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
15889 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
15891 2017-04-25 Julian Brown <julian@codesourcery.com>
15892 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15894 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
15895 (thunderx2t99_sha): New Reservation.
15897 2017-04-25 Julian Brown <julian@codesourcery.com>
15898 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15900 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
15901 type for 1-element load.
15903 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
15905 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
15907 2017-04-24 Martin Jambor <mjambor@suse.cz>
15909 PR tree-optimization/80293
15910 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
15911 char arrays not totally scalarizable if it is false.
15912 (analyze_all_variable_accesses): Pass correct value in the new
15913 parameter. Add a statistics counter.
15915 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
15917 PR middle-end/79931
15918 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
15920 2017-04-24 Richard Biener <rguenther@suse.de>
15922 PR tree-optimization/80494
15923 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
15924 out for complex types.
15926 2017-04-24 Richard Biener <rguenther@suse.de>
15928 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
15929 * tree-ssa-sccvn.c (print_scc): Print SCC size.
15930 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
15931 (DFS): Adjust and never fail.
15932 (sccvn_dom_walker::fail): Remove.
15933 (sccvn_dom_walker::before_dom_children): Adjust.
15934 (run_scc_vn): Likewise and never fail.
15935 * tree-ssa-pre.c (pass_pre::execute): Adjust.
15936 (pass_fre::execute): Likewise.
15938 2017-04-24 Richard Biener <rguenther@suse.de>
15940 PR tree-optimization/79725
15941 * tree-ssa-sink.c (statement_sink_location): Return whether
15942 failure reason was zero uses. Move that check later.
15943 (sink_code_in_bb): Deal with zero uses by removing the stmt
15946 2017-04-24 Richard Biener <rguenther@suse.de>
15949 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
15950 pointer-based references.
15952 2017-04-24 Richard Biener <rguenther@suse.de>
15955 * pass_manager.h (pass_manager::operator new): Remove.
15956 (pass_manager::operator delete): Likewise.
15957 * passes.c (pass_manager::operator new): Remove.
15958 (pass_manager::operator delete): Likewise.
15959 (pass_manager::pass_manager): Zero individual pass members.
15961 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
15964 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
15965 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
15966 Check "XEXP (src, 1)" operand here.
15967 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
15968 Check "XEXP (src, 1)" operand here.
15969 (dimode_scalar_chain::make_vector_copies): Detect count register
15970 of a shift instruction. Zero extend count register from QImode
15971 to DImode to satisfy vector shift pattern count operand predicate.
15972 Substitute vector shift count operand with a DImode copy.
15973 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
15976 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
15978 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
15979 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
15980 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
15981 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
15982 (UNSPEC_NOREX_MEM): Remove definition.
15984 2017-04-21 Richard Biener <rguenther@suse.de>
15986 PR tree-optimization/79547
15987 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
15988 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
15989 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
15990 without any constraints.
15992 2017-04-21 Richard Biener <rguenther@suse.de>
15994 PR tree-optimization/78847
15995 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
15997 2017-04-21 Richard Biener <rguenther@suse.de>
15999 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
16000 (build_distinct_type_copy): Likewise.
16001 (build_variant_type_copy): Likewise.
16002 * tree.c (build_qualified_type): Pass down mem-stat info.
16003 (build_distinct_type_copy): Likewise.
16004 (build_variant_type_copy): Likewise.
16006 2017-04-21 Richard Biener <rguenther@suse.de>
16008 PR tree-optimization/80237
16009 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
16011 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
16012 for a simplified result.
16014 2016-04-21 Richard Biener <rguenther@suse.de>
16016 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
16017 sth as strict as a simple_iv but a chrec without symbols and an
16018 operand defined in the loop we are peeling (and not some subloop).
16019 (propagate_constants_for_unrolling): Propagate all constants.
16021 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
16024 * config/i386/i386.c (print_reg): Remove assert for disalowed
16025 regno values, call output_operand_lossage instead.
16027 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
16030 * config/i386/constraints.md (Yc): New register constraint.
16031 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
16032 Use Yc constraint for alternative 2 of operand 0. Remove
16033 preferred_for_speed attribute.
16035 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
16037 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
16038 lastprivate clauses in SIMT case.
16040 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
16042 * doc/invoke.texi (-Wextra-semi): Document new warning option.
16044 2017-04-20 Richard Biener <rguenther@suse.de>
16046 PR tree-optimization/57796
16047 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
16048 as N scalar stores.
16049 (vect_model_load_cost): Cost gathers as N scalar loads.
16051 2017-04-20 Richard Biener <rguenther@suse.de>
16053 * ggc-page.c (ggc_allocated_p): Rename to ...
16054 (safe_lookup_page_table_entry): ... this and return the lookup
16056 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
16058 2017-04-20 Richard Biener <rguenther@suse.de>
16060 PR tree-optimization/80453
16061 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
16062 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
16063 from the conditions.
16064 (vn_phi_eq): Pass them down.
16065 (vn_phi_lookup): Record them.
16066 (vn_phi_insert): Likewise.
16068 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
16070 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
16071 uninitialized variable warning to avoid buffer overrun.
16073 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
16076 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
16077 is suppressed for '{ 0 }' in C.
16079 2017-04-20 Jakub Jelinek <jakub@redhat.com>
16081 * BASE-VER: Set to 8.0.0.
16083 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
16085 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
16086 priority .init_array and .fini_array section with SECTION_NOTYPE
16089 2017-04-20 Jakub Jelinek <jakub@redhat.com>
16091 PR middle-end/80423
16092 * tree.h (build_array_type): Add typeless_storage default argument.
16093 * tree.c (type_cache_hasher::equal): Also compare
16094 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
16095 (build_array_type): Add typeless_storage argument, set
16096 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
16098 (build_nonshared_array_type): Adjust build_array_type_1 caller.
16099 (build_array_type): Likewise. Add typeless_storage argument.
16101 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
16102 Jakub Jelinek <jakub@redhat.com>
16104 PR tree-optimization/80426
16105 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
16106 operation on symbolic operands, also compute the overflow for the
16107 invariant part when the operation degenerates into a negation.
16109 2017-04-19 Jakub Jelinek <jakub@redhat.com>
16112 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
16113 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
16116 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
16118 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
16121 * config/avr/avr.c (tree.h): Include it.
16122 (cgraph.h): Include it.
16123 (avr_encode_section_info): Don't warn for uninitialized progmem
16124 variable if it's just an alias.
16126 2017-04-19 Richard Biener <rguenther@suse.de>
16129 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
16130 when needed by AutoPGO.
16132 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
16135 * doc/lto.texi: Remove an extra 'that'.
16137 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
16139 PR rtl-optimization/80429
16140 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
16141 are only used in debug insns.
16143 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
16144 Vladimir Makarov <vmakarov@redhat.com>
16146 * config/sparc/predicates.md (input_operand): Add comment. Return
16147 true for any memory operand when LRA is in progress.
16148 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
16150 2017-04-18 Jeff Law <law@redhat.com>
16153 * mips.md ({return,simple_return}_internal): Do not overwrite
16156 2017-04-18 Jakub Jelinek <jakub@redhat.com>
16158 PR tree-optimization/80443
16159 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
16160 instead of adding 1, subtract -1 and similarly instead of subtracting
16163 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
16165 PR rtl-optimization/80357
16166 * haifa-sched.c (tmp_bitmap): New variable.
16167 (model_recompute): Handle duplicate use records.
16168 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
16169 (free_global_sched_pressure_data): Free it.
16171 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
16174 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
16175 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
16176 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
16177 instead of SYSTEM_HEADER_DIR.
16179 2017-04-18 Jeff Law <law@redhat.com>
16181 PR middle-end/80422
16182 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
16183 predecessors after walking up the insn chain.
16185 2017-04-18 Jakub Jelinek <jakub@redhat.com>
16188 * dwarf2out.c (modified_type_die): Try harder not to emit internal
16189 sizetype type into debug info.
16191 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
16194 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
16195 unneeded test for TARGET_UPPER_REGS_SF.
16196 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
16198 2017-04-18 Jakub Jelinek <jakub@redhat.com>
16201 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
16202 instead of gsi_after_labels.
16204 2017-04-18 Jeff Law <law@redhat.com>
16206 * regcprop.c (maybe_mode_change): Avoid creating copies of the
16210 2017-04-13 Jeff Law <law@redhat.com>
16211 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
16212 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
16214 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
16217 * config/avr/avr.c (intl.h): Include it.
16218 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
16220 2017-04-18 Martin Liska <mliska@suse.cz>
16222 PR gcov-profile/78783
16223 * gcov-tool.c (gcov_output_files): Validate that destination
16224 file is either removed by the tool or by a user.
16226 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
16227 Guy Benyei <guybe@mellanox.com>
16229 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
16230 block, and do not negate it, the stored id is already negative.
16232 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
16234 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
16236 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
16239 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
16240 masks of options that should be turned off if the VSX vector
16241 options are turned off.
16242 (OTHER_P8_VECTOR_MASKS): Likewise.
16243 (OTHER_VSX_VECTOR_MASKS): Likewise.
16244 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
16245 rs6000_disable_incompatible_switches to validate no type switches
16247 (rs6000_incompatible_switch): New function to disallow turning on
16248 other vector options if -mno-vsx, -mno-power8-vector, or
16249 -mno-power9-vector are specified.
16251 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
16253 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
16255 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
16257 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
16258 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
16259 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
16260 (ARG_POINTER_CFA_OFFSET): Likewise.
16262 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
16264 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
16265 conditions to take advantage of various optimizations.
16267 2017-04-13 Jeff Law <law@redhat.com>
16269 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
16270 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
16271 (zero_extendsidi2_dext): Likewise.
16273 2017-04-13 Jakub Jelinek <jakub@redhat.com>
16276 * fold-const.c (fold_ternary_loc): Revert
16277 use op0 instead of fold_convert_loc (loc, type, arg0) part of
16280 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
16282 PR rtl-optimization/80343
16283 * lra-remat.c (update_scratch_ops): Assign original hard reg to
16284 new scratch pseudo.
16286 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
16289 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
16290 to ubsan_encode_value.
16292 2017-04-13 Jeff Law <law@redhat.com>
16294 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
16295 appearing in DEBUG_INSNs.
16297 2017-04-13 Martin Liska <mliska@suse.cz>
16299 PR gcov-profile/80413
16300 * gcov-io.c (gcov_write_string): Copy to buffer just when
16301 allocated size is greater than zero.
16303 2017-04-13 Jakub Jelinek <jakub@redhat.com>
16306 * dwarf2out.c (decls_for_scope): Ignore declarations of
16307 current_function_decl in BLOCK_NONLOCALIZED_VARS.
16309 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
16312 * ipa-visibility.c (non_local_p): Fix typos.
16313 (localize_node): When localizing symbol in same comdat group,
16314 dissolve the group only when we know external symbols are going
16316 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
16318 2017-04-12 Jakub Jelinek <jakub@redhat.com>
16320 PR tree-optimization/79390
16321 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
16322 order does not result in usable sequence, retry with reversed operand
16328 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
16329 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
16330 op0 instead of fold_convert_loc (loc, type, arg0).
16332 2017-04-12 Jeff Law <law@redhat.com>
16334 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
16335 has a delay slot in the generated code.
16337 * config/cris/cris.md (cris_preferred_reload_class): Return
16338 GENNONACR_REGS rather than GENERAL_REGS.
16340 2017-04-12 Jakub Jelinek <jakub@redhat.com>
16343 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
16344 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
16345 signedness of the result type.
16347 2017-04-12 Richard Biener <rguenther@suse.de>
16348 Jeff Law <law@redhat.com>
16350 PR tree-optimization/80359
16351 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
16352 trim stores to TARGET_MEM_REFs.
16354 2017-04-12 Richard Biener <rguenther@suse.de>
16356 PR tree-optimization/79390
16357 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
16358 threading case even more.
16360 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
16363 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
16364 for quad_address_p for TImode, instead of just not indexed_address.
16366 2017-04-12 Richard Biener <rguenther@suse.de>
16367 Bernd Edlinger <bernd.edlinger@hotmail.de>
16369 PR middle-end/79671
16370 * alias.c (component_uses_parent_alias_set_from): Handle
16371 TYPE_TYPELESS_STORAGE.
16372 (get_alias_set): Likewise.
16373 * tree-core.h (tree_type_common): Add typeless_storage flag.
16374 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
16375 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
16376 for types containing members with TYPE_TYPELESS_STORAGE.
16377 (place_field): Likewise.
16378 (layout_type): Likewise for ARRAY_TYPE.
16379 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
16380 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
16381 TYPE_TYPELESS_STORAGE.
16382 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
16384 2017-04-12 Jakub Jelinek <jakub@redhat.com>
16387 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
16388 first argument to type.
16390 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16394 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
16395 CONST0_RTX (mode) rather than const0_rtx where appropriate.
16396 (rs6000_expand_binop_builtin): Likewise.
16397 (rs6000_expand_ternop_builtin): Likewise; also add missing
16398 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
16399 vshasigma built-ins.
16400 * doc/extend.texi: Document that vec_xxpermdi's third argument
16401 must be a constant.
16403 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
16405 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
16406 Use shift_const cost parameter when calculating gain of STV shifts.
16408 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
16410 PR rtl-optimization/70478
16411 * lra-constraints.c (process_alt_operands): Check memory for
16412 disfavoring memory insn operand.
16414 2017-04-11 Jakub Jelinek <jakub@redhat.com>
16416 PR middle-end/80100
16417 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
16418 left shift in unsigned HOST_WIDE_INT type.
16420 PR rtl-optimization/80385
16421 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
16422 (not (neg X)) into (plus X -1) for complex or non-integral modes.
16425 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
16426 if they have any depend clauses.
16428 2017-04-11 Martin Liska <mliska@suse.cz>
16431 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
16432 * ipa-split.c (split_function): Create a local comdat symbol
16433 if caller is in a comdat group.
16435 2017-04-11 Martin Liska <mliska@suse.cz>
16438 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
16441 2017-04-11 Martin Sebor <msebor@redhat.com>
16443 PR middle-end/80364
16444 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
16445 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
16447 (directive::set_width, directive::set_precision, format_character):
16449 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
16452 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
16455 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
16456 conflict, set target->arch_name instead of target->cpu_name.
16458 2017-04-11 Richard Biener <rguenther@suse.de>
16460 PR tree-optimization/80374
16461 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
16462 build_zero_cst, remove fold_convertible_p check again.
16464 2017-04-11 Martin Liska <mliska@suse.cz>
16467 * ubsan.c (instrument_object_size): Do not instrument register
16470 2017-04-11 Jakub Jelinek <jakub@redhat.com>
16473 * config/i386/i386-builtin-types.def
16474 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
16475 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
16476 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
16477 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
16478 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
16479 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
16480 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
16481 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
16482 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
16483 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
16484 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
16485 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
16486 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
16487 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
16488 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
16489 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
16490 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
16491 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
16492 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
16493 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
16494 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
16495 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
16496 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
16497 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
16498 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
16499 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
16500 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
16501 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
16502 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
16503 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
16504 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
16505 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
16506 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
16507 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
16508 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
16509 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
16510 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
16511 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
16512 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
16513 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
16514 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
16515 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
16516 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
16517 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
16518 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
16519 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
16521 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
16522 flag to second_arg_count, handle 4 argument function type _COUNT
16523 aliases, handle second_arg_count on second argument rather than last.
16525 2017-04-10 Jeff Law <law@redhat.com>
16527 PR tree-optimization/80374
16528 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
16529 record anything if we can not convert integer_zero_node to the
16532 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
16535 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16536 Enhance special handling given to the TARGET_P9_MINMAX option in
16537 relation to certain other options.
16539 2017-04-10 Bin Cheng <bin.cheng@arm.com>
16541 PR tree-optimization/80153
16542 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
16543 remove POINTER_PLUS_EXPR's base part directly, rather than through
16546 2017-04-10 Richard Biener <rguenther@suse.de>
16547 Bin Cheng <bin.cheng@arm.com>
16549 PR tree-optimization/80153
16550 * tree-affine.c (aff_combination_to_tree): Get base pointer from
16551 the first element of pointer type aff_tree. Build result expr in
16553 (add_elt_to_tree): Convert to type unconditionally. Remove other
16554 fold_convert calls.
16555 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
16556 (rewrite_use_nonlinear_expr): Check invariant using iv information.
16558 2017-04-10 Richard Biener <rguenther@suse.de>
16560 * tree-ssa-structalias.c (find_func_aliases): Properly handle
16563 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
16565 PR rtl-optimization/70478
16566 * lra-constraints.c (curr_small_class_check): New.
16567 (update_and_check_small_class_inputs): New.
16568 (process_alt_operands): Update curr_small_class_check. Disfavor
16569 alternative insn memory operands. Check available regs for small
16572 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
16575 * config/mips/mips.opt (-mvirt): Update description.
16576 * doc/invoke.texi (-mvirt): Likewise.
16578 2017-04-10 Richard Biener <rguenther@suse.de>
16580 PR middle-end/80362
16581 * fold-const.c (fold_binary_loc): Look at unstripped ops when
16582 looking for NEGATE_EXPR in -A / -B to A / B folding.
16584 2017-04-10 Martin Liska <mliska@suse.cz>
16586 PR gcov-profile/80224
16587 * gcov.c (print_usage): Fix usage string.
16588 (get_gcov_intermediate_filename): Remove.
16589 (output_gcov_file): Use both for normal and intermediate format.
16590 (generate_results): Do not initialize special file for
16591 intermediate format.
16593 2017-04-10 Richard Biener <rguenther@suse.de>
16595 PR tree-optimization/80304
16596 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
16599 2017-04-10 Nathan Sidwell <nathan@acm.org>
16602 * config/rs6000/rs6000.c (rs6000_vector_type): New.
16603 (rs6000_init_builtins): Use it.
16605 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16607 * config/arm/arm.md (<mrc>): Add mode to SET source.
16608 (<mrrc>): Likewise.
16610 2017-04-10 Richard Biener <rguenther@suse.de>
16612 PR middle-end/80344
16613 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
16615 2017-04-10 Jakub Jelinek <jakub@redhat.com>
16618 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
16619 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
16620 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
16621 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
16622 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
16623 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
16624 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
16625 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
16626 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
16627 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
16628 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
16629 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
16630 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
16631 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
16632 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
16633 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
16634 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
16635 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
16636 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
16637 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
16638 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
16639 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
16640 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
16642 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
16644 PR rtl-optimization/70478
16645 * lra-constraints.c: Reverse the last patch.
16647 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
16649 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
16650 Add comment for WCHAR_T.
16652 2017-04-08 Martin Liska <mliska@suse.cz>
16655 2017-04-07 Martin Liska <mliska@suse.cz>
16658 * ipa-split.c (split_function): Add function part to a same comdat
16661 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16664 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
16666 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
16668 * rs6000/rs6000.c (vec_load_pendulum): Rename...
16669 (vec_pairing): ...to this.
16670 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
16671 (rs6000_sched_init): Adjust for name change.
16672 (struct rs6000_sched_context): Likewise.
16673 (rs6000_init_sched_context): Likewise.
16674 (rs6000_set_sched_context): Likewise.
16676 2017-04-07 Jakub Jelinek <jakub@redhat.com>
16682 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
16684 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
16685 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
16686 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
16688 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
16690 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
16692 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
16694 PR rtl-optimization/70703
16695 * ira-color.c (update_conflict_hard_regno_costs): Use
16696 int64_t instead of HOST_WIDE_INT.
16698 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
16700 PR rtl-optimization/70478
16701 * lra-constraints.c (process_alt_operands): Disfavor alternative
16702 insn memory operands.
16704 2017-04-07 Jeff Law <law@redhat.com>
16706 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
16707 CALL and NOTE_INSN_CALL_ARG_LOCATION.
16709 2017-04-07 Martin Liska <mliska@suse.cz>
16712 * config/aarch64/aarch64.c (aarch64_process_target_attr):
16713 Show error message instead of an ICE.
16715 2017-04-07 Martin Liska <mliska@suse.cz>
16718 * ipa-split.c (split_function): Add function part to a same comdat
16721 2017-04-07 Richard Biener <rguenther@suse.de>
16723 PR middle-end/80341
16724 * tree.c (get_unwidened): Also handle ! for_type case for
16726 * convert.c (do_narrow): Split out from ...
16727 (convert_to_integer_1): ... here. Do not pass final truncation
16728 type to get_unwidened for TRUNC_DIV_EXPR.
16730 2017-04-07 Richard Biener <rguenther@suse.de>
16732 * tree-affine.c (wide_int_ext_for_comb): Take type rather
16734 (aff_combination_const): Adjust.
16735 (aff_combination_scale): Likewise.
16736 (aff_combination_add_elt): Likewise.
16737 (aff_combination_add_cst): Likewise.
16738 (aff_combination_convert): Likewise.
16739 (add_elt_to_tree): Likewise. Remove unused argument.
16740 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
16742 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
16744 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
16746 * config/arm/arm.c (arm_default_short_enums): Use
16747 ARM_DEFAULT_SHORT_ENUMS.
16748 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
16750 2017-04-06 Jakub Jelinek <jakub@redhat.com>
16753 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
16754 members with redundant out-of-class redeclaration.
16756 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
16759 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
16760 * config/i386/i386.md (*zero_extendsidi2):
16761 Add (?*x,*x) and (?*v,*v) alternatives.
16763 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
16766 * config/i386/i386.c (ix86_expand_builtin)
16767 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
16768 mode from insn data. Convert operands to insn operand mode.
16769 Copy operands that don't satisfy insn predicate to a register.
16771 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
16773 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
16776 2017-04-06 Richard Biener <rguenther@suse.de>
16778 PR tree-optimization/80334
16779 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
16780 preserve alignment of accesses.
16782 2017-04-06 Richard Biener <rguenther@suse.de>
16784 PR tree-optimization/80262
16785 * tree-sra.c (build_ref_for_offset): Preserve address-space
16787 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
16788 Drop useless address-space information on MEM_REF offsets.
16790 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
16792 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
16794 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
16796 PR rtl-optimization/70703
16797 * ira-color.c (update_conflict_hard_regno_costs): Use
16798 HOST_WIDE_INT instead of long.
16800 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
16803 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
16804 not defined for x86_64 target. Add -mmmx target option when __SSE2__
16806 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
16807 for x86_64 target. Handle -m3dnowa option.
16809 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
16811 PR rtl-optimization/70703
16812 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
16813 (update_conflict_hard_regno_costs): Use long instead of unsigned
16814 arithmetic for cost calculation.
16816 2017-04-05 Jakub Jelinek <jakub@redhat.com>
16817 Bernd Edlinger <bernd.edlinger@hotmail.de>
16820 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
16823 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
16826 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
16827 ptr_mode with Pmode throughout.
16828 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
16829 into probe_stack_range and use DImode.
16831 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
16834 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
16835 call_eh_return is true.
16837 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16839 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
16840 Initialize last_match_fntype_index.
16842 2017-04-05 Jakub Jelinek <jakub@redhat.com>
16845 * tree-nvr.c: Include internal-fn.h.
16846 (pass_return_slot::execute): Ignore internal calls without
16849 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16850 Richard Biener <rguenther@suse.de>
16853 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
16854 captures used multiple times, except for the last use.
16855 * generic-match-head.c: Include gimplify.h.
16857 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16859 PR tree-optimization/79390
16860 * target.h (struct noce_if_info): Declare.
16861 * targhooks.h (default_noce_conversion_profitable_p): Declare.
16862 * target.def (noce_conversion_profitable_p): New target hook.
16863 * ifcvt.h (struct noce_if_info): New type, moved from ...
16864 * ifcvt.c (struct noce_if_info): ... here.
16865 (noce_conversion_profitable_p): Renamed to ...
16866 (default_noce_conversion_profitable_p): ... this. No longer
16868 (noce_try_store_flag_constants, noce_try_addcc,
16869 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
16870 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
16871 instead of noce_conversion_profitable_p.
16872 * config/i386/i386.c: Include ifcvt.h.
16873 (ix86_option_override_internal): Don't override
16874 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
16875 (ix86_noce_conversion_profitable_p): New function.
16876 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
16877 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
16878 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
16879 * doc/tm.texi: Regenerated.
16881 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16883 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
16886 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16889 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
16890 instructions for small multiply cores.
16892 2017-04-04 Jeff Law <law@redhat.com>
16894 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
16896 (mips_expand_vec_perm_const): Initialize elements in orig_perm
16897 that are not set by the loop over the elements.
16899 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16902 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
16903 int mode, convert_modes it to mode as unsigned, otherwise use
16904 lowpart_subreg to mode rather than SImode.
16905 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
16906 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
16907 Use DImode instead of SImode for the shift count operand.
16908 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
16911 2017-04-04 Richard Biener <rguenther@suse.de>
16913 PR middle-end/80281
16914 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
16915 arithmetic done for the negate or the plus. Simplify.
16916 (A - (-B) -> A + B): Likewise.
16917 * fold-const.c (split_tree): Make sure to not negate pointers.
16919 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
16921 PR rtl-optimization/60818
16922 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
16923 a compare of comparisons with the thing compared if this results
16924 in a different machine mode.
16926 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
16928 * alias.c (base_alias_check): Fix typo in comment.
16929 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
16930 * cgraphunit.c (symbol_table::compile): Likewise.
16931 * collect2.c (maybe_run_lto_and_relink): Likewise.
16932 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
16933 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
16934 * config/avr/avr.c (avr_map_op_t): Likewise.
16935 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
16936 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
16937 * config/epiphany/epiphany.md (movcc): Likewise.
16938 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
16939 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
16941 * config/mips/mips.c (mips_save_restore_reg): Likewise.
16942 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
16943 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
16944 * config/sh/sh.c (sh_rtx_costs): Likewise.
16945 * fold-const.c (fold_truth_andor): Likewise.
16946 * genautomata.c (collapse_flag): Likewise.
16947 * gengtype.h (struct type::u::s): Likewise.
16948 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
16949 * input.c (FORMAT_AMOUNT): Likewise.
16950 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
16951 (known_aggs_to_agg_replacement_list): Likewise.
16952 * ipa-inline-analysis.c: Likewise.
16953 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
16954 * ipa-polymorphic-call.c
16955 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
16956 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
16957 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
16959 * modulo-sched.c (apply_reg_moves): Likewise.
16960 * omp-expand.c (build_omp_regions_1): Likewise.
16961 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
16962 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
16963 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
16964 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
16965 * value-prof.c: Likewise.
16966 * var-tracking.c (val_reset): Likewise.
16968 2017-04-03 Richard Biener <rguenther@suse.de>
16970 PR tree-optimization/80275
16971 * fold-const.c (split_address_to_core_and_offset): Handle
16974 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
16976 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
16977 descriptors is at least equal to that of functions.
16979 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
16981 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
16983 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
16986 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
16987 (mov<IMOD4:mode>): New expander.
16988 (*mov<IMOD4:mode>_internal): New insn and split pattern.
16990 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
16992 PR rtl-optimization/79405
16993 * fwprop.c (propagations_left): New variable.
16994 (forward_propagate_into): Decrement it.
16995 (fwprop_init): Initialize it.
16996 (fw_prop): If the variable has reached zero, stop propagating.
16997 (fwprop_addr): Ditto.
16999 2017-03-31 Jakub Jelinek <jakub@redhat.com>
17002 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
17003 a FUNCTION_DECL, pass it as decl instead of origin to
17006 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
17008 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
17011 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
17014 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
17015 TARGET_VSX_SMALL_INTEGER.
17017 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17019 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17020 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
17022 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
17024 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
17025 extraction from odd-numbered MSA register.
17027 2017-03-31 Jakub Jelinek <jakub@redhat.com>
17029 PR middle-end/80173
17030 * expmed.c (store_bit_field_1): Don't attempt to create
17031 a word subreg out of hard registers wider than word if they
17032 have HARD_REGNO_NREGS of 1 for their mode.
17034 PR middle-end/80163
17035 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
17036 conversions to integer types wider than word and pointer.
17039 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
17040 (rtx_equal_for_cselib_p): Pass 0 to it.
17041 * cselib.c (cselib_hasher::equal): Likewise.
17042 (rtx_equal_for_cselib_1): Add depth argument. If depth
17043 is 128, don't look up VALUE locs and punt. Increment
17044 depth in recursive calls when walking VALUE locs.
17046 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
17048 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
17049 (make_gcov_file_name): Use the canonical path name for generating
17051 (read_line): Fix handling of files with ascii null bytes.
17053 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
17055 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
17056 to initialise a vector register instead
17057 of using a const_int.
17059 2017-03-30 Jakub Jelinek <jakub@redhat.com>
17061 PR translation/80189
17062 * gimplify.c (omp_default_clause): Use %qs instead of %s in
17063 diagnostic messages.
17065 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
17068 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
17069 (dfp_diex_<mode>): Update mode of operand 1.
17070 * doc/extend.texi (dxex, dxexq): Document change to return type.
17071 (diex, diexq): Document change to argument type.
17073 2017-03-30 Martin Jambor <mjambor@suse.cz>
17076 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
17077 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
17078 it reflects the signature changes performed at the callee side.
17079 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
17080 to cgraph_build_function_type_skip_args.
17081 (build_function_decl_skip_args): Adjust call to the above function.
17083 2017-03-30 Jakub Jelinek <jakub@redhat.com>
17086 * config/i386/sse.md
17087 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
17088 register as dest whenever it is a MEM not rtx_equal_p to the
17089 corresponding dup operand, and when forcing into reg move the
17090 reg into the memory afterwards.
17091 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
17092 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
17093 for the force_reg mode.
17094 (avx512vl_vextractf128<mode>): Use register as dest either
17095 always when a MEM, or when it is a MEM not rtx_equal_p to the
17096 corresponding dup operand, or even not when it is a CONST_VECTOR
17097 depending on the mode and lo vs. hi.
17098 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
17100 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
17101 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
17102 Likewise. Require that operands[2] is even.
17103 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
17104 Remove extraneous parens. Require that operands[2] is a multiple
17106 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
17107 operands[0] is a MEM if <mask_applied>, the predicates/constraints
17108 disallow memory then.
17110 2017-03-30 Richard Biener <rguenther@suse.de>
17112 PR tree-optimization/77498
17113 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
17114 to non-constants over backedges.
17116 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
17118 PR rtl-optimization/80233
17119 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
17120 as last_combined_insn. Do not test for BARRIER_P separately.
17122 2017-03-29 Andreas Schwab <schwab@suse.de>
17125 * calls.c (prepare_call_address): Convert funexp to Pmode before
17126 copying to temp reg.
17128 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17130 PR tree-optimization/80158
17131 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
17132 Handle possible future case of more than one alternate
17134 (replace_rhs_if_not_dup): Likewise.
17135 (replace_one_candidate): Likewise.
17137 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
17139 PR rtl-optimization/80193
17140 * ira.c (ira): Do not check allocation for LRA.
17142 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
17144 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
17145 (nvptx_output_simt_exit): Declare.
17146 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
17147 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
17148 (init_softstack_frame): Move initialization of crtl->is_leaf to...
17149 (nvptx_declare_function_name): ...here. Emit declaration of local
17150 memory space buffer for omp_simt_enter insn.
17151 (nvptx_output_unisimt_switch): New.
17152 (nvptx_output_softstack_switch): New.
17153 (nvptx_output_simt_enter): New.
17154 (nvptx_output_simt_exit): New.
17155 * config/nvptx/nvptx.h (struct machine_function): New fields
17156 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
17157 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
17158 (UNSPECV_SIMT_EXIT): Ditto.
17159 (omp_simt_enter_insn): New insn.
17160 (omp_simt_enter): New expansion.
17161 (omp_simt_exit): New insn.
17162 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
17164 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
17165 (expand_GOMP_SIMT_ENTER_ALLOC): New.
17166 (expand_GOMP_SIMT_EXIT): New.
17167 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
17168 (GOMP_SIMT_ENTER_ALLOC): Ditto.
17169 (GOMP_SIMT_EXIT): Ditto.
17170 * target-insns.def (omp_simt_enter): New insn.
17171 (omp_simt_exit): Ditto.
17172 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
17174 (lower_rec_simd_input_clauses): Implement SIMT privatization.
17175 (lower_rec_input_clauses): Likewise.
17176 (lower_lastprivate_clauses): Handle SIMT privatization.
17178 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
17179 (ompdevlow_adjust_simt_enter): New.
17180 (find_simtpriv_var_op): New.
17181 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
17182 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
17184 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
17185 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
17186 (copy_decl_for_dup_finish): Ditto.
17188 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
17190 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
17193 * config/i386/i386.c (ix86_option_override_internal): Always
17194 allow -mpreferred-stack-boundary=3 for 64-bit targets.
17196 2017-03-28 Bin Cheng <bin.cheng@arm.com>
17198 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
17200 2017-03-28 Bin Cheng <bin.cheng@arm.com>
17202 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
17203 mark new edge's irreducible flag accordign to it.
17204 (vect_do_peeling): Check loop preheader edge's irreducible flag
17205 and pass it to function slpeel_add_loop_guard.
17207 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
17209 PR tree-optimization/80218
17210 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
17211 Update block frequencies and counts.
17213 2017-03-28 Richard Biener <rguenther@suse.de>
17215 PR tree-optimization/78644
17216 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
17217 of a simplification result we may not use it at all.
17219 2017-03-28 Richard Biener <rguenther@suse.de>
17222 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
17223 without arguments, generate default definition of a SSA name.
17225 2017-03-28 Richard Biener <rguenther@suse.de>
17227 PR middle-end/80222
17228 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
17229 TYPE_REF_CAN_ALIAS_ALL references.
17230 * fold-const.c (fold_indirect_ref_1): Likewise.
17232 2017-03-28 Martin Liska <mliska@suse.cz>
17235 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
17236 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
17238 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
17239 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
17241 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
17242 (EXTRA_SPECS): Define.
17243 (SUBTARGET_EXTRA_SPECS): Likewise.
17244 (SUBTARGET_CPP_SPEC): Likewise.
17245 * config/arc/elf.h (EXTRA_SPECS): Renamed to
17246 SUBTARGET_EXTRA_SPECS.
17247 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
17249 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
17251 * config/arc/simdext.md (vst64_insn): Update pattern.
17252 (vld32wh_insn): Likewise.
17253 (vld32wl_insn): Likewise.
17254 (vld64_insn): Likewise.
17255 (vld32_insn): Likewise.
17257 2017-03-28 Marek Polacek <polacek@redhat.com>
17260 * fold-const.c (fold_comparison): Use protected_set_expr_location
17261 instead of SET_EXPR_LOCATION.
17263 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
17265 * tree.c (add_expr): Avoid name lookup warning.
17267 2017-03-27 Jeff Law <law@redhat.com>
17269 PR tree-optimization/80216
17270 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
17271 function name. Limit recursion depth.
17272 (record_temporary_equivalences): Corresponding changes.
17274 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
17276 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
17279 2017-03-27 Jakub Jelinek <jakub@redhat.com>
17282 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
17284 * cfgcleanup.c (reg_note_cfa_p): New array.
17285 (insns_have_identical_cfa_notes): New function.
17286 (old_insns_match_p): Don't cross-jump in between /f
17287 and non-/f instructions. If both i1 and i2 are frame related,
17288 verify all CFA notes, their order and content.
17290 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
17293 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
17294 HImode and SImode with zero extend to DImode to one insn.
17295 (bswap<mode>2_extenddi): Likewise.
17296 (bswapsi2_extenddi): Likewise.
17297 (bswaphi2_extendsi): Likewise.
17298 (bswaphi2): Combine bswap HImode and SImode into one insn.
17299 Separate memory insns from swapping register.
17300 (bswapsi2): Likewise.
17301 (bswap<mode>2): Likewise.
17302 (bswaphi2_internal): Delete, no longer used.
17303 (bswapsi2_internal): Likewise.
17304 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
17305 store, and gpr<-gpr swap insns.
17306 (bswap<mode>2_store): Likewise.
17307 (bswaphi2_reg): Register only splitter, combine with the splitter.
17308 (bswaphi2 splitter): Likewise.
17309 (bswapsi2_reg): Likewise.
17310 (bswapsi2 splitter): Likewise.
17311 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
17312 the insns into load, store, and register/register insns.
17313 (bswapdi2_ldbrx): Likewise.
17314 (bswapdi2_load): Likewise.
17315 (bswapdi2_store): Likewise.
17316 (bswapdi2_reg): Likewise.
17318 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
17320 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
17321 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
17323 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
17326 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
17328 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
17329 special handling for target option conflicts between dform
17330 options (-mpower9-dform, -mpower9-dform-vector,
17331 -mpower9-dform-scalar) and -mno-direct-move.
17333 2017-03-27 Richard Biener <rguenther@suse.de>
17335 PR tree-optimization/80181
17336 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
17338 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
17340 * config/arc/predicates.md (move_double_src_operand): Replace the
17341 call to move_double_src_operand with a call to address_operand.
17343 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
17345 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
17346 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
17347 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
17349 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
17351 * config/arc/predicates.md (long_immediate_loadstore_operand):
17352 Consider scaled addresses cases.
17354 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
17356 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
17357 restored when in interrupt.
17358 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
17359 doesn't have delay slot.
17361 2017-03-27 Richard Biener <rguenther@suse.de>
17364 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
17365 inlined thunk clones.
17367 2017-03-27 Jakub Jelinek <jakub@redhat.com>
17370 * asan.c (instrument_derefs): Copy over last operand from
17371 original COMPONENT_REF to the new COMPONENT_REF with
17372 DECL_BIT_FIELD_REPRESENTATIVE.
17373 * ubsan.c (instrument_object_size): Likewise.
17375 2017-03-27 Richard Biener <rguenther@suse.de>
17377 PR tree-optimization/80170
17378 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
17379 sure DR/SCEV didnt fold in constants we do not see when looking
17380 at the reference base alignment.
17382 2017-03-27 Richard Biener <rguenther@suse.de>
17384 PR middle-end/80171
17385 * gimple-fold.c (fold_ctor_reference): Properly guard against
17386 NULL return value from canonicalize_constructor_val.
17388 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
17391 * config/i386/i386.c (ix86_expand_builtin)
17392 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
17393 flags reg setting and flags reg using instructions.
17394 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
17395 clobbering instructions to zero extend op2.
17397 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
17399 * doc/install.texi (Configuration) <--with-aix-soname>:
17400 Update link to AIX ld.
17402 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
17404 PR rtl-optimization/80160
17405 PR rtl-optimization/80159
17406 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
17407 reg_alternate_class into account.
17409 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
17412 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
17413 to consider in curr_insn_transform.
17415 2017-03-24 Jakub Jelinek <jakub@redhat.com>
17417 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
17418 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
17419 and emit_mode_inner.
17421 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17423 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
17424 argument to the overloaded builtin variants. Use the new flag to
17425 deprecate certain builtin variants.
17426 * config/s390/s390-builtin-types.def: Add new builtin types.
17427 * config/s390/s390-builtins.h: Support new flags field for
17428 overloaded builtins.
17429 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
17430 (s390_macro_to_expand): Enable vector float data type.
17431 (s390_cpu_cpp_builtins_internal): Indicate support of the new
17432 builtins by incrementing the __VEC__ version number.
17433 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
17435 (s390_resolve_overloaded_builtin): Emit error messages depending
17436 on the builtin flags.
17437 * config/s390/s390.c (s390_expand_builtin): Support additional
17438 flags argument. Change error message to match the messages
17439 emitted in s390-c.c.
17440 * config/s390/s390.md: New UNSPEC_* constants.
17441 (op_type): Add new instruction types.
17442 * config/s390/vecintrin.h: Add new builtins and test data class
17444 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
17445 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
17446 (VEC_INEXACT, VEC_NOINEXACT): New constants.
17447 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
17448 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
17449 ("vec_mergel<mode>"): V_HW -> VEC_HW.
17451 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
17452 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
17453 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
17454 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
17456 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
17457 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
17458 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
17459 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
17461 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
17462 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
17463 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
17464 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
17465 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
17466 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
17467 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
17469 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
17470 ("vec_scatter_element<V_HW_4:mode>_DI")
17471 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
17472 ("vec_fpint<mode>", "vflls")
17473 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
17474 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
17475 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
17476 ("*vec_cmphe<mode>_cc"): ... these.
17478 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
17479 mode constant instead of magic value.
17481 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17483 * config/s390/s390.c (s390_expand_vec_compare): Support other
17484 vector floating point modes than just V2DF.
17485 (s390_expand_vcond): Likewise.
17486 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
17487 (s390_cannot_change_mode_class): Prevent mode changes between TF
17488 and V1TF in vector registers.
17489 * config/s390/s390.md (DF, SF): New mode attributes.
17490 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
17491 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
17492 SFmode support for VRs.
17493 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
17495 (VFT, VF_HW): New mode iterators.
17496 (vw, sdx): New mode attributes.
17497 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
17498 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
17499 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
17500 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
17501 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
17502 also the new vector floating point modes. Renaming to ...
17504 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
17505 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
17506 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
17507 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
17508 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
17509 ("vec_unordered<mode>"): ... these.
17511 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
17512 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
17513 ("*vec_extendv2df"): New insn definitions.
17515 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17517 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
17518 ("mulditi3_2", "*muldi3_sign"): New patterns.
17519 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
17520 rename the pattern definition.
17522 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17524 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
17526 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
17528 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17530 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
17531 instruction if possible.
17532 * config/s390/vector.md (vec_halfnumelts): New mode
17534 ("*vec_vllezlf<mode>"): New pattern.
17536 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17538 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
17539 ("popcountv4si2", "popcountv2di2"): Rename to ...
17540 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
17541 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
17543 ("popcount<mode>2_vxe"): New pattern.
17545 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17547 * common/config/s390/s390-common.c (processor_flags_table): Add
17549 * config.gcc: Add arch12.
17550 * config/s390/driver-native.c (s390_host_detect_local_cpu):
17551 Default to arch12 for unknown CPU model numbers.
17552 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
17553 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
17554 PROCESSOR_max sanity check.
17555 * config/s390/s390-opts.h (enum processor_type): Add
17557 * config/s390/s390.c (processor_table): Add arch12.
17558 (s390_expand_builtin): Add check for B_VXE flag.
17559 (s390_issue_rate): Add PROCESSOR_ARCH12.
17560 (s390_get_sched_attrmask): Likewise.
17561 (s390_get_unit_mask): Likewise.
17562 (s390_sched_score): Enable z13 scheduling for arch12.
17563 (s390_sched_reorder): Likewise.
17564 (s390_sched_variable_issue): Likewise.
17565 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
17567 (s390_tune_attr): Use z13 scheduling also for arch12.
17568 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
17569 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
17570 (TARGET_VXE_P): New macros.
17571 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
17572 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
17573 * config/s390/s390.opt: Add arch12 as processor_type.
17575 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17577 * config/s390/s390.md
17578 ("fixuns_truncdddi2", "fixuns_trunctddi2")
17579 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
17580 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
17582 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
17583 Rename expanders to ...
17585 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
17586 ("fixuns_truncdddi2_emu"): ... these.
17588 ("fixuns_trunc<mode>si2_emu"): New expander.
17590 ("*fixuns_truncdfdi2_z13"): Rename to ...
17591 ("*fixuns_truncdfdi2_vx"): ... this.
17593 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17595 * config/s390/2964.md: Remove the single element vector compare
17596 instructions which are no longer used.
17597 * config/s390/s390.c (s390_select_ccmode): Remove handling of
17599 (s390_canonicalize_comparison): Remove handling of DFmode
17601 (s390_expand_vec_compare_scalar): Remove function.
17602 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
17603 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
17605 ("*cmp<mode>_ccs"): Add wfcdb instruction.
17607 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17609 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
17611 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
17612 will anyway by matched by mov<mode>_64dfp.
17614 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17616 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
17617 vlef/vstef. Add missing operand to vleif.
17619 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17621 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
17622 pair for all vector types with 64 bit elements.
17623 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
17624 * config/s390/vector.md (V_HW_64): ... here.
17625 (V_128_NOSINGLE): New mode iterator.
17626 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
17627 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
17628 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
17629 ("*vec_load_pairv2di"): Change to ...
17630 ("*vec_load_pair<mode>"): ... this one.
17632 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17634 * config/s390/constraints.md: Add comments.
17635 (jKK): Reject element sizes > 8 bytes.
17636 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
17638 * config/s390/s390.md: Add the s_operand checks formerly in
17639 s390_split_ok_p to various splitters where they are still
17641 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
17642 for 128 bit vectors. Plus two splitters.
17644 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17646 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
17649 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17652 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
17653 error if the boundary argument is not constant.
17655 2017-03-24 Jakub Jelinek <jakub@redhat.com>
17657 PR rtl-optimization/80112
17658 * loop-doloop.c (doloop_condition_get): Don't check condition
17659 if cmp isn't SET with IF_THEN_ELSE src.
17661 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17663 PR tree-optimization/80158
17664 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
17665 replacing a candidate statement, also replace it for the
17666 candidate's alternate interpretation.
17667 (replace_rhs_if_not_dup): Likewise.
17668 (replace_one_candidate): Likewise.
17670 2017-03-24 Richard Biener <rguenther@suse.de>
17672 PR tree-optimization/80167
17673 * graphite-isl-ast-to-gimple.c
17674 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
17676 (translate_isl_ast_to_gimple::get_rename): Likewise.
17678 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
17680 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
17681 handling of certain combinations of target options, including the
17682 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
17683 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
17685 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17688 * config/arm/arm.md (*load_multiple): Add reload_completed to
17689 matching condition.
17691 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17692 Richard Biener <rguenth@suse.de>
17694 PR tree-optimization/79908
17695 PR tree-optimization/80136
17696 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
17697 been cast away, gimplify_and_add suffices.
17699 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
17701 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
17703 2017-03-23 Richard Biener <rguenther@suse.de>
17705 PR tree-optimization/80032
17706 * gimplify.c (gimple_push_cleanup): Forced unconditional
17707 cleanups still have to go to the conditional_cleanups
17710 2017-03-22 Jakub Jelinek <jakub@redhat.com>
17712 PR tree-optimization/80072
17713 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
17715 (next_operand_entry_id): Change type to unsigned int.
17716 (sort_by_operand_rank): Make sure to return the right return value
17717 even if unsigned fields are bigger than INT_MAX.
17718 (struct oecount): Change cnt and id type to unsigned int.
17719 (oecount_hasher::equal): Formatting fix.
17720 (oecount_cmp): Make sure to return the right return value
17721 even if unsigned fields are bigger than INT_MAX.
17722 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
17725 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
17726 TREE_READONLY on result if writing it more than once.
17729 * doc/invoke.texi (-fsanitize=thread): Document that with
17730 -fnon-call-exceptions atomics are not able to throw
17734 * tsan.c: Include tree-eh.h.
17735 (instrument_builtin_call): Call maybe_clean_eh_stmt or
17736 maybe_clean_or_replace_eh_stmt where needed.
17737 (instrument_memory_accesses): Add cfg_changed argument.
17738 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
17739 if it returned true.
17740 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
17742 PR rtl-optimization/63191
17743 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
17744 wrapper function, moved the whole old content into ...
17745 (ix86_delegitimize_address_1): ... this. New inline function.
17746 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
17747 true as last argument instead of ix86_delegitimize_address.
17749 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
17751 * config/aarch64/aarch64.c (generic_branch_cost): Copy
17752 cortexa57_branch_cost.
17754 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
17756 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
17758 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17761 * doc/md.texi (Constraints): Document wA constraint.
17762 * config/rs6000/constraints.md (wA): New.
17763 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
17764 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
17765 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
17766 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
17768 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
17771 * gimplify.c (is_oacc_declared): New function.
17772 (oacc_default_clause): Use it to set default flags for acc declared
17773 variables inside parallel regions.
17774 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
17775 declared variables.
17776 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
17777 declare attribute to any decl as necessary.
17779 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
17782 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
17783 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
17784 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
17785 (arm_arch_lpae): This.
17786 * config/arm/arm.c (arm_arch7ve): Rename into ...
17787 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
17788 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
17791 2017-03-22 Martin Liska <mliska@suse.cz>
17794 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
17795 error message instead of an ICE.
17797 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17799 * doc/extend.texi (6.11 Additional Floating Types): Revise.
17801 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
17803 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
17805 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
17808 2017-03-21 Martin Sebor <msebor@redhat.com>
17810 * doc/extend.texi: Use "cannot" instead of "can't."
17811 * doc/hostconfig.texi: Same.
17812 * doc/install.texi: Same.
17813 * doc/invoke.texi: Same.
17814 * doc/loop.texi: Same.
17815 * doc/md.texi: Same.
17816 * doc/objc.texi: Same.
17817 * doc/rtl.texi: Same.
17818 * doc/tm.texi: Same.
17819 * doc/tm.texi.in: Same.
17820 * doc/trouble.texi: Same.
17822 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
17825 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
17826 (collect_checksum_attributes): Set it.
17827 (die_checksum_ordered): Use it.
17829 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17831 PR tree-optimization/79908
17832 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
17833 change: For a VA_ARG whose LHS has been cast away, use
17834 force_gimple_operand to construct the side effects.
17836 2017-03-21 David Malcolm <dmalcolm@redhat.com>
17838 PR translation/80001
17839 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
17840 more amenable to translation.
17841 (oacc_loop_auto_partitions): Likewise.
17843 2017-03-21 Marek Polacek <polacek@redhat.com>
17844 Martin Sebor <msebor@redhat.com>
17846 PR tree-optimization/80109
17847 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
17848 on INTEGRAL_TYPE_P.
17850 2017-03-21 Jakub Jelinek <jakub@redhat.com>
17851 Segher Boessenkool <segher@kernel.crashing.org>
17854 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
17855 check reg_used_between_p between insn and one of succ or succ2
17856 depending on if succ is artificial insn not inserted into insn
17859 2017-03-21 Martin Liska <mliska@suse.cz>
17861 PR gcov-profile/80081
17862 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
17863 * doc/gcc.texi: Include gcov-dump stuff.
17864 * doc/gcov-dump.texi: New file.
17866 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
17868 PR rtl-optimization/79150
17869 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
17870 conditional jump, if the jump is the last insn of the loop.
17872 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17873 Richard Biener <rguenth@suse.de>
17875 PR tree-optimization/79908
17876 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
17877 been cast away, use force_gimple_operand to construct the side
17880 2017-03-21 Martin Liska <mliska@suse.cz>
17882 PR libfortran/79956
17883 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
17886 2017-03-21 Brad Spengler <spender@grsecurity.net>
17889 * plugin.c (htab_hash_plugin): New function.
17890 (add_new_plugin): Use it and adjust.
17891 (parse_plugin_arg_opt): Adjust.
17892 (init_one_plugin): Likewise.
17894 2017-03-21 Richard Biener <rguenther@suse.de>
17896 PR tree-optimization/80032
17897 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
17898 if set force the cleanup to happen unconditionally.
17899 (gimplify_target_expr): Push inserted clobbers with force_uncond
17900 to avoid them being removed by control-dependent DCE.
17902 2017-03-21 Richard Biener <rguenther@suse.de>
17904 PR tree-optimization/80122
17905 * tree-inline.c (copy_bb): Do not expans va-arg packs or
17906 va_arg_pack_len when the inlined call stmt requires pack
17908 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
17910 2017-03-21 Jakub Jelinek <jakub@redhat.com>
17913 * tsan.c (instrument_builtin_call): If the memory model argument
17914 is not a constant, assume it is valid.
17917 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
17920 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
17922 PR rtl-optimization/79910
17923 * combine.c (can_combine_p): Do not allow combining an I0 or I1
17924 if its dest is used by an insn before I2 (other than the combined
17925 insns themselves, which are properly handled already).
17927 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
17930 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
17932 * combine.c (record_used_regs): New static function.
17933 (try_combine): Handle situations where there is an additional
17934 instruction between I2 and I3 which needs to have a LOG_LINK
17938 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
17940 * combine.c (try_combine): Delete redundant i1 test. Call
17941 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
17943 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17946 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
17947 alternatives 13/14.
17949 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17951 PR tree-optimization/80054
17952 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
17953 the optimization if a PHI or any of its arguments is not dominated
17954 by the candidate's basis. Use gphi* rather than gimple* as
17956 (replace_profitable_candidates): Clean up a gimple* variable that
17957 should be a gphi* variable.
17959 2017-03-20 Martin Sebor <msebor@redhat.com>
17962 * doc/extend.texi (attribute constructor): Document present limitation.
17964 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
17967 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
17968 __POWER9_VECTOR__ #ifdef control, change template definition to
17969 use Power9-specific built-in function.
17970 (vec_any_eq): Likewise.
17971 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
17972 to control outcomes from this test.
17973 (vector_ae_<mode>p): For VEC_F modes, likewise.
17975 2017-03-20 Ian Lance Taylor <iant@google.com>
17977 * config/i386/i386.c (ix86_function_regparm): Save an extra
17978 register for -fsplit-stack with DECL_STATIC_CHAIN.
17980 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
17983 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
17984 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
17986 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
17988 * config/riscv/riscv.c (riscv_print_operand): Use "fence
17990 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
17993 2017-03-20 Marek Polacek <polacek@redhat.com>
17996 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
17998 2017-03-20 Richard Biener <rguenther@suse.de>
18000 PR tree-optimization/80113
18001 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
18002 allocate extra SSA name for PHI def.
18003 (add_close_phis_to_outer_loops): Likewise.
18004 (add_close_phis_to_merge_points): Likewise.
18005 (copy_loop_close_phi_args): Likewise.
18006 (copy_cond_phi_nodes): Likewise.
18008 2017-03-20 Martin Liska <mliska@suse.cz>
18010 PR middle-end/79753
18011 * tree-chkp.c (chkp_build_returned_bound): Do not build
18012 returned bounds for a LHS that's not a BOUNDED_P type.
18014 2017-03-20 Martin Liska <mliska@suse.cz>
18018 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
18019 COMPLEX_CST and VECTOR_CST.
18021 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18024 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
18025 target operand. A new splitter adds the clobber statement in case
18026 the target operand is dead anyway.
18028 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
18030 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
18031 to age-old versions of binutils and glibc.
18033 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
18035 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
18037 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
18039 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
18041 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
18043 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
18044 requirement for binutils 2.13.
18046 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
18048 * combine.c (try_combine): Delete redundant i1 test. Call
18049 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
18051 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
18053 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
18054 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
18056 <riscv64-*-elf>: Re-arrange section
18057 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
18058 <riscv32-*-linux>: Likewise.
18059 <riscv64-*-elf>: Likewise
18060 <riscv64-*-linux>: Likewise.
18062 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
18065 * aarch64.opt(verbose-cost-dump): Fix typo.
18067 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
18070 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
18071 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
18073 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
18075 * reload.c (find_reloads): When reloading a nonoffsettable address,
18076 use RELOAD_OTHER for it and its address reloads.
18078 PR rtl-optimization/79910
18079 * combine.c (record_used_regs): New static function.
18080 (try_combine): Handle situations where there is an additional
18081 instruction between I2 and I3 which needs to have a LOG_LINK
18084 2017-03-17 Jeff Law <law@redhat.com>
18086 PR tree-optimization/71437
18087 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
18088 conditional in the hash table first.
18089 (vrp_dom_walker::before_dom_children): Extract condition from
18090 ASSERT_EXPR. Record condition, its inverion and any implied
18091 conditions as well.
18093 2017-03-17 Marek Polacek <polacek@redhat.com>
18094 Markus Trippelsdorf <markus@trippelsdorf.de>
18096 PR tree-optimization/80079
18097 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
18100 2017-03-17 Richard Biener <rguenther@suse.de>
18102 PR middle-end/80075
18103 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
18104 Properly verify the LHS before the RHS possibly claims to be
18106 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
18109 2017-03-17 Martin Jambor <mjambor@suse.cz>
18111 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
18112 (List of -O2 options): Likewise.
18113 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
18116 2017-03-17 Tom de Vries <tom@codesourcery.com>
18118 * gcov-dump.c (print_usage): Print bug_report_url.
18120 2017-03-17 Richard Biener <rguenther@suse.de>
18122 PR middle-end/80050
18123 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
18124 (parser::peek): Likewise.
18126 2017-03-17 Richard Biener <rguenther@suse.de>
18128 PR tree-optimization/80048
18129 * sese.c (free_sese_info): Properly release rename_map and
18130 copied_bb_map elements.
18132 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
18134 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
18135 Add linked-list forward and backlinks. Insert on
18136 construction, remove on destruction.
18137 (class pass_store_merging): Add m_stores_head field.
18138 (pass_store_merging::terminate_and_process_all_chains):
18139 Iterate over m_stores_head list.
18140 (pass_store_merging::terminate_all_aliasing_chains):
18142 (pass_store_merging::execute): Check for debug stmts first.
18143 Push new chains onto the m_stores_head stack.
18145 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
18148 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
18149 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
18150 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
18152 2017-03-16 Jeff Law <law@redhat.com>
18154 PR tree-optimization/71437
18155 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
18156 member function. Implementation moved into after_dom_children
18157 member function and into the threader's thread_outgoing_edges
18159 (dom_opt_dom_walker::after_dom_children): Simplify by moving
18160 some code into new thread_outgoing_edges.
18161 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
18162 definition. Simplify marker handling (do it here). Assume we always
18163 have the available expression and the const/copies tables.
18164 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
18166 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
18167 * tree-vrp.c (equiv_stack): No longer file scoped.
18168 (vrp_dom_walker): New class.
18169 (vrp_dom_walker::before_dom_children): New member function.
18170 (vrp_dom_walker::after_dom_children): Likewise.
18171 (identify_jump_threads): Setup domwalker. Use it rather than
18172 walking edges in a random order by hand. Simplify setup/finalization.
18173 (finalize_jump_threads): Remove.
18174 (vrp_finalize): Do not call identify_jump_threads here.
18175 (execute_vrp): Do it here instead and call thread_through_all_blocks
18178 PR tree-optimization/71437
18179 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
18181 (simplify_stmt_for_jump_threading): Add basic_block argument. All
18183 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
18184 (dom_opt_dom_walker::thread_across_edge): Remove
18185 handle_dominating_asserts argument. All callers changed.
18186 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
18187 changes. Remove calls to lhs_of_dominating_assert. Other
18188 uses of handle_dominating_asserts turn into unconditional code
18189 (simplify_control_stmt_condition_1): Likewise.
18190 (simplify_control_stmt_condition): Likewise.
18191 (thread_through_normal_block, thread_across_edge): Likewise.
18192 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
18193 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
18194 object if it is not an SSA_NAME.
18195 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
18196 before calling into the VRP specific simplifiers.
18197 (identify_jump_threads): Remove handle_dominating_asserts
18200 2017-03-16 Jakub Jelinek <jakub@redhat.com>
18203 * tree-diagnostic.c (default_tree_printer): No longer static.
18204 * tree-diagnostic.h (default_tree_printer): New prototype.
18206 2017-03-16 Tamar Christina <tamar.christina@arm.com>
18208 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
18209 Change ins into fmov.
18211 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18213 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
18214 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
18215 Use h_con constraint for operand 1.
18216 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
18217 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
18219 2017-03-15 Jeff Law <law@redhat.com>
18221 PR tree-optimization/71437
18222 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
18223 (record_temporary_equivalences): Use it.
18225 PR tree-optimization/71437
18226 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
18227 tree-ssa-scopedtables.
18228 (lookup_avail_expr, build_and_record_new_cond): Likewise.
18229 (record_conditions, record_cond, vuse_eq): Likewise.
18230 (record_edge_info): Adjust to API tweak of record_conditions.
18231 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
18232 (record_temporary_equivalences, optimize_stmt): Likewise.
18233 (eliminate_redundant_computations): Likewise.
18234 (record_equivalences_from_stmt): Likewise.
18235 * tree-ssa-scopedtables.c: Include options.h and params.h.
18236 (vuse_eq): New function, moved from tree-ssa-dom.c
18237 (build_and_record_new_cond): Likewise.
18238 (record_conditions): Likewise. Accept vector of conditions rather
18239 than edge_equivalence structure for first argument.
18240 for the first argument.
18241 (avail_exprs_stack::lookup_avail_expr): New member function, moved
18242 from tree-ssa-dom.c.
18243 (avail_exprs_stack::record_cond): Likewise.
18244 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
18245 from tree-ssa-dom.c.
18246 (avail_exprs_stack): Add new member functions lookup_avail_expr
18248 (record_conditions): Declare.
18250 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
18253 * lra-constraints.c (process_alt_operands): Increase reject for
18254 reloading an input/output operand.
18256 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
18259 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
18260 insns to convert from signed/unsigned char/short to IEEE 128-bit
18262 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
18264 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
18267 * config/i386/i386.c (ix86_vector_duplicate_value): Create
18268 subreg of inner mode for values already in registers.
18270 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
18272 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
18273 iteration reg is used after the loop.
18275 2017-03-14 Martin Sebor <msebor@redhat.com>
18277 PR tree-optimization/79800
18278 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
18279 precision in negative-positive range.
18280 (format_floating): Call non-const overload with adjusted precision.
18282 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
18285 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
18288 2017-03-14 Martin Sebor <msebor@redhat.com>
18290 PR middle-end/80020
18291 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
18292 * builtins.def (aligned_alloc): Use it.
18295 * Makefile.in (GTFILES): Add calls.c.
18296 * calls.c: Include "gt-calls.h".
18298 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
18300 PR rtl-optimization/79728
18301 * regs.h (struct target_regs): New field
18302 x_contains_allocatable_regs_of_mode.
18303 (contains_allocatable_regs_of_mode): New macro.
18304 * reginfo.c (init_reg_sets_1): Initialize it, and change
18305 contains_reg_of_mode so it includes global regs as well.
18306 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
18307 rather than contains_regs_of_mode.
18309 2017-03-14 Martin Liska <mliska@suse.cz>
18311 * doc/invoke.texi: Document options that can't be combined with
18312 -fcheck-pointer-bounds.
18314 2017-03-14 Martin Liska <mliska@suse.cz>
18316 PR middle-end/79831
18317 * doc/invoke.texi (-Wchkp): Document the option.
18319 2017-03-14 Martin Liska <mliska@suse.cz>
18321 * Makefile.in: Install gcov-dump.
18323 2017-03-14 Martin Liska <mliska@suse.cz>
18325 * multiple_target.c (expand_target_clones): Bail out for
18326 an invalid attribute.
18328 2017-03-14 Richard Biener <rguenther@suse.de>
18330 * alias.c (struct alias_set_entry): Pack properly.
18331 * cfgloop.h (struct loop): Likewise.
18332 * cse.c (struct set): Likewise.
18333 * ipa-utils.c (struct searchc_env): Likewise.
18334 * loop-invariant.c (struct invariant): Likewise.
18335 * lra-remat.c (struct cand): Likewise.
18336 * recog.c (struct change_t): Likewise.
18337 * rtl.h (struct address_info): Likewise.
18338 * symbol-summary.h (function_summary): Likewise.
18339 * tree-loop-distribution.c (struct partition): Likewise.
18340 * tree-object-size.c (struct object_size_info): Likewise.
18341 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
18342 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
18343 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
18344 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
18345 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
18346 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
18347 (struct _stmt_vec_info): Likewise.
18349 2017-03-14 Martin Liska <mliska@suse.cz>
18352 * multiple_target.c (create_dispatcher_calls): Check that
18353 a target can create a function dispatcher.
18355 2017-03-14 Martin Liska <mliska@suse.cz>
18358 * multiple_target.c (expand_target_clones): Drop local.local
18359 flag for default implementation.
18361 2017-03-14 Richard Biener <rguenther@suse.de>
18363 PR tree-optimization/80030
18364 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
18366 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
18368 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
18369 gcc_fallthrough() instead of __attribute__((fallthrough));
18371 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
18373 * doc/gcc.texi: Remove "up" link to (DIR).
18374 * doc/gccint.texi: Ditto.
18376 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
18378 * doc/install.texi (Specific) <avr>: Remove reference to
18381 2017-03-13 Jeff Law <law@redhat.com>
18383 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
18384 attribute rather than comments.
18386 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
18387 match_scratch operand is highest.
18389 2017-03-13 Martin Liska <mliska@suse.cz>
18391 PR middle-end/78339
18392 * ipa-pure-const.c (warn_function_noreturn): If the declarations
18393 is a CHKP clone, use original declaration.
18395 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18397 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
18398 (arc_conditional_register_usage): Use a different allocation order
18399 when optimizing for size.
18400 * common/config/arc/arc-common.c (arc_option_optimization_table):
18401 Section anchors default on when optimizing for size.
18403 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18405 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
18407 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18409 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
18410 * config/arc/arc.md (cpu_facility): Add cd variant.
18411 (*movqi_insn): Add code density variant.
18412 (*movhi_insn): Likewise.
18413 (*movqi_insn): Likewise.
18414 (*addsi3_mixed): Likewise.
18415 (subsi3_insn): Likewise.
18417 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18419 * config/arc/arc.md (movsi_cond_exec): Update constraint.
18421 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18423 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
18424 expressions with MINUS and UNARY ops.
18426 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18429 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
18431 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
18432 between vec_select and vector argument.
18433 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
18434 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
18435 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
18436 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
18437 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
18438 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
18440 2017-03-13 Richard Biener <rguenther@suse.de>
18443 * params.def (vect-max-peeling-for-alignment): Fix typo.
18445 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
18447 * doc/install.texi (Specific) <mips-*-*>: Remove description of
18448 issue that only occurred with binutils below 2.18.
18450 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
18452 * doc/install.texi (Specific) <cris-axis-elf>: No longer
18453 refer to binutils 2.11/2.12 minimum.
18455 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
18457 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
18458 ftp.kernel.org and simplify binutils requirement.
18460 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
18462 * doc/invoke.texi (Warning Options): Fix spelling of link-time
18464 (Optimize Options): Ditto. Also remove redundancy.
18466 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18468 PR translation/79848
18469 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
18471 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
18472 to G_ to avoid double translation.
18474 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18476 PR translation/79923
18477 * auto-profile.c (get_combined_location): Convert leading
18478 character of diagnostics to lower case and remove trailing period.
18479 (read_profile): Likewise for various diagnostics.
18480 * config/arm/arm.c (arm_option_override): Remove trailing period
18481 from various diagnostics.
18482 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
18483 (msp430_expand_delay_cycles): Likewise.
18485 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18488 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
18489 full command-line argument, rather than just "str".
18490 (aarch64_validate_march): Likewise.
18491 (aarch64_validate_mtune): Likewise.
18493 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
18495 PR rtl-optimization/78911
18496 * lra-assigns.c (must_not_spill_p): New function.
18497 (spill_for): Use it.
18499 2017-03-10 Jakub Jelinek <jakub@redhat.com>
18501 PR tree-optimization/79981
18502 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
18503 ATOMIC_COMPARE_EXCHANGE ifn result.
18504 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
18505 IFN_ATOMIC_COMPARE_EXCHANGE.
18507 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18510 * opts.c (parse_sanitizer_options): Add missing question mark to
18511 "did you mean" message.
18513 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18515 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
18517 (VMULEUH_UNS): Likewise.
18518 (VMULOUB_UNS): Likewise.
18519 (VMULOUH_UNS): Likewise.
18520 * config/rs6000/rs6000.c (builtin_function_type): Remove
18521 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
18523 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18526 * read-rtl-function.c (function_reader::read_rtx_operand): Update
18527 x with result of extra_parsing_for_operand_code_0.
18528 (function_reader::extra_parsing_for_operand_code_0): Convert
18529 return type from void to rtx, returning x. When reading
18530 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
18531 larger size containing struct block_symbol.
18533 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
18535 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
18536 -mfloat128-hardware without -m64.
18538 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
18541 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
18542 entries to the case statement that marks unsigned arguments to
18543 overloaded functions.
18545 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
18547 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
18548 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
18550 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
18553 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
18554 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
18556 2017-03-10 Martin Liska <mliska@suse.cz>
18560 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
18561 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
18564 2017-03-10 Olivier Hainque <hainque@adacore.com>
18566 * tree-switch-conversion (array_value_type): Start by resetting
18567 candidate type to it's main variant.
18569 2017-03-10 Jakub Jelinek <jakub@redhat.com>
18571 PR rtl-optimization/79909
18572 * combine.c (try_combine): Use simplify_replace_rtx on individual
18573 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
18574 of the whole CALL_INSN_FUNCTION_USAGE.
18576 PR tree-optimization/79972
18577 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
18578 get_range_info on SSA_NAMEs. Formatting fixes.
18580 2017-03-10 Richard Biener <rguenther@suse.de>
18581 Jakub Jelinek <jakub@redhat.com>
18583 PR tree-optimization/77975
18584 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
18585 edge to be constant.
18586 (get_val_for): For constant x return it. Formatting fix.
18587 (loop_niter_by_eval): Avoid pointless looping if the next iteration
18588 would use the same bases as the current one.
18590 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18592 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
18593 instead of vec_select for V1TImode.
18594 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
18596 (VSX_LE_128): Add V1TI to this mode iterator.
18597 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
18598 (*vsx_le_perm_store_<mode>): Likewise.
18599 (pre-reload splitter for VSX stores): Likewise.
18600 (post-reload splitter for VSX stores): Likewise.
18601 (*vsx_xxpermdi2_le_<mode>): Likewise.
18602 (*vsx_lxvd2x2_le_<mode>): Likewise.
18603 (*vsx_stxvd2x2_le_<mode>): Likewise.
18605 2017-03-09 Michael Eager <eager@eagercon.com>
18607 Correct failures with --enable-checking=yes,rtl.
18609 * config/microblaze/microblaze.c (microblaze_expand_shift):
18610 Replace GET_CODE test with CONST_INT_P and INTVAL test with
18611 test for const0_rtx.
18612 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
18613 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
18615 2017-03-09 Richard Biener <rguenther@suse.de>
18617 PR tree-optimization/79977
18618 * graphite-scop-detection.c (scop_detection::merge_sese):
18619 Handle the case of extra exits to blocks dominating the entry.
18621 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
18623 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
18626 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
18628 PR rtl-optimization/79949
18629 * lra-constraints.c (process_alt_operands): Check memory when
18630 trying to predict a cycle. Print about the overall increase.
18632 2017-03-09 Richard Biener <rguenther@suse.de>
18634 PR middle-end/79971
18635 * gimple-expr.c (useless_type_conversion_p): Preserve
18636 TYPE_SATURATING for fixed-point types.
18638 2017-03-09 Richard Biener <rguenther@suse.de>
18641 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
18642 alignment of BLKmode params.
18644 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18647 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
18648 (VALL_NO_V2Q): Likewise.
18650 * config/aarch64/aarch64-simd.md
18651 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
18653 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
18654 VALL_NO_V2Q mode iterator.
18655 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
18657 2017-03-09 Martin Liska <mliska@suse.cz>
18659 PR tree-optimization/79631
18660 * tree-chkp-opt.c (chkp_is_constant_addr): Call
18661 tree_int_cst_sign_bit just for INTEGER constants.
18663 2017-03-09 Martin Liska <mliska@suse.cz>
18667 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
18670 2017-03-09 Marek Polacek <polacek@redhat.com>
18673 * tree.c (inchash::add_expr): Handle TREE_VEC.
18675 2017-03-09 Martin Liska <mliska@suse.cz>
18678 (chkp_narrow_size_and_offset): New function.
18679 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
18680 (void chkp_parse_bit_field_ref): New function.
18681 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
18682 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
18684 2017-03-09 Martin Liska <mliska@suse.cz>
18687 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
18688 (chkp_find_bounds_1): Remove gcc_unreachable.
18690 2017-03-09 Jakub Jelinek <jakub@redhat.com>
18693 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
18694 BUILT_IN_SYNC*, determine the access type from the size suffix and
18695 always build a MEM_REF with that type. Handle forgotten
18696 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
18699 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
18700 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
18701 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
18702 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
18703 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
18704 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
18705 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
18706 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
18707 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
18708 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
18709 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
18710 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
18711 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
18712 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
18713 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
18714 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
18715 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
18716 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
18717 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
18718 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
18719 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
18720 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
18721 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
18722 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
18723 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
18724 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
18725 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
18726 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
18727 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
18728 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
18729 definitions outside of __OPTIMIZE__ guarded section.
18732 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
18733 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
18734 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
18735 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
18738 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18740 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
18741 ("vfenez<mode>"): Add missing constraints.
18743 2017-03-08 Martin Sebor <msebor@redhat.com>
18746 * config/nds32/nds32.c (nds32_option_override):
18747 Fix misspelled diagnostic.
18749 2017-03-08 Jakub Jelinek <jakub@redhat.com>
18752 * gimplify.c (gimplify_omp_for): Replace index var in outer
18753 taskloop statement with an artificial variable and add
18754 OMP_CLAUSE_PRIVATE clause for it.
18756 2017-03-08 Richard Biener <rguenther@suse.de>
18758 PR tree-optimization/79955
18759 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
18760 for accesses that are completely outside of the variable.
18762 2017-03-08 Andrew Haley <aph@redhat.com>
18764 PR tree-optimization/79943
18765 * tree-ssa-loop-split.c (compute_new_first_bound): When
18766 calculating the new upper bound, (END-BEG) should be added, not
18769 2017-03-08 Jakub Jelinek <jakub@redhat.com>
18771 * config/avr/avr.md (setmemhi): Make sure match_dup
18772 operand number comes before match_scratch.
18774 2017-03-08 Richard Biener <rguenther@suse.de>
18776 PR tree-optimization/79920
18777 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
18778 with ncopies == 1 to ...
18779 (vect_transform_slp_perm_load): ... here. Properly compute
18780 all element loads by iterating VF times over the group. Do
18781 not handle ncopies (computed in a broken way) in
18782 vect_create_mask_and_perm.
18784 2017-03-08 Jakub Jelinek <jakub@redhat.com>
18787 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
18788 is a uniform vector, use uniform_vector_p return value instead of
18789 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
18791 2017-03-07 Marek Polacek <polacek@redhat.com>
18793 PR middle-end/79809
18794 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
18795 (alloca_call_type): Likewise.
18797 2017-03-07 Martin Liska <mliska@suse.cz>
18799 * gcov.c (process_args): Put comment to correct location.
18801 2017-03-07 Martin Liska <mliska@suse.cz>
18803 PR middle-end/68270
18804 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
18805 Use array_at_struct_end_p instead of DECL_CHAIN (field).
18806 (chkp_narrow_bounds_for_field): Likewise.
18807 (chkp_parse_array_and_component_ref): Pass one more argument to
18810 2017-03-07 Richard Biener <rguenther@suse.de>
18812 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
18815 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
18817 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
18818 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
18820 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18823 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
18824 to end of description.
18825 (PARAM_MAX_STORES_TO_MERGE): Likewise.
18827 2017-03-07 Jakub Jelinek <jakub@redhat.com>
18829 PR rtl-optimization/79901
18830 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
18832 (*avx512f_<code><mode>3<mask_name>): ... this.
18833 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
18834 iterator instead of VI8_AVX2_AVX512BW.
18836 PR rtl-optimization/79901
18837 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
18838 min/max expander, expand it using expand_vec_cond_expr.
18841 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
18844 2017-03-06 Jakub Jelinek <jakub@redhat.com>
18847 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
18848 to void * for PCH reasons.
18849 * dwarf2out.c (output_loc_operands, output_die): Cast
18850 v.val_vec.array to unsigned char *.
18852 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
18855 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
18858 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
18860 PR rtl-optimization/79571
18861 * lra-constraints.c (process_alt_operands): Calculate static
18862 reject and subtract it from overall when only addresses will be
18865 2017-03-06 Julia Koval <julia.koval@intel.com>
18868 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
18869 incoming stack boundary to 128 for 64-bit targets.
18871 2017-03-06 Richard Biener <rguenther@suse.de>
18873 PR tree-optimization/79894
18874 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
18875 to NULL after folding it.
18877 2017-03-06 Richard Biener <rguenther@suse.de>
18879 PR tree-optimization/79824
18880 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
18881 check disabling peeling for gaps.
18883 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
18885 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
18886 attributes): Document gettimeofday.
18888 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
18890 * config/s390/s390.c (s390_option_override_internal): Set
18891 PARAM_MIN_VECT_LOOP_BOUND
18893 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
18895 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
18896 * config/s390/s390.md: Likewise.
18898 2017-03-06 Jakub Jelinek <jakub@redhat.com>
18901 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
18902 (<avx2_avx512>_perm<mode>): Rename to ...
18903 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
18905 (<avx512>_perm<mode>_mask): Rename to ...
18906 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
18908 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
18909 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
18910 instead of VI8F_256_512.
18911 (avx512f_perm<mode>): New define_expand.
18912 (avx512f_perm<mode>_mask): Likewise.
18913 (avx512f_perm<mode>_1<mask_name>): New define_insn.
18914 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
18916 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18918 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
18919 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
18921 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
18923 2017-03-06 Martin Liska <mliska@suse.cz>
18926 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
18927 when having a SSA NAME w/o VAR_DECL assigned to it.
18929 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18931 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
18932 msa_dpsub_<su>_d): Fix MODE for vec_select.
18934 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18936 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
18938 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
18940 2017-03-06 Richard Biener <rguenther@suse.de>
18942 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
18943 * plugin.c (register_plugin_info): Likewise.
18944 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
18946 2017-03-05 Jakub Jelinek <jakub@redhat.com>
18948 * config/i386/sse.md (sse_storehps, sse_storelps,
18949 avx_<castmode><avxsizesuffix>_<castmode>,
18950 avx512f_<castmode><avxsizesuffix>_<castmode>,
18951 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
18952 in condition that at least one operand is not a MEM.
18954 2017-03-03 Jakub Jelinek <jakub@redhat.com>
18956 PR middle-end/79805
18957 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
18958 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
18960 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
18961 gimple_call_nothrow_p flag based on whether original builtin can throw.
18962 If it can, emit following stmts on the fallthrough edge.
18963 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
18964 don't create new bb if inserting just debug stmts on the edge, try to
18965 insert them on the fallthru bb or just reset debug stmts.
18967 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
18970 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
18971 restore recog_data (including the operand rtxes inside it) around
18972 the call to get_insn_template.
18974 2017-03-03 Martin Sebor <msebor@redhat.com>
18976 PR tree-optimization/79699
18977 * context.c (context::~context): Free MPFR caches to avoid
18978 a memory leak on program exit.
18980 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18982 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
18983 Use wide_int::ulow () instead of .elt (0).
18985 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
18987 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
18988 (*pushxf): Limit oF constraint to 32bit targets and add oC
18989 constraint for 64bit targets.
18990 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
18991 (*pushdf): Change rmF constraint to rmC.
18993 2017-03-03 Martin Liska <mliska@suse.cz>
18995 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
18996 Remove unused variable.
18998 2017-03-03 Jakub Jelinek <jakub@redhat.com>
19001 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
19002 is a memory operand, increase num_memory.
19003 (ix86_expand_args_builtin): Likewise.
19005 2017-03-03 Jan Hubicka <jh@suse.cz>
19008 * ipa-devirt.c (maybe_record_node): Properly handle
19009 __cxa_pure_virtual visibility.
19011 2017-03-03 Martin Liska <mliska@suse.cz>
19013 PR tree-optimization/79803
19014 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
19016 (pass_loop_prefetch::execute): Disabled optimization if an
19017 assumption about L1 cache size is not met.
19019 2017-03-03 Martin Liska <mliska@suse.cz>
19021 PR rtl-optimization/79574
19022 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
19023 (hash_scan_set): Likewise.
19024 (dump_hash_table): Likewise.
19025 (hoist_code): Likewise.
19027 2017-03-03 Richard Biener <rguenther@suse.de>
19029 * fixed-value.c (fixed_from_string): Restore use of elt (1)
19030 in place of uhigh ().
19031 (fixed_convert_from_real): Likewise.
19033 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
19036 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
19038 2017-03-03 Richard Biener <rguenther@suse.de>
19040 PR middle-end/79818
19041 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
19042 TYPE_OVERFLOW_UNDEFINED check.
19044 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19046 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
19048 (vector_ae_<mode>_p): Likewise.
19049 (vector_nez_<mode>_p): Likewise.
19050 (vector_ne_v2di_p): Likewise.
19051 (vector_ae_v2di_p): Likewise.
19052 (vector_ne_<mode>_p): Likewise.
19053 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
19055 (vsx_tsqrt<mode>2_fe): Likewise.
19057 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
19060 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
19062 2017-03-02 Jakub Jelinek <jakub@redhat.com>
19064 PR rtl-optimization/79780
19065 * cprop.c (one_cprop_pass): When second and further conditional trap
19066 in a single basic block is turned into an unconditional trap, turn it
19067 into a deleted note to avoid RTL verification failures.
19069 2017-03-02 Richard Biener <rguenther@suse.de>
19071 * fold-const.c (const_binop): Use ulow () instead of elt (0).
19073 2017-03-02 Richard Biener <rguenther@suse.de>
19075 PR tree-optimization/79345
19077 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
19078 param and abort the walk, returning -1 if it is hit.
19079 (walk_aliased_vdefs): Take a limit param and pass it on.
19080 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
19081 defaulting to 0 and return a signed int.
19082 * tree-ssa-uninit.c (struct check_defs_data): New struct.
19083 (check_defs): New helper.
19084 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
19085 about uninitialized memory.
19086 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
19087 bogus uninitialized warning.
19088 (fixed_convert_from_real): Likewise.
19090 2017-03-02 Bin Cheng <bin.cheng@arm.com>
19092 PR tree-optimization/66768
19093 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
19094 iv_use if base object can't be determined.
19096 2017-03-02 Jakub Jelinek <jakub@redhat.com>
19098 PR tree-optimization/79345
19099 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
19100 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
19101 (get_pattern_stats): Initialize it.
19102 * genemit.c (gen_expand): Verify match_scratch numbers come after
19103 match_operand/match_dup numbers.
19104 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
19105 match_scratch numbers.
19106 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
19108 * config/s390/s390.md (trunctdsd2): Likewise.
19110 2017-03-02 Richard Biener <rguenther@suse.de>
19112 * wide-int.h (wide_int_storage::operator=): Implement in terms
19115 2017-03-02 Richard Biener <rguenther@suse.de>
19117 PR tree-optimization/79777
19118 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
19119 the to insert expression to sth existing.
19121 2017-03-01 Martin Sebor <msebor@redhat.com>
19123 PR middle-end/79692
19124 * gimple-ssa-sprintf.c
19125 (directive::known_width_and_precision): New function.
19126 (format_integer): Use it.
19127 (get_mpfr_format_length): Consider the full range of precision
19128 when computing %g output with the # flag. Set the likely byte
19129 count to 3 rather than 1 when precision is indeterminate.
19130 (format_floating): Correct the lower bound of precision.
19132 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19134 * doc/invoke.texi: Document default code model for 64-bit Linux.
19136 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19139 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
19140 udiv rather than div since input pattern is unsigned.
19142 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
19144 * config/i386/i386.c (print_reg): Warn for values of
19145 unsupported size in integer register.
19147 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
19150 * config/rs6000/predicates.md (current_file_function_operand): Do
19151 not allow self calls to be local if the function is replaceable.
19153 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
19156 * config/rs6000/altivec.h (vec_ctz and others): Change the
19157 preprocessor macro that controls conditional compilation from
19158 _ARCH_PWR9 to __POWER9_VECTOR__.
19159 (vec_all_ne): Change parameterization of __altivec_scalar_pred
19160 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
19161 control (instead of _ARCH_PWR9 control) so that template
19162 definition uses power9-specific function.
19163 (vec_any_eq): Likewise.
19164 (vec_all_ne): Change macro definition to use a power9-specific
19165 expansion under #ifdef __POWER9_VECTOR__ control (instead of
19166 _ARCH_PWR9 control).
19167 (vec_any_eq) Likewise.
19168 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
19169 expansion for CMPNEF to remove support for xvcmpnesp instruction.
19170 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
19171 support for xvcmpnedp instruction.
19172 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
19173 macro expansion so that Power9 implementation of vec_all_ne does
19174 not use the AltiVec predicate framework.
19175 (VCMPNEH_P): Likewise.
19176 (VCMPNEW_P): Likewise.
19177 (VCMPNED_P): Likewise.
19178 (VCMPNEFP_P): Likewise.
19179 (VCMPNEDP_P): Likewise.
19180 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
19181 implementation of vec_any_eq to not use AltiVec predicate
19183 (VCMPAEH_P): Likewise.
19184 (VCMPAEW_P): Likewise.
19185 (VCMPAED_P): Likewise.
19186 (VCMPAEFP_P): Likewise.
19187 (VCMPAEDP_P): Likewise.
19188 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
19189 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
19190 not use the AltiVec predicate framework.
19191 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
19192 of vec_any_eq to not use AltiVec predicate framework.
19193 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
19194 support for predefined __POWER9_VECTOR__ macro to indicate that
19195 Power9 instruction selection is enabled.
19196 (altivec_overloaded_builtins): Remove extraneous
19197 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
19198 function argument types RS6000_BTI_bool_V16QI and
19199 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
19200 entry for overloaded function argument types RS6000_BTI_bool_V4SI
19201 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
19202 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
19203 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
19204 Power9 for implementations of vec_cmpne. Change the signature for
19205 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
19206 (representing vec_all_ne) to remove the previously described first
19207 argument of type RS6000_BTI_INTSI, as this was an artifact of
19208 reliance on the AltiVec predicate framework, which is no longer
19209 used in the implementation of these functions. Add
19210 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
19211 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
19212 since, unlike the AltiVec predicate framework implementation, we
19213 do not share function descriptors between vec_alle and vec_anyeq.
19214 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
19215 set of modes that receive special treatment even when
19216 TARGET_P9_VECTOR is true. The special treatment emits code that
19217 does not depend on Power9 instructions.
19218 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
19219 define_expand to not rely on AltiVec predicate framework.
19220 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
19222 (vector_ne_v2di_p): Change this define_expand to not rely on
19223 AltiVec predicate framework.
19224 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
19226 (vector_ne_<mode>_p): Change this define_expand to not rely on
19227 AltiVec predicate framework.
19228 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
19230 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
19231 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
19232 define_insn pattern.
19233 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
19234 define_insn pattern because the xvcmpne<VSs>. instruction is not
19236 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
19237 instruction is not supported.
19239 2017-03-01 Jakub Jelinek <jakub@redhat.com>
19241 * config/nvptx/nvptx.c: Include intl.h.
19243 2017-03-01 Martin Jambor <mjambor@suse.cz>
19246 * ipa-prop.h (ipa_bits): Removed field known.
19247 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
19248 to pointers. Adjusted their comments to warn about their sharing.
19249 (ipcp_transformation_summary): Change bits to a vector of pointers.
19250 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
19251 (ipa_get_ipa_bits_for_value): Declare.
19252 * tree-vrp.h (value_range): Mark as GTY((for_user)).
19253 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
19254 (ipa_bits_hash_table): Likewise.
19255 (ipa_vr_ggc_hash_traits): Likewise.
19256 (ipa_vr_hash_table): Likewise.
19257 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
19258 being pointers and vr_known being removed.
19259 (ipa_set_jf_unknown): Likewise.
19260 (ipa_get_ipa_bits_for_value): New function.
19261 (ipa_set_jfunc_bits): Likewise.
19262 (ipa_get_value_range): New overloaded functions.
19263 (ipa_set_jfunc_vr): Likewise.
19264 (ipa_compute_jump_functions_for_edge): Use the above functions to
19265 construct bits and vr parts of jump functions.
19266 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
19267 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
19269 (ipcp_grow_transformations_if_necessary): Also allocate
19270 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
19272 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
19273 them. Fix too long lines.
19274 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
19275 vr_known being removed.
19276 (ipa_read_jump_function): Use new setter functions to construct bits
19277 and vr parts of jump functions or set them to NULL.
19278 (write_ipcp_transformation_info): Adjust for bits being pointers.
19279 (read_ipcp_transformation_info): Likewise.
19280 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
19282 Include gt-ipa-prop.h.
19283 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
19285 (ipcp_store_bits_results): Likewise.
19286 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
19287 Do not write to existing jump functions but use a temporary instead.
19289 2017-03-01 Jakub Jelinek <jakub@redhat.com>
19292 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
19293 attempt to use its first operand as BIT_FIELD_REF base.
19295 2017-03-01 Richard Biener <rguenther@suse.de>
19297 PR middle-end/79721
19298 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
19299 interpolating formula in wrapping arithmetic.
19300 (chrec_apply): Convert chrec_evaluate return value to wanted type.
19302 2017-03-01 Jakub Jelinek <jakub@redhat.com>
19304 PR tree-optimization/79734
19305 * tree-vect-generic.c (expand_vector_condition): Optimize
19306 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
19307 Handle VEC_COND_EXPR where comparison has different inner width from
19308 type's inner width.
19310 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
19312 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
19313 markup, and similar issues. Remove @opindex entries for things
19314 that aren't options. Add missing -mmpy-option entries.
19316 2017-02-28 Jakub Jelinek <jakub@redhat.com>
19318 PR tree-optimization/79737
19319 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
19320 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
19321 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
19322 instead of byte_size. Formatting fix.
19323 (shift_bytes_in_array_right): Formatting fix.
19325 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
19328 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
19329 condition on optimize for the leaf function test.
19331 2017-02-28 Martin Liska <mliska@suse.cz>
19334 * read-rtl-function.c (function_reader::handle_unknown_directive):
19335 Bail out when one uses -flto.
19337 2017-02-28 Martin Liska <mliska@suse.cz>
19339 * common.opt: Replace space with tabular for options of <number>
19341 * config/i386/i386.opt: Show <number> value for
19342 -mlarge-data-threshold.
19343 * opts.c (print_filtered_help): Do not display number in hexadecimal
19346 2017-02-28 Martin Liska <mliska@suse.cz>
19348 * common.opt: Fix --help=option -Q for options which are of
19351 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
19353 * config/i386/i386.c (print_reg): Error out for values
19354 of 8-bit size in invalid integer register.
19356 2017-02-28 Martin Sebor <msebor@redhat.com>
19358 PR tree-optimization/79691
19359 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
19361 2017-02-28 Jakub Jelinek <jakub@redhat.com>
19364 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
19365 gcc_unreachable with output_operand_lossage.
19367 2017-02-28 Richard Biener <rguenther@suse.de>
19369 PR tree-optimization/79740
19370 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
19372 (visit_nary_op): Insert the nary into the hashtable if we
19373 pattern-matched sth.
19374 * tree-ssa-pre.c (eliminate_insert): Robustify.
19376 2017-02-28 Richard Biener <rguenther@suse.de>
19378 PR middle-end/79731
19379 * fold-const.c (decode_field_reference): Reject out-of-bound
19382 2017-02-28 Jakub Jelinek <jakub@redhat.com>
19384 * config/i386/i386.c: Include intl.h.
19385 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
19386 instead of just cond ? "..." : "...".
19387 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
19388 * coverage.c (read_counts_file): Likewise.
19389 * omp-offload.c: Include intl.h.
19390 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
19391 of just cond ? "..." : "...".
19392 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
19393 of just cond ? "..." : "...".
19395 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
19398 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
19400 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
19401 'tune for' CPU name.
19402 * config/arm/arm-cpu-data.h: Regenerated.
19404 2017-02-28 Richard Biener <rguenther@suse.de>
19406 PR tree-optimization/79732
19407 * tree-inline.c (expand_call_inline): Do not shadow var.
19409 2017-02-28 Richard Biener <rguenther@suse.de>
19411 PR tree-optimization/79723
19412 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
19413 address-space properly.
19415 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
19417 * doc/optinfo.texi (Optimization groups): Fix option used for
19419 * doc/invoke.texi (-fopt-info): Document "omp".
19420 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
19421 (OPTGROUP_ALL): Add OPTGROUP_OMP.
19422 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
19423 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
19424 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
19426 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
19428 * dumpfile.c (optgroup_options): Instead of "openmp", associate
19429 OPTGROUP_OMP with "omp".
19431 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
19434 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
19435 for arithmetic shift of unsigned V2DI.
19437 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
19439 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
19440 arc/linux.h headers.
19441 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
19442 (LINK_SPEC): Likewise.
19443 (ARC_TLS_EXTRA_START_SPEC): Likewise.
19444 (EXTRA_SPECS): Likewise.
19445 (STARTFILE_SPEC): Likewise.
19446 (ENDFILE_SPEC): Likewise.
19447 (LIB_SPEC): Likewise.
19448 (TARGET_SDATA_DEFAULT): Likewise.
19449 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
19450 (MULTILIB_DEFAULTS): Likewise.
19451 (DWARF2_UNWIND_INFO): Likewise.
19452 * config/arc/big.h: New file.
19453 * config/arc/elf.h: Likewise.
19454 * config/arc/linux.h: Likewise.
19455 * config/arc/t-uClibc: Remove.
19457 2017-02-27 Bin Cheng <bin.cheng@arm.com>
19459 PR tree-optimization/77536
19460 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
19461 (tree_transform_and_unroll_loop): Use above function to compute the
19462 estimated niter of unrolled loop and use it when scaling profile.
19463 Also use count info rather than frequency if it's non-zero.
19464 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
19465 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
19466 (vect_transform_loop): Call above function.
19468 2017-02-27 Richard Biener <rguenther@suse.de>
19470 PR tree-optimization/45397
19471 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
19472 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
19473 (visit_nary_op): Add pattern matching for CSEing sign-changed
19474 or truncated operations with wider ones.
19476 2017-02-27 Richard Biener <rguenther@suse.de>
19478 PR tree-optimization/79690
19479 * tree-vect-stmts.c (vectorizable_store): Use vector type
19480 built from the DR with address-space.
19482 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
19484 * doc/invoke.texi (Optimize Options): Refine the description
19485 of asan-use-after-return.
19487 2017-02-25 Alan Modra <amodra@gmail.com>
19489 PR rtl-optimization/79584
19490 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
19491 base, not ad->base_term, the reg within base. Remove assertion
19492 that ad->base == ad->base_term. Replace gen_int_mode using
19493 bogus mode with const0_rtx.
19495 2017-02-25 Jakub Jelinek <jakub@redhat.com>
19497 PR middle-end/79396
19498 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
19499 FMA_EXPR like tcc_binary or tcc_unary.
19501 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
19504 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
19506 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
19507 (output_loc_operands): Handle DW_OP_call_ref and
19508 DW_OP_GNU_variable_value.
19509 (struct variable_value_struct): New type.
19510 (struct variable_value_hasher): Likewise.
19511 (variable_value_hash): New variable.
19512 (string_types): Remove.
19513 (copy_loc_descr): New function.
19514 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
19515 (prepend_loc_descr_to_each): New function.
19516 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
19517 instead of add_loc_descr_to_each if the first argument is single
19518 location list and the second has multiple.
19519 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
19520 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
19521 when looking for variable value which doesn't have other location info.
19522 (loc_list_from_tree): Formatting fix.
19523 (gen_array_type_die): Simplify DW_AT_string_length handling.
19524 (adjust_string_types): Remove.
19525 (gen_subprogram_die): Don't call adjust_string_types nor test/set
19526 string_types. Call resolve_variable_values.
19527 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
19528 (resolve_addr_in_expr): Likewise. Add A argument.
19529 (copy_deref_exprloc): Remove deref argument. Adjust for the
19530 original expression being DW_OP_GNU_variable_value with optionally
19531 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
19532 optionally after it.
19533 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
19534 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
19535 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
19536 (variable_value_hasher::hash, variable_value_hasher::equal): New
19538 (resolve_variable_value_in_expr, resolve_variable_value,
19539 resolve_variable_values, note_variable_value_in_expr,
19540 note_variable_value): New functions.
19541 (dwarf2out_early_finish): Call note_variable_value on all toplevel
19544 2017-02-24 Jakub Jelinek <jakub@redhat.com>
19547 * opts.h (handle_generated_option): Add GENERATED_P argument.
19548 * opts-common.c (handle_option): Adjust function comment.
19549 (handle_generated_option): Add GENERATED_P argument, pass it to
19551 (control_warning_option): Pass false to handle_generated_option
19553 * opts.c (maybe_default_option): Pass true to handle_generated_option
19555 * optc-gen.awk: Likewise.
19557 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19559 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
19560 a REG, look at the REG it is a SUBREG of.
19561 (splitter for cmpeqsi_t): Ditto.
19563 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19565 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
19566 the special USEs with the pattern of the insn, not the insn itself.
19568 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
19571 * doc/invoke.texi: Document -mload-store-pairs.
19573 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19574 Sandra Loosemore <sandra@codesourcery.com>
19576 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
19577 argument isn't a CONST_INT.
19578 (nios2_alternate_compare_const): Assert op is a CONST_INT.
19579 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
19580 (nios2_validate_compare): Bypass alternate compare logic if *op2
19581 is not a CONST_INT.
19582 (ldstwm_operation_p): Return false if first_base is not a REG or
19583 if first_offset is not a CONST_INT.
19585 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19587 * config/cris/cris.md: Use correct operand in a define_peephole2.
19589 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19591 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
19593 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19595 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
19596 this_insn if it is an INSN or JUMP_INSN.
19597 (force_offsettable): Look at base, not at addr.
19598 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
19599 on things that aren't necessarily CONST_INTs.
19601 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
19603 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
19604 -mfpmath=sse is the default also for x86-32 targets with SSE2
19605 instruction set when @option{-ffast-math} is enabled
19607 2017-02-24 Jeff Law <law@redhat.com>
19609 PR rtl-optimizatoin/79286
19610 * ira.c (update_equiv_regs): Drop may_trap_p exception to
19613 2017-02-24 Richard Biener <rguenther@suse.de>
19615 PR tree-optimization/79389
19616 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
19619 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
19621 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
19622 function comment to reflect reality.
19623 (loop_exits_before_overflow): Fix typo in function description.
19625 2017-02-24 Richard Biener <rguenther@suse.de>
19627 PR tree-optimization/79389
19628 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
19629 properly that a threading opportunity exists. Detect conditional
19630 copy/constant propagation opportunities.
19632 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
19634 * config/visium/visium.md (type): Add trap.
19635 (b): New mode attribute.
19636 (*btst): Rename into...
19637 (*btst<mode>): ...this and adjust.
19638 (*cbranchsi4_btst_insn): Rename into...
19639 (*cbranch<mode>4_btst_insn): ...this and adjust.
19640 (trap): New define_insn.
19642 2017-02-23 Jakub Jelinek <jakub@redhat.com>
19644 PR tree-optimization/79389
19645 * ifcvt.c (struct noce_if_info): Add rev_cond field.
19646 (noce_reversed_cond_code): New function.
19647 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
19648 reversed_comparison_code. Formatting fix.
19649 (noce_try_store_flag): Test rev_cond != NULL in addition to
19650 reversed_comparison_code.
19651 (noce_try_store_flag_constants): Likewise.
19652 (noce_try_store_flag_mask): Likewise.
19653 (noce_try_addcc): Use rev_cond if non-NULL instead of
19654 reversed_comparison_code.
19655 (noce_try_cmove_arith): Likewise. Formatting fixes.
19656 (noce_try_minmax, noce_try_abs): Clear rev_cond.
19657 (noce_find_if_block): Initialize rev_cond.
19658 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
19659 instead of false as last argument never attempt to reverse it
19662 2017-02-23 Bin Cheng <bin.cheng@arm.com>
19664 PR tree-optimization/79663
19665 * tree-predcom.c (combine_chains): Process refs in reverse order
19666 only for ZERO length chains, and add explaining comment.
19668 2017-02-23 Jeff Law <law@redhat.com>
19670 PR tree-optimization/79578
19671 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
19672 in call to operand_equal_p.
19674 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
19677 * config/i386/cpuid.h: Fix another undefined behavior.
19679 2017-02-23 Richard Biener <rguenther@suse.de>
19681 PR tree-optimization/79683
19682 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
19683 vector types for data-refs.
19685 2017-02-23 Martin Liska <mliska@suse.cz>
19687 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
19689 2017-02-23 Jakub Jelinek <jakub@redhat.com>
19691 PR middle-end/79665
19692 * internal-fn.c (get_range_pos_neg): Moved to ...
19693 * tree.c (get_range_pos_neg): ... here. No longer static.
19694 * tree.h (get_range_pos_neg): New prototype.
19695 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
19696 are known to be in between 0 and signed maximum inclusive, try to
19697 expand both unsigned and signed divmod and use the cheaper one from
19700 2017-02-22 Jeff Law <law@redhat.com>
19702 PR tree-optimization/79578
19703 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
19704 to compare base operands.
19706 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
19709 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
19710 gpc_reg_operand instead of fpr_reg_operand.
19712 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
19714 * config/mips/mips.c (mips_return_in_memory): Force FP
19715 vector types to be returned in memory for o32 ABI.
19717 2017-02-22 Jakub Jelinek <jakub@redhat.com>
19719 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
19720 instead of DW_TAG_member for static data member declarations and don't
19721 set no_linkage_name for static inline data members.
19722 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
19725 2017-02-22 Martin Liska <mliska@suse.cz>
19727 * doc/invoke.texi: Replace inequality signs with square brackets
19730 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19732 PR tree-optimization/68644
19733 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
19735 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
19738 * lra-constraints.c (simplify_operand_subreg): Handle
19739 WORD_REGISTER_OPERATIONS targets.
19741 2017-02-22 Jakub Jelinek <jakub@redhat.com>
19744 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
19745 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
19746 elimination by swapping fld*.
19748 2017-02-22 Richard Biener <rguenther@suse.de>
19750 PR tree-optimization/79673
19751 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
19752 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
19753 irrelevant address-space qualifiers and avoiding a
19754 ADDR_SPACE_CONVERT_EXPR from fold_convert.
19756 2017-02-22 Richard Biener <rguenther@suse.de>
19758 PR tree-optimization/79666
19759 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
19760 to not symbolically negate if that may introduce undefined
19763 2017-02-22 Martin Liska <mliska@suse.cz>
19766 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
19767 * data-streamer-out.c (streamer_write_gcov_count_stream):
19769 * value-prof.c (stream_out_histogram_value): Make assert more
19770 precise based on type of counter.
19772 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
19775 * config/i386/i386.md (standard_x87sse_constant_load splitter):
19776 Use nonimmediate_operand instead of memory_operand for operand 1.
19777 (float-extend standard_x87sse_constant_load splitter): Ditto.
19779 2017-02-21 Jeff Law <law@redhat.com>
19781 PR tree-optimization/79621
19782 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
19783 blocks with edges to themselves.
19785 2017-02-21 Jakub Jelinek <jakub@redhat.com>
19788 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
19789 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
19790 Use gimple_call_builtin_p.
19793 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
19794 on temporarily removed DEBUG_INSNs.
19796 PR tree-optimization/79649
19797 * tree-loop-distribution.c (classify_partition): Give up on
19798 non-generic address space loads/stores.
19800 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
19802 * doc/loop.texi (Loop manipulation): Remove nonexistent
19803 tree_ssa_loop_version from the documentation.
19804 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
19806 2017-02-21 Jakub Jelinek <jakub@redhat.com>
19809 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
19810 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
19811 * config/rs6000/rs6000.c: Include except.h.
19812 (rs6000_expand_split_stack_prologue): Call
19813 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
19815 2017-02-21 Martin Jambor <mjambor@suse.cz>
19818 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
19819 have been analyzed.
19821 2017-02-21 Martin Jambor <mjambor@suse.cz>
19823 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
19824 for backward compatibility only.
19825 * doc/invoke.texi (Option Summary): Remove all references to
19826 -fipa-cp-alignment.
19828 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
19832 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19834 * lra-constraints.c (curr_insn_transform): Handle
19835 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
19837 2017-02-21 Martin Liska <mliska@suse.cz>
19839 * config/i386/i386.opt: Replace -masm-dialect with -masm.
19841 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
19843 PR translation/79638
19844 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
19846 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
19849 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
19850 (arm_function_ok_for_sibcall): Return false for an indirect call by
19851 descriptor if all the argument registers are used.
19852 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
19853 alignment of the function.
19855 2017-02-21 Jakub Jelinek <jakub@redhat.com>
19857 PR tree-optimization/61441
19858 * simplify-rtx.c (simplify_const_unary_operation): For
19859 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
19860 the sNaN unmodified.
19862 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
19864 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
19865 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
19866 instead of SYSTEM_HEADER_DIR.
19868 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
19869 Martin Liška <mliska@suse.cz>
19871 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
19872 Fix typos and grammar, use active voice, and clarify.
19874 2017-02-20 Marek Polacek <polacek@redhat.com>
19876 PR middle-end/79537
19877 * gimplify.c (gimplify_expr): Handle unused *&&L;.
19880 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
19882 2017-02-20 Jakub Jelinek <jakub@redhat.com>
19885 * config/i386/i386.c (ix86_expand_builtin): Handle
19886 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
19887 ix86_builtins_isa[fcode].isa as a requirement of those
19888 flags and any other flag in the bitmask.
19889 (ix86_init_mmx_sse_builtins): Use 0 instead of
19890 ~OPTION_MASK_ISA_64BIT as mask.
19891 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
19892 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
19893 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
19894 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
19896 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19899 * lra-constraints.c (split_reg): Check requested split mode
19900 is supported by the register.
19902 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19904 * lra-constraints.c (simplify_operand_subreg): Remove early
19907 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19910 * lra-constraints.c (curr_insn_transform): Tighten condition
19911 for converting SUBREG reloads from OP_OUT to OP_INOUT.
19913 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19916 * lra-constraints.c (curr_insn_transform): Handle
19917 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
19919 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
19922 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
19924 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
19926 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
19929 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
19932 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19934 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
19937 2017-02-18 Jakub Jelinek <jakub@redhat.com>
19939 * final.c (last_columnnum, override_columnnum): New variables.
19940 (final_start_function): Set last_columnnum, pass it to begin_prologue
19941 hook and pass 0 to dwarf2out_begin_prologue.
19942 (final_scan_insn): Update override_columnnum. Pass last_columnnum
19943 to source_line debug hook.
19944 (notice_source_line): Compute last_columnnum and for debug_column_info
19945 return true on column changes.
19946 * debug.h (struct gcc_debug_hooks): Add column argument to
19947 source_line and begin_prologue hooks.
19948 (debug_nothing_int_charstar_int_bool): Remove prototype.
19949 (debug_nothing_int_int_charstar,
19950 debug_nothing_int_int_charstar_int_bool): New prototypes.
19951 (dwarf2out_begin_prologue): Add column argument.
19952 * debug.c (do_nothing_debug_hooks): Adjust source_line and
19953 begin_prologue hooks.
19954 (debug_nothing_int_charstar_int_bool): Remove.
19955 (debug_nothing_int_int_charstar,
19956 debug_nothing_int_int_charstar_int_bool): New functions.
19957 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
19958 through to dwarf2out_source_line.
19959 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
19960 (dwarf2out_source_line): Add column argument, emit it if requested.
19961 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
19963 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
19964 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
19965 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
19966 through to dwarf2out_begin_prologue.
19967 (vmsdbgout_source_line): Add column argument, pass it through to
19968 dwarf2out_source_line.
19969 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
19970 dbxout_source_line caller.
19971 (dbxout_source_line): Add column argument.
19973 * common.opt (gno-column-info, gcolumn-info): New options.
19974 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
19975 (check_die): Also test for multiple DW_AT_decl_column attributes.
19976 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
19977 DW_AT_decl_column if requested.
19978 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
19980 (gen_variable_die): Likewise.
19981 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
19982 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
19985 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
19986 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
19987 (ix86_handle_option): Handle OPT_m3dnowa.
19988 * doc/invoke.texi (-m3dnowa): Document.
19989 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
19990 -m3dnowa instead of -m3dnow -march=athlon.
19993 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
19994 instead of gcc_assert for K, r and R code checks. Formatting fixes.
19996 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19999 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
20000 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
20001 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
20002 generator for vsx_xxpermdi_<mode>_be.
20003 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
20004 force big-endian semantics.
20005 (vsx_xxpermdi_<mode>_be): New define_expand with same
20006 implementation as previous version of vsx_xxpermdi_<mode>.
20008 2017-02-17 Jakub Jelinek <jakub@redhat.com>
20010 PR tree-optimization/79327
20011 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
20012 variable, its initialization and use.
20014 2017-02-17 Julia Koval <julia.koval@intel.com>
20016 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
20017 (OPTION_MASK_ISA_PKU_UNSET): New.
20018 (ix86_handle_option): Handle -mrdpid.
20019 * config/i386/cpuid.h (bit_RDPID): New.
20020 * config/i386/driver-i386.c (host_detect_local_cpu):
20021 Detect RDPID feature.
20022 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
20023 * config/i386/i386-c.c (ix86_target_macros_internal):
20025 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
20026 (ix86_valid_target_attribute_inner_p): Add "rdpid".
20027 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
20028 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
20029 * config/i386/i386.md (define_insn "rdpid"): New.
20030 * config/i386/i386.opt Add -mrdpid.
20031 * config/i386/immintrin.h (_rdpid_u32): New.
20033 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
20035 PR rtl-optimization/79541
20036 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
20037 instead of transforming it into USE.
20039 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
20041 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
20042 If HONOR_SNANS (SFmode) force the input to a register.
20043 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
20044 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
20045 an frsp or similar insn.
20047 2017-02-17 Martin Liska <mliska@suse.cz>
20049 PR rtl-optimization/79577
20050 * params.def (selsched-max-sched-times): Increase minimum to 1.
20052 2017-02-17 Martin Liska <mliska@suse.cz>
20054 PR rtl-optimization/79574
20055 * gcse.c (want_to_gcse_p): Prevent integer overflow.
20057 2017-02-17 Martin Liska <mliska@suse.cz>
20059 PR tree-optimization/79529
20060 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
20061 ssa_defined_default_def_p to handle cases which are implicitly
20063 * tree-ssa.c (ssa_defined_default_def_p): New function.
20064 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
20065 which are implicitly defined.
20066 * tree-ssa.h (ssa_defined_default_def_p): Declare.
20068 2017-02-17 Richard Biener <rguenther@suse.de>
20070 PR middle-end/79576
20071 * params.def (max-ssa-name-query-depth): Limit to 10.
20073 2017-02-17 Richard Biener <rguenther@suse.de>
20075 PR tree-optimization/79552
20076 * tree-ssa-structalias.c (visit_loadstore): Properly verify
20079 2017-02-17 Richard Biener <rguenther@suse.de>
20082 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
20084 2017-02-17 Marek Polacek <polacek@redhat.com>
20086 PR middle-end/79536
20087 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
20088 (fold_negate_expr): New wrapper.
20090 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
20092 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
20093 Correct terminology and de-emphasize pre-standard behavior.
20095 2017-02-16 Alan Modra <amodra@gmail.com>
20097 PR rtl-optimization/79286
20098 * ira.c (def_dominates_uses): New function.
20099 (update_equiv_regs): Don't create an equivalence for insns that
20100 may trap where the register def does not dominate the use.
20102 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
20104 PR rtl-optimization/78127
20105 * lra.c (lra): Call lra_eliminate before finish the loop after
20108 2017-02-16 Richard Biener <rguenther@suse.de>
20110 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
20112 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
20113 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
20114 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
20115 (isl_val_int_from_wi): New function.
20116 (extract_affine_gmp): Rename to ...
20117 (extract_affine_wi): ... this, take a widest_int.
20118 (extract_affine_int): Just wrap extract_affine_wi.
20119 (add_param_constraints): Use isl_val_int_from_wi.
20120 (add_loop_constraints): Likewise, and extract_affine_wi.
20122 2017-02-15 Jeff Law <law@redhat.com>
20124 PR middle-end/79521
20125 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
20126 ira_init_register_move_cost_if_necessary.
20128 2017-02-15 Martin Sebor <msebor@redhat.com>
20130 PR middle-end/32003
20131 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
20132 removed in a prior commit.
20134 2017-02-15 Bin Cheng <bin.cheng@arm.com>
20136 PR tree-optimization/79347
20137 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
20138 counters during peeling.
20140 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
20142 * Makefile.in (site.exp): Remove "set ISLVER".
20144 2017-02-15 Jakub Jelinek <jakub@redhat.com>
20147 * real.c (real_from_integer): Call real_convert even for decimal.
20149 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
20152 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
20154 2017-02-14 Andrew Pinski <apinski@cavium.com>
20156 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
20157 cores and change the partno/implementer to be correct.
20158 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
20159 the 'B" as the implementer.
20160 * config/aarch64/aarch64-tune.md: Regenerate.
20162 2017-02-14 Carl Love <cel@us.ibm.com>
20164 * config/rs6000/rs6000.c: Add case statement entry to make the
20165 xvcvuxdsp built-in argument unsigned.
20166 * config/rs6000/vsx.md: Fix the source and return operand types so they
20167 match the instruction definitions from the ISA document. Fix typo
20168 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
20171 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
20174 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
20175 member early_clobber_alts.
20176 * lra-lives.c (reg_early_clobber_p): New.
20177 (process_bb_lives): Use it.
20178 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
20179 (debug_operand_data): Initialize early_clobber_alts.
20180 (setup_operand_alternative): Set up early_clobber_alts.
20181 (collect_non_operand_hard_regs): Ditto. Pass early clobber
20182 alternatives to new_insn_reg.
20183 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
20185 (lra_update_insn_regno_info): Pass the new arg.
20187 2017-02-14 Jakub Jelinek <jakub@redhat.com>
20189 PR middle-end/79505
20190 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
20191 (new_oacc_loop_raw): Don't clear already cleared fields.
20194 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
20195 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
20196 _mm512_prefetch_i64gather_ps): New inline functions and macros.
20198 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
20201 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
20203 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
20206 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
20207 the extra instruction to the right place to store 128-bit constant
20210 2017-02-14 Martin Sebor <msebor@redhat.com>
20212 PR middle-end/79448
20213 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
20214 warning for strings of unknown length.
20216 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
20218 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
20220 2017-02-14 Jeff Law <law@redhat.com>
20223 * ira-costs.c (scan_one_insn): Initialize register move costs
20224 for pseudos seen in USE/CLOBBER insns.
20226 PR tree-optimization/79095
20227 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
20228 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
20229 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
20230 if the operands are known to be not equal, then the resulting range
20232 (intersect_ranges): If the new range is ~[0,0] and the old range is
20233 wide, then prefer ~[0,0].
20234 * tree-vrp.c (overflow_comparison_p_1): New function.
20235 (overflow_comparison_p): New function.
20236 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
20237 if NAME is used in an overflow test.
20238 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
20239 overflow check that can be expressed as an equality test, then adjust
20240 ops to be that equality test.
20242 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20244 * config/s390/s390-builtin-types.def: Remove flags argument.
20245 * config/s390/s390.c (s390_init_builtins): Likewise.
20247 2017-02-14 Martin Liska <mliska@suse.cz>
20249 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
20250 vector. Fix trailing white spaces.
20252 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
20254 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
20257 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20259 PR rtl-optimization/68664
20260 * config/arm/arm.c (arm_sched_can_speculate_insn):
20261 New function. Declare prototype.
20262 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
20264 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20266 PR rtl-optimization/68664
20267 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
20269 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
20271 2017-02-14 Amit Pawar <amit.pawar@amd.com>
20273 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
20274 max skip bytes for function, loop and jump.
20276 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20278 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
20279 ABS_EXPR for gimple dump.
20281 2017-02-14 Jakub Jelinek <jakub@redhat.com>
20284 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
20286 PR tree-optimization/79408
20287 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
20288 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
20289 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
20290 also if rhs1 is INTEGER_CST.
20292 2017-02-14 Richard Biener <rguenther@suse.de>
20294 PR middle-end/79432
20295 * tree-into-ssa.c (insert_phi_nodes): When the function can
20296 have abnormal edges rewrite SSA names with broken use-def
20297 dominance out of SSA and register them for PHI insertion.
20299 2017-02-13 Martin Sebor <msebor@redhat.com>
20301 PR middle-end/79496
20302 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
20303 clearing info.nowrite flag when snprintf size argument is a range.
20305 2017-02-13 Jakub Jelinek <jakub@redhat.com>
20307 * cprop.c (cprop_jump): Add missing space in string literal.
20308 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
20309 (get_constraint_for_component_ref): Likewise.
20310 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
20311 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
20312 * lra-constraints.c (process_alt_operands): Likewise.
20313 * ipa-inline.c (inline_small_functions): Likewise.
20314 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
20315 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
20316 * trans-mem.c (diagnose_tm_1_op): Likewise.
20317 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
20318 (grid_parallel_clauses_gridifiable): Likewise.
20320 * config/nvptx/mkoffload.c (process): Add space in between
20323 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
20324 "MOD4_SSE_REGS" and "ALL_REGS".
20326 * spellcheck.c (test_data): Add , in between "foo" and "food".
20328 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20331 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
20332 boundary crossing check and subsequent code generation agree.
20334 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20336 * config/aarch64/aarch64.c (has_memory_op): Delete.
20337 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
20340 2017-02-13 Jakub Jelinek <jakub@redhat.com>
20342 PR rtl-optimization/79388
20343 PR rtl-optimization/79450
20344 * combine.c (distribute_notes): When removing TEM_INSN for which
20345 corresponding dest has last value recorded, invalidate that last
20348 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20350 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
20351 of explicit '@'. Add missing assembly comment marker on branch costs
20354 2017-02-13 Nathan Sidwell <nathan@acm.org>
20356 * gengtype-lex.l (<in_struct>): Add '/'.
20358 2017-02-13 Martin Liska <mliska@suse.cz>
20361 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
20363 2017-02-13 Richard Biener <rguenther@suse.de>
20365 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
20367 * configure: Re-generate.
20368 * config.in: Likewise.
20369 * graphite-dependences.c: Simplify as if
20370 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
20371 * graphite-isl-ast-to-gimple.c: Likewise.
20372 * graphite-optimize-isl.c: Likewise.
20373 * graphite-poly.c: Likewise.
20374 * graphite-sese-to-poly.c: Likewise.
20375 * graphite.h: Likewise.
20376 * toplev.c: Include isl/version.h and use isl_version () for
20377 printing the ISL version.
20378 * doc/install.texi: Update ISL requirement.
20380 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
20382 * doc/standards.texi (Standards): Update reference to
20385 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
20387 * doc/extend.texi (Named Address Spaces): sourceware.org now
20389 * doc/install.texi (Binaries): Ditto.
20392 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
20394 * doc/cpp.texi: Replace "stringify"/"stringification" with C
20395 standard terminology "stringize"/"stringizing" throughout.
20396 * doc/cppinternals.texi: Likewise.
20398 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
20400 * doc/extend.texi: Fix some spelling mistakes and typos.
20401 * doc/invoke.texi: Likewise.
20403 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
20406 * params.def (inline-min-speedup) Change from 10 to 8.
20408 2017-02-11 Jakub Jelinek <jakub@redhat.com>
20410 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
20413 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
20416 * ipa-inline-analysis.c (get_minimal_bb): New function.
20417 (record_modified): Use it.
20418 (remap_edge_change_prob): Handle also ancestor functions.
20420 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
20422 * doc/contrib.texi (Contributors): Remove broken link into
20423 the Mauve CVS repository.
20425 2017-02-11 Jakub Jelinek <jakub@redhat.com>
20427 PR middle-end/79454
20428 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
20429 result computation whenever lhs doesn't have vector mode, not
20430 just when it has BLKmode.
20432 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
20434 * doc/makefile.texi (profiledbootstrap): Refer to the
20435 installation instructions only in textual form.
20437 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20440 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
20442 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
20444 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
20445 (Specific): Update mingw-w64 reference.
20447 (Specific): Remove broken link to Renesas RX processor.
20449 2017-02-10 Richard Biener <rguenther@suse.de>
20451 * toplev.c (process_options): Do not mention obsolete graphite
20452 options when printing sorry message about missing graphite support.
20453 Mention -floop-nest-optimize.
20455 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
20457 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
20458 (vtst_p16): Likewise.
20459 (vtstq_p8): Likewise.
20460 (vtstq_p16): Likewise.
20462 (vtstq_p64): Likewise.
20463 * config/arm/arm_neon.h (vgetq_lane_p64): New.
20464 (vset_lane_p64): New.
20465 (vsetq_lane_p64): New.
20467 2017-02-10 Jakub Jelinek <jakub@redhat.com>
20469 PR tree-optimization/79411
20470 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
20471 stmt operands are SSA_NAMEs used in abnormal phis.
20472 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
20475 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
20478 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
20481 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
20483 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
20485 2017-02-09 Jakub Jelinek <jakub@redhat.com>
20487 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
20491 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
20492 not arbitrary TREE_CONSTANT.
20495 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
20496 "omp declare target link" attribute unless is_global_var.
20497 * omp-offload.c (find_link_var_op): Likewise.
20499 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
20500 Chung-Lin Tang <cltang@codesourcery.com>
20502 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
20504 (gimplify_adjust_omp_clauses): Don't delete TILE.
20505 (gimplify_omp_for): Deal with TILE.
20506 * internal-fn.c (expand_GOACC_TILE): New function.
20507 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
20509 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
20510 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
20512 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
20513 avoid DIV for outermost collapse var.
20514 (expand_oacc_for): Insert tile element loop as needed. Adjust.
20515 Remove out of date comments, fix whitespace.
20516 * omp-general.c (omp_extract_for_data): Deal with tiling.
20517 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
20518 adjust OLF_DIM_BASE value.
20519 (struct omp_for_data): Add tiling field.
20520 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
20521 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
20522 for auto loops. Remove default auto determining, moved to
20523 oacc_loop_fixed_partitions.
20524 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
20525 stmts, add e_mask field.
20526 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
20527 (oacc_thread_numbers): Use oacc_dim_call.
20528 (oacc_xform_tile): New.
20529 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
20530 (finish_oacc_loop): Adjust for ifns vector.
20531 (oacc_loop_discover_walk): Append loop abstraction sites to list,
20532 add case for GOACC_TILE fns.
20533 (oacc_loop_xform_loop): Delete.
20534 (oacc_loop_process): Iterate over call list directly, and add
20535 handling for GOACC_TILE fns.
20536 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
20538 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
20539 vector partitioning to outer loops. Assign 2 partitions to loops
20540 when available. Add TILE handling.
20541 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
20542 (execite_oacc_device_lower): Process GOACC_TILE fns,
20543 ignore unknown specs.
20544 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
20545 * tree.c (omp_clause_num_ops): Adjust TILE ops.
20546 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
20548 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
20550 * configure.ac (ACX_BUGURL): Update.
20551 * configure: Regenerate.
20553 2017-02-09 Richard Biener <rguenther@suse.de>
20555 PR tree-optimization/69823
20556 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
20557 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
20559 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
20561 * config/arc/arc-c.def: Add __NPS400__ definition.
20562 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
20563 (TARGET_NPS400): Define.
20565 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
20567 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
20569 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
20570 pointer, arch_info.
20571 (arc_cpu_types): Fill the arch_info field with a pointer into the
20572 arc_arch_types table.
20573 (arc_selected_cpu): Declare.
20574 * config/arc/arc.c (arc_selected_cpu): Make global.
20575 (arc_selected_arch): Delete.
20576 (arc_base_cpu): Delete.
20577 (arc_override_options): Remove references to deleted variables,
20578 update access to arch information.
20579 (ARC_OPT): Update access to arch information.
20580 (ARC_OPTX): Likewise.
20581 * config/arc/arc.h (arc_base_cpu): Remove declaration.
20582 (TARGET_ARC600): Update access to arch information.
20583 (TARGET_ARC601): Likewise.
20584 (TARGET_ARC700): Likewise.
20585 (TARGET_EM): Likewise.
20586 (TARGET_HS): Likewise.
20587 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
20590 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
20593 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
20594 condition/operands for integer GE/LE/GEU/LEU operations.
20596 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
20598 PR translation/79397
20599 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
20602 2017-02-08 Martin Jambor <mjambor@suse.cz>
20605 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
20606 whether allocation happened.
20607 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
20608 nothing was allocated.
20610 2017-02-08 Jakub Jelinek <jakub@redhat.com>
20612 PR tree-optimization/79408
20613 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
20614 constant, but SSA_NAME with a known integer range, use the minimum
20615 of that range instead of op1 to determine if modulo can be replaced
20616 with its first operand.
20618 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20620 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
20622 2017-02-08 Richard Biener <rguenther@suse.de>
20624 PR tree-optimization/71824
20625 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
20626 Check all loops contained in the merged region.
20628 2017-02-07 Andrew Pinski <apinski@cavium.com>
20630 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
20632 2017-02-07 Andrew Pinski <apinski@cavium.com>
20634 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
20635 (thunderxt88): Likewise.
20636 (thunderxt81): Disable LSE and change v8.1 to v8.
20637 (thunderxt83): Likewise.
20639 2017-02-07 Jakub Jelinek <jakub@redhat.com>
20640 Richard Biener <rguenther@suse.de>
20642 PR middle-end/79399
20643 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
20644 type from int to size_t.
20645 * ira-costs.c (struct_costs_size): Change type from int to size_t.
20647 2017-02-07 Jakub Jelinek <jakub@redhat.com>
20649 PR rtl-optimization/79386
20650 * cprop.c (bypass_conditional_jumps): Initialize
20651 bypass_last_basic_block already before splitting bbs after
20652 unconditional traps...
20653 (bypass_conditional_jumps): ... rather than here.
20656 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
20657 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
20658 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
20659 fix -masm=intel patterns.
20661 2017-02-07 Richard Biener <rguenther@suse.de>
20663 PR tree-optimization/79256
20664 PR middle-end/79278
20665 * builtins.c (get_object_alignment_2): Use min_align_of_type
20666 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
20667 and ADJUST_FIELD_ALIGN.
20669 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
20671 * doc/tm.texi: Regenerate.
20672 * stor-layout.c (layout_decl): Adjust.
20673 (update_alignment_for_field): Likewise.
20674 (place_field): Likewise.
20675 (min_align_of_type): Likewise.
20676 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
20677 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
20678 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
20679 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
20680 * config/frv/frv.c (frv_adjust_field_align): Likewise.
20681 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
20682 * config/i386/i386.c (x86_field_alignment): Likewise.
20683 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
20684 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
20685 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
20686 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
20687 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
20688 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
20692 2017-01-30 Richard Biener <rguenther@suse.de>
20694 PR tree-optimization/79256
20695 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
20696 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
20699 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
20701 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
20702 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
20703 builtins to SImode and emit a zero-extend, if necessary.
20705 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
20707 * docs/invoke.texi (RISC-V Options): Alphabetize.
20709 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
20711 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
20714 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
20716 * config/riscv/riscv.c: New file.
20717 * common/config/riscv/riscv-common.c: Likewise.
20718 * config.gcc: Likewise.
20719 * config/riscv/constraints.md: Likewise.
20720 * config/riscv/elf.h: Likewise.
20721 * config/riscv/generic.md: Likewise.
20722 * config/riscv/linux.h: Likewise.
20723 * config/riscv/multilib-generator: Likewise.
20724 * config/riscv/peephole.md: Likewise.
20725 * config/riscv/pic.md: Likewise.
20726 * config/riscv/predicates.md: Likewise.
20727 * config/riscv/riscv-builtins.c: Likewise.
20728 * config/riscv/riscv-c.c: Likewise.
20729 * config/riscv/riscv-ftypes.def: Likewise.
20730 * config/riscv/riscv-modes.def: Likewise.
20731 * config/riscv/riscv-opts.h: Likewise.
20732 * config/riscv/riscv-protos.h: Likewise.
20733 * config/riscv/riscv.h: Likewise.
20734 * config/riscv/riscv.md: Likewise.
20735 * config/riscv/riscv.opt: Likewise.
20736 * config/riscv/sync.md: Likewise.
20737 * config/riscv/t-elf-multilib: Likewise.
20738 * config/riscv/t-linux: Likewise.
20739 * config/riscv/t-linux-multilib: Likewise.
20740 * config/riscv/t-riscv: Likewise.
20741 * configure.ac: Likewise.
20742 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
20743 Waterman as RISC-V maintainers.
20744 * doc/install.texi: Add RISC-V entries.
20745 * doc/invoke.texi: Add RISC-V options section.
20746 * doc/md.texi: Add RISC-V constraints section.
20747 * configure: Regenerated.
20749 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
20752 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
20753 false values to be constant vectors with all 0 or all 1 bits set.
20754 (vcondu<mode><mode>): Likewise.
20755 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
20757 (fpmask_comparison_operator): Update comment.
20758 (vecint_comparison_operator): New predicate.
20759 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
20760 vector conditionals when the true and false values are constant
20761 vectors with all 0 bits or all 1 bits set.
20763 2017-02-06 Martin Sebor <msebor@redhat.com>
20765 PR tree-optimization/79376
20766 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
20768 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
20770 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
20771 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
20772 to simplify split condition.
20774 2017-02-06 Jakub Jelinek <jakub@redhat.com>
20776 * omp-expand.c (oxpand_omp_atomic_fetch_op,
20777 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
20780 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
20782 PR rtl-optimization/68664
20783 * target.def (can_speculate_insn): New hook.
20784 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
20785 * doc/tm.texi: Regenerate.
20786 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
20787 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
20788 (rs6000_sched_can_speculate_insn): New function.
20790 2017-02-06 Jakub Jelinek <jakub@redhat.com>
20792 PR tree-optimization/79284
20793 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
20794 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
20795 vectorizable_mask_load_store, vectorizable_operation,
20796 vect_is_simple_cond, get_same_sized_vectype): Use it instead
20797 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
20798 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
20799 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
20800 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
20801 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
20802 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
20803 is_gimple_assign (stmt). Replace another such test with
20804 is_gimple_assign (stmt).
20806 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
20809 * config/avr/avr.c (rtl-iter.h): Include it.
20810 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
20811 (avr_legitimate_combined_insn): ...and implementation.
20813 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
20815 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
20816 * config/s390/s390.c (s390_const_operand_ok)
20817 (s390_canonicalize_comparison, s390_extract_part)
20818 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
20819 (s390_contiguous_bitmask_p, s390_rtx_costs)
20820 (legitimize_pic_address): Likewise.
20821 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
20822 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
20823 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
20824 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
20825 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
20827 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
20829 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
20830 REGNO($0) == REGNO($1).
20832 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20834 * config/s390/linux.h(SIZE_TYPE): Add comment.
20836 2017-02-06 Julian Brown <julian@codesourcery.com>
20837 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20838 Virendra Pathak <virendra.pathak@broadcom.com>
20840 * config/aarch64/aarch64-cores.def: Change the scheduler
20842 * config/aarch64/aarch64.md: Include thunderx2t99.md.
20843 * config/aarch64/thunderx2t99.md: New file.
20845 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20847 * doc/standards.texi (Go Language): Update link to language
20850 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
20852 * tree-eh.c (lower_resx): Sanitize profile.
20853 (cleanup_empty_eh_move_lp): Likewise.
20855 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
20858 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
20860 * cfgloopmanip.h (loop_version): Update prototype.
20861 * modulo-sched.c (sms_schedule): Update call of loop_version.
20862 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
20863 * tree-parloops.c (gen_parallel_loop): Likewise.
20864 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
20865 * tree-ssa-loop-split.c (split_loop): Likewise.
20866 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
20867 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
20869 2017-02-05 Martin Liska <mliska@suse.cz>
20872 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
20874 (print_operand_address): Initialize a struct to zero.
20876 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20878 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
20879 garbage collector only in textual form.
20881 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20883 * doc/extend.texi (x86 specific memory model extensions for
20884 transactional memory): Simplify a phrase.
20886 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
20889 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
20890 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
20891 (atomic_storedi_1): Likewise.
20893 2017-02-04 Jakub Jelinek <jakub@redhat.com>
20895 PR tree-optimization/79338
20896 * tree-parloops.c (gather_scalar_reductions): Don't call
20897 vect_analyze_loop_form for loop->inner before destroying loop's
20900 2017-02-03 Martin Sebor <msebor@redhat.com>
20902 PR tree-optimization/79327
20903 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
20904 when precision has resulted in leading zeros.
20905 (format_integer): Adjust the likely counter to assume an unknown
20906 argument that may be zero is non-zero.
20908 2017-02-03 Jason Merrill <jason@redhat.com>
20911 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
20912 avoid copying non-taken branch.
20914 2017-02-03 Jakub Jelinek <jakub@redhat.com>
20916 PR tree-optimization/79340
20917 * tree-vect-loop.c (vectorizable_reduction): Release
20918 vec_defs elements after safe_splicing them into other vectors.
20921 PR tree-optimization/79327
20922 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
20923 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
20925 (format_integer): Use wide_int_to_tree instead of build_int_cst
20926 + to_?hwi. If argmin is NULL, just set argmin and argmax to
20927 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
20928 of shortest and longest sequence.
20930 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
20932 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
20933 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
20935 2017-02-03 Walter Lee <walt@tilera.com>
20938 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
20939 after initial stackframe link reg save.
20940 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
20942 2017-02-03 Jakub Jelinek <jakub@redhat.com>
20945 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
20946 wu for stxssp alternative.
20948 2017-02-03 Martin Sebor <msebor@redhat.com>
20950 PR tree-optimization/79352
20951 * gimple-fold.c (get_range_strlen): Add argument.
20952 (get_range_strlen): Change return type to bool.
20953 (get_maxval_strlen): Pass in a dummy argument.
20954 * gimple-fold.h (get_range_strlen): Change return type to bool.
20955 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
20956 * tree.h (array_at_struct_end_p): Add argument.
20957 * tree.c (array_at_struct_end_p): Handle it.
20959 2017-02-03 Martin Liska <mliska@suse.cz>
20962 * multiple_target.c (create_dispatcher_calls): Redirect edge
20963 from a caller of a dispatcher.
20964 (expand_target_clones): Make the clones local.
20965 (ipa_target_clone): Do both target clones and resolvers.
20966 (ipa_dispatcher_calls): Remove the pass.
20967 (pass_dispatcher_calls::gate): Likewise.
20968 (make_pass_dispatcher_calls): Likewise.
20969 * passes.def (pass_target_clone): Put as very first IPA early
20972 2017-02-03 Martin Liska <mliska@suse.cz>
20974 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
20975 in case of a function with ifunc attribute.
20977 2017-02-03 Martin Liska <mliska@suse.cz>
20979 * cgraph.c (cgraph_node::dump): Dump function version info.
20980 * symtab.c (symtab_node::dump_base): Add missing new line.
20982 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
20984 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
20985 (ifcombine_ifandif): Use it.
20987 2017-02-03 Martin Liska <mliska@suse.cz>
20989 * doc/invoke.texi: Document default value for
20990 use-after-scope-direct-emission-threshold.
20992 2017-02-03 Martin Liska <mliska@suse.cz>
20994 PR tree-optimization/79339
20995 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
20996 (format_floating): Likewise.
20998 2017-02-03 Martin Liska <mliska@suse.cz>
21001 * ipa-prop.c (ipa_node_params_t::insert): Remove current
21003 (ipa_node_params_t::remove): Likewise.
21004 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
21005 initialization from removed ipa_node_params_t::insert.
21006 (ipa_node_params::~ipa_node_params): Move from removed
21007 ipa_node_params_t::release.
21008 * symbol-summary.h (symbol_summary::m_released): New member.
21009 Do not release a summary twice. Do not allow to call finalizer
21010 for types of a summary that live in GGC memory.
21012 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21014 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
21017 2017-02-02 Martin Sebor <msebor@redhat.com>
21019 PR middle-end/79275
21020 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
21021 (format_string): Tighten up the range of output for non-constant
21022 strings and correct the expected range for wide non-constant strings.
21024 2017-02-02 Martin Sebor <msebor@redhat.com>
21026 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
21028 PR middle-end/32003
21029 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
21031 (-fdump-tree-@var): Add to index and document how to come up
21032 with pass-specific option and dump file names.
21033 (-fdump-passes): Clarify where to look for output.
21035 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
21037 PR middle-end/77445
21038 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
21039 statistics of the analyzed path; allow threading for speed when
21040 any of BBs along the path are optimized for speed.
21042 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
21044 PR middle-end/78468
21045 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
21046 settings of the virtual registers.
21049 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
21051 * explow.c (get_dynamic_stack_size): Take known alignment of stack
21052 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
21055 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21057 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
21058 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
21060 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21062 * config/s390/s390.md: Add missing comments with the expanded
21064 * config/s390/vector.md: Likewise.
21065 * config/s390/vx-builtins.md: Likewise.
21067 2017-02-02 Jakub Jelinek <jakub@redhat.com>
21070 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
21071 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
21072 conditions on a single line.
21074 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21076 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
21077 __S390_VX__ to __VX__.
21079 2017-02-01 Andrew Pinski <apinski@cavium.com>
21081 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
21082 stmt_info to record_stmt_cost.
21083 (vect_get_known_peeling_cost): Pass stmt_info if known to
21085 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
21086 cpu_vector_cost field into
21087 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
21088 field into vec_int_stmt_cost and vec_fp_stmt_cost.
21089 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
21090 splitting of scalar_stmt_cost and vec_stmt_cost.
21091 (thunderx_vector_cost): Likewise.
21092 (cortexa57_vector_cost): LIkewise.
21093 (exynosm1_vector_cost): Likewise.
21094 (xgene1_vector_cost): Likewise.
21095 (thunderx2t99_vector_cost): Improve after the splitting of the two
21097 (aarch64_builtin_vectorization_cost): Update for the splitting of
21098 scalar_stmt_cost and vec_stmt_cost.
21100 2017-02-01 Torvald Riegel <triegel@redhat.com>
21101 Richard Henderson <rth@redhat.com>
21103 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
21104 conditional on existance of a fast atomic load.
21105 * optabs-query.c (can_atomic_load_p): New function.
21106 * optabs-query.h (can_atomic_load_p): Declare it.
21107 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
21108 no fast atomic load is available for the particular size of access.
21109 (expand_atomic_compare_and_swap): Likewise.
21110 (expand_atomic_load): Likewise.
21111 (expand_atomic_store): Likewise.
21112 (expand_atomic_fetch_op): Likewise.
21113 * testsuite/lib/target-supports.exp
21114 (check_effective_target_sync_int_128): Remove x86 because it provides
21115 no fast atomic load.
21116 (check_effective_target_sync_int_128_runtime): Likewise.
21118 2017-02-01 Richard Biener <rguenther@suse.de>
21120 * graphite.c: Include tree-vectorizer.h for find_loop_location.
21121 (graphite_transform_loops): Provide opt-info for optimized nests.
21122 * tree-parloop.c (parallelize_loops): Provide opt-info for
21123 parallelized loops.
21125 2017-02-01 Richard Biener <rguenther@suse.de>
21127 PR middle-end/79315
21128 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
21129 was not set before.
21131 2017-02-01 Richard Biener <rguenther@suse.de>
21133 PR tree-optimization/71824
21134 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
21135 Verify the loops are valid in the merged SESE region.
21136 (scop_detection::can_represent_loop_1): Check analyzing the
21137 evolution of the number of iterations in the region succeeds.
21139 2017-01-31 Ian Lance Taylor <iant@golang.org>
21141 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
21142 REG_ARGS_SIZE note to 32-bit push insns and call insn.
21144 2017-01-31 David Malcolm <dmalcolm@redhat.com>
21146 PR preprocessor/79210
21147 * input.c (get_substring_ranges_for_loc): Replace line_width
21148 assertion with error-handling.
21150 2017-01-31 Richard Biener <rguenther@suse.de>
21152 PR tree-optimization/77318
21153 * graphite-sese-to-poly.c (extract_affine): Fix assert.
21154 (create_pw_aff_from_tree): Take loop parameter.
21155 (add_condition_to_pbb): Pass loop of the condition to
21156 create_pw_aff_from_tree.
21158 2017-01-31 Jakub Jelinek <jakub@redhat.com>
21160 * config/s390/s390.c (s390_asan_shadow_offset): New function.
21161 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
21163 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
21167 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
21169 (convert_int_to_float128): Likewise.
21170 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
21171 (convert_int_to_float128): Likewise.
21172 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
21173 (UNSPEC_IEEE128_CONVERT): Likewise.
21174 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
21175 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
21176 Use local variables for IBM extended format.
21177 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
21178 (fix_trunc<mode>si2_fprs): Likewise.
21179 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
21180 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
21181 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
21182 to know that we can now have integers of all sizes in vector
21184 (fix<uns>_<mode>di2_hw): Likewise.
21185 (float<uns>_<mode>si2_hw): Likewise.
21186 (fix_<mode>si2_hw): Likewise.
21187 (fixuns_<mode>si2_hw): Likewise.
21188 (float<uns>_<mode>di2_hw): Likewise.
21189 (float_<mode>di2_hw): Likewise.
21190 (float_<mode>si2_hw): Likewise.
21191 (floatuns_<mode>di2_hw): Likewise.
21192 (floatuns_<mode>si2_hw): Likewise.
21193 (xscvqp<su>wz_<mode>): Delete, no longer used.
21194 (xscvqp<su>dz_<mode>): Likewise.
21195 (xscv<su>dqp_<mode>): Likewise.
21196 (ieee128_mfvsrd_64bit): Likewise.
21197 (ieee128_mfvsrd_32bit): Likewise.
21198 (ieee128_mfvsrwz): Likewise.
21199 (ieee128_mtvsrw): Likewise.
21200 (ieee128_mtvsrd_64bit): Likewise.
21201 (ieee128_mtvsrd_32bit): Likewise.
21203 2017-01-31 Martin Liska <mliska@suse.cz>
21206 * ipa-prop.c (ipa_free_all_node_params): Call release method
21207 instead of ~sumbol_summary to not to trigger double times
21210 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
21212 PR tree-optimization/71691
21213 * bitmap.h (class auto_bitmap): New.
21214 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
21215 is_maybe_undefined instead of ssa_undefined_value_p.
21217 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21219 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
21220 __S390_ARCH_LEVEL__ to __ARCH__.
21222 2017-01-31 Jakub Jelinek <jakub@redhat.com>
21224 PR tree-optimization/79267
21225 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
21226 if should_remove_lhs_p is true.
21228 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
21231 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
21232 (add_alignment_attribute): New.
21233 (base_type_die): Add alignment attribute.
21234 (subrange_type_die): Likewise.
21235 (modified_type_die): Likewise.
21236 (gen_array_type_die): Likewise.
21237 (gen_descr_array_type_die: Likewise.
21238 (gen_enumeration_type_die): Likewise.
21239 (gen_subprogram_die): Likewise.
21240 (gen_variable_die): Likewise.
21241 (gen_field_die): Likewise.
21242 (gen_ptr_to_mbr_type_die): Likewise.
21243 (gen_struct_or_union_type_die): Likewise.
21244 (gen_subroutine_type_die): Likewise.
21245 (gen_typedef_die): Likewise.
21246 (base_type_cmp): Compare alignment attribute.
21248 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21251 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
21252 (setb_unsigned) New pattern for setb with CCUNS.
21253 * config/rs6000/rs6000.c (expand_block_compare): Use a different
21254 subfc./subfe sequence to avoid overflow problems. Generate a
21255 shorter sequence with cmpld/setb for power9.
21256 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
21257 for generating subfc. instruction.
21258 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
21259 now uses this instruction.
21261 2017-01-30 Ian Lance Taylor <iant@google.com>
21264 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
21265 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
21267 2017-01-30 Martin Sebor <msebor@redhat.com>
21269 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
21270 Move constant to the right of a relational operator.
21271 (get_mpfr_format_length, format_character, format_string): Ditto.
21272 (should_warn_p, maybe_warn): Same.
21274 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
21276 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
21279 * asan.c (get_translation_unit_decl): Remove function.
21280 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
21282 2017-01-30 Martin Liska <mliska@suse.cz>
21284 PR gcov-profile/79259
21285 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
21286 -fprofile-generate.
21288 2017-01-30 Martin Liska <mliska@suse.cz>
21291 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
21292 Initialize variables with NULL value.
21294 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
21297 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
21299 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
21301 2017-01-30 Richard Biener <rguenther@suse.de>
21303 PR tree-optimization/79276
21304 * tree-vrp.c (process_assert_insertions): Properly adjust common
21305 when removing a duplicate.
21307 * gcc.dg/torture/pr79276.c: New testcase.
21309 2017-01-30 Richard Biener <rguenther@suse.de>
21311 PR tree-optimization/79256
21312 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
21313 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
21315 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
21317 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
21320 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
21321 ("*r<noxa>sbg_<mode>_sll_bitmask")
21322 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
21323 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
21324 Use contiguous_bitmask_nowrap_operand.
21326 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21329 * config/rs6000/altivec.h (vec_xl): Revise #define.
21330 (vec_xst): Likewise.
21332 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
21334 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
21336 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
21338 PR rtl-optimization/79194
21339 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
21340 traps before call to bypass_conditional_jumps.
21342 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
21344 PR tree-optimization/71374
21345 * lra-constraints.c (check_conflict_input_operands): New.
21346 (match_reload): Use it.
21348 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
21351 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
21352 account to calculate conflict_set.
21354 2017-01-27 Bin Cheng <bin.cheng@arm.com>
21356 PR rtl-optimization/78559
21357 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
21358 other_insn in combine.
21360 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
21362 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
21363 uint16_type_node for BT_UINT16.
21365 2017-01-27 David Malcolm <dmalcolm@redhat.com>
21367 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
21368 "RTL Tests" to menu.
21369 (GIMPLE Tests): New node.
21370 (RTL Tests): New node.
21372 2017-01-27 Richard Biener <rguenther@suse.de>
21374 PR tree-optimization/79245
21375 * tree-loop-distribution.c (distribute_loop): Apply cost
21376 modeling also to detected patterns.
21378 2017-01-27 Richard Biener <rguenther@suse.de>
21380 PR tree-optimization/71433
21381 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
21382 (compare_assert_loc): New function.
21383 (process_assert_insertions): Sort and optimize assert locations
21384 to remove duplicates and push down identical assertions on
21385 edges to their destination block.
21387 2017-01-27 Richard Biener <rguenther@suse.de>
21389 PR tree-optimization/79244
21390 * tree-vrp.c (remove_range_assertions): Forcefully propagate
21391 out SSA names even if abnormal.
21393 2017-01-27 Jakub Jelinek <jakub@redhat.com>
21395 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
21396 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
21397 instead of MPFR_RNDN.
21399 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
21402 * arm.c (arm_option_override): Don't call build_target_option_node
21403 until after doing all option overrides.
21404 (arm_valid_target_attribute_tree): Likewise.
21406 2017-01-27 Martin Liska <mliska@suse.cz>
21408 * doc/invoke.texi (-fprofile-arcs): Document profiling support
21409 for {cd}tors and C++ {cd}tors.
21411 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
21413 * config/s390/s390.md ("*setmem_long_and")
21414 ("*setmem_long_and_31z"): Use zero_extend instead of and.
21416 2017-01-26 Martin Sebor <msebor@redhat.com>
21418 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
21421 2017-01-26 Martin Sebor <msebor@redhat.com>
21423 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
21424 HAVE_DFmode before using XFmode or DFmode.
21425 (parse_directive): Avoid using the z length modifier to avoid
21426 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
21428 PR middle-end/78703
21429 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
21430 to accept adjustment as an array.
21431 (get_int_range): New function.
21432 (struct directive): Make width and prec arrays.
21433 (directive::set_width, directive::set_precision): Call get_int_range.
21434 (format_integer, format_floating): Handle width and precision ranges.
21435 (format_string, parse_directive): Same.
21437 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21440 * dwarf2out.c (generate_skeleton_bottom_up): For children with
21441 comdat_type_p set, just clone them, but keep the children in the
21445 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
21446 which have direct callers with -fvar-tracking-assignments enabled
21448 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
21449 inside of type units.
21451 2017-01-26 Martin Sebor <msebor@redhat.com>
21453 PR middle-end/78703
21454 * gimple-ssa-sprintf.c (struct result_range): Add likely and
21456 (struct format_result): Replace number_chars, number_chars_min,
21457 and number_chars_max with a single member of struct result_range.
21459 (format_result::operator+=): Adjust.
21460 (struct fmtresult): Remove bounded. Handle likely and unlikely
21462 (fmtresult::adjust_for_width_or_precision): New function.
21463 (fmtresult:type_max_digits): New function.
21464 (bytes_remaining): Handle likely and unlikely counters.
21465 (min_bytes_remaining): Remove.
21466 (format_percent): Simplify.
21467 (format_integer, format_floating): Set likely and unlikely counters.
21468 (get_string_length, format_character, format_string): Same.
21469 (format_plain, should_warn_p): New function.
21470 (maybe_warn): Call should_warn_p. Update diagnostic messages
21471 and handle those for all directives, including plain strings.
21472 (format_directive): Handle likely and unlikely counters.
21473 Remove unnecessary quoting from diagnostics. Add an informational
21475 (add_bytes): Remove.
21476 (pass_sprintf_length::compute_format_length): Simplify.
21477 (try_substitute_return_value): Handle likely and unlikely counters.
21479 2017-01-26 Carl Love <cel@us.ibm.com>
21481 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
21482 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
21484 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
21487 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
21488 endianess for subregs into account.
21489 * lra-constraints.c (lra_constraints): Do risky transformations
21490 always on the first iteration.
21491 * lra-lives.c (check_pseudos_live_through_calls): Add arg
21492 last_call_used_reg_set.
21493 (process_bb_lives): Define and use last_call_used_reg_set.
21494 * lra.c (lra): Always continue after lra_constraints on the first
21497 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
21499 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
21501 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
21503 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21505 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
21506 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
21507 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
21508 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
21509 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
21510 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
21511 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
21512 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
21513 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
21515 2017-01-26 Marek Polacek <polacek@redhat.com>
21518 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
21519 for the third operand.
21521 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21523 PR middle-end/79236
21524 * omp-low.c (struct omp_context): Add simt_stmt field.
21525 (scan_omp_for): Return omp_context *.
21526 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
21527 context to the _simt_ SIMD stmt.
21528 (lower_omp_for): For combined SIMD with sibling _simt_
21529 SIMD, make sure to use the same decls in _looptemp_
21530 clauses as in the sibling.
21532 2017-01-26 David Sherwood <david.sherwood@arm.com>
21534 PR middle-end/79212
21535 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
21538 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21541 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
21542 emit fld b; fld a; if possible.
21544 * brig-builtins.def: Update copyright years.
21545 * config/arm/arm_acle_builtins.def: Update copyright years.
21547 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
21550 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
21551 constraint instead of o for the stxsd instruction.
21553 2017-01-25 Carl Love <cel@us.ibm.com>
21555 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
21556 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
21558 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
21560 * doc/invoke.texi (C++ Dialect Options): Fix typo.
21562 2017-01-25 Richard Biener <rguenther@suse.de>
21564 PR tree-optimization/69264
21565 * target.def (vector_alignment_reachable): Improve documentation.
21566 * doc/tm.texi: Regenerate.
21567 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
21569 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
21570 earlier changes with respect to TYPE_USER_ALIGN.
21571 (vector_alignment_reachable_p): Likewise. Improve dumping.
21573 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21576 * config/arm/arm.md (xordi3): Force constant operand into a register
21579 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21581 * doc/invoke.texi (-fstore-merging): Correct default optimization
21582 levels at which it is enabled.
21583 (-O): Move -fstore-merging from list to...
21586 2017-01-25 Richard Biener <rguenther@suse.de>
21589 * omp-expand.c: Include debug.h.
21590 (expand_omp_taskreg): Make sure to generate early debug before
21591 outlining anything from a function.
21592 (expand_omp_target): Likewise.
21593 (grid_expand_target_grid_body): Likewise.
21595 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
21598 * asan.c (get_translation_unit_decl): New function.
21599 (asan_add_global): Extract modules file name from globals
21600 TRANSLATION_UNIT_DECL name.
21602 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
21605 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
21606 for long calls with APCS frame and VFP.
21608 2017-01-24 David Malcolm <dmalcolm@redhat.com>
21610 * cfg.c (original_copy_tables_initialized_p): New function.
21611 * cfg.h (original_copy_tables_initialized_p): New decl.
21612 * cfgrtl.c (relink_block_chain): Guard the call to
21613 free_original_copy_tables with a call to
21614 original_copy_tables_initialized_p.
21615 * cgraph.h (symtab_node::native_rtl_p): New decl.
21616 * cgraphunit.c (symtab_node::native_rtl_p): New function.
21617 (symtab_node::needed_p): Don't assert for early assembly output
21618 for __RTL functions.
21619 (cgraph_node::finalize_function): Set "force_output" for __RTL
21621 (cgraph_node::analyze): Bail out early for __RTL functions.
21622 (analyze_functions): Update assertion to support __RTL functions.
21623 (cgraph_node::expand): Bail out early for __RTL functions.
21624 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
21626 * function.h (struct function): Update comment for field
21628 * gimple-expr.c: Include "tree-pass.h".
21629 (gimple_has_body_p): Return false for __RTL functions.
21630 * Makefile.in (OBJS): Add run-rtl-passes.o.
21631 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
21633 (gcc::pass_manager::get_clean_slate): New accessor.
21634 * passes.c: Include "insn-addr.h".
21635 (should_skip_pass_p): Add logging. Update logic for running
21636 "expand" to be compatible with both __GIMPLE and __RTL. Guard
21637 property-provider override so it is only done for gimple passes.
21639 (skip_pass): New function.
21640 (execute_one_pass): Call skip_pass when skipping passes.
21641 * read-md.c (md_reader::read_char): Support filtering
21642 the input to a subset of line numbers.
21643 (md_reader::md_reader): Initialize fields
21644 m_first_line and m_last_line.
21645 (md_reader::read_file_fragment): New function.
21646 * read-md.h (md_reader::read_file_fragment): New decl.
21647 (md_reader::m_first_line): New field.
21648 (md_reader::m_last_line): New field.
21649 * read-rtl-function.c (function_reader::create_function): Only
21650 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
21651 curr_properties. Set DECL_INITIAL to a dummy block.
21652 (read_rtl_function_body_from_file_range): New function.
21653 * read-rtl-function.h (read_rtl_function_body_from_file_range):
21655 * run-rtl-passes.c: New file.
21656 * run-rtl-passes.h: New file.
21658 2017-01-24 Jeff Law <law@redhat.com>
21660 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
21663 2017-01-24 Bin Cheng <bin.cheng@arm.com>
21665 PR tree-optimization/79159
21666 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
21667 (record_nonwrapping_iv): Improve boundary using above function if no
21668 value range information.
21670 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
21671 Martin Jambor <mjambor@suse.cz>
21673 * brig-builtins.def: New file.
21674 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
21675 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
21676 (DEF_HSAIL_SAT_BUILTIN): Likewise.
21677 (DEF_HSAIL_INTR_BUILTIN): Likewise.
21678 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
21679 * builtin-types.def (BT_INT8): New.
21680 (BT_INT16): Likewise.
21681 (BT_UINT8): Likewise.
21682 (BT_UINT16): Likewise.
21683 (BT_FN_ULONG): Likewise.
21684 (BT_FN_UINT_INT): Likewise.
21685 (BT_FN_UINT_ULONG): Likewise.
21686 (BT_FN_UINT_LONG): Likewise.
21687 (BT_FN_UINT_PTR): Likewise.
21688 (BT_FN_ULONG_PTR): Likewise.
21689 (BT_FN_INT8_FLOAT): Likewise.
21690 (BT_FN_INT16_FLOAT): Likewise.
21691 (BT_FN_UINT32_FLOAT): Likewise.
21692 (BT_FN_UINT16_FLOAT): Likewise.
21693 (BT_FN_UINT8_FLOAT): Likewise.
21694 (BT_FN_UINT64_FLOAT): Likewise.
21695 (BT_FN_UINT16_UINT32): Likewise.
21696 (BT_FN_UINT32_UINT16): Likewise.
21697 (BT_FN_UINT16_UINT16_UINT16): Likewise.
21698 (BT_FN_INT_PTR_INT): Likewise.
21699 (BT_FN_UINT_PTR_UINT): Likewise.
21700 (BT_FN_LONG_PTR_LONG): Likewise.
21701 (BT_FN_ULONG_PTR_ULONG): Likewise.
21702 (BT_FN_VOID_UINT64_UINT64): Likewise.
21703 (BT_FN_UINT8_UINT8_UINT8): Likewise.
21704 (BT_FN_INT8_INT8_INT8): Likewise.
21705 (BT_FN_INT16_INT16_INT16): Likewise.
21706 (BT_FN_INT_INT_INT): Likewise.
21707 (BT_FN_UINT_FLOAT_UINT): Likewise.
21708 (BT_FN_FLOAT_UINT_UINT): Likewise.
21709 (BT_FN_ULONG_UINT_UINT): Likewise.
21710 (BT_FN_ULONG_UINT_PTR): Likewise.
21711 (BT_FN_ULONG_ULONG_ULONG): Likewise.
21712 (BT_FN_UINT_UINT_UINT): Likewise.
21713 (BT_FN_VOID_UINT_PTR): Likewise.
21714 (BT_FN_UINT_UINT_PTR: Likewise.
21715 (BT_FN_UINT32_UINT64_PTR): Likewise.
21716 (BT_FN_INT_INT_UINT_UINT): Likewise.
21717 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
21718 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
21719 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
21720 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
21721 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
21722 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
21723 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
21724 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
21725 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
21726 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
21727 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
21728 * doc/frontends.texi: List BRIG FE.
21729 * doc/install.texi (Testing): Add BRIG tesring requirements.
21730 * doc/invoke.texi (Overall Options): Mention BRIG.
21731 * doc/standards.texi (Standards): Doucment BRIG HSA version.
21733 2017-01-24 Richard Biener <rguenther@suse.de>
21735 PR translation/79208
21736 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
21738 2017-01-24 Martin Jambor <mjambor@suse.cz>
21741 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
21742 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
21743 and known_contexts.
21745 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
21747 PR middle-end/79123
21748 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
21749 casts from signed to unsigned really don't have a range.
21751 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
21753 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
21754 GMP_RNDx for compatiblity.
21756 2017-01-24 Martin Liska <mliska@suse.cz>
21759 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
21760 that would prevent us to call alloca with -1 as argument.
21762 2017-01-24 Jakub Jelinek <jakub@redhat.com>
21764 * dwarf2out.c (output_compilation_unit_header, output_file_names):
21765 Avoid -Wformat-security warning.
21767 2017-01-23 Andrew Pinski <apinski@cavium.com>
21769 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
21772 2017-01-23 Martin Sebor <msebor@redhat.com>
21774 PR middle-end/78703
21775 * gimple-ssa-sprintf.c (warn_level): New global.
21776 (format_integer): Use it here and throughout the rest of the file.
21777 Use the same switch to compute sign as base.
21778 (maybe_warn): New function.
21779 (format_directive): Factor out warnings into maybe_warn.
21780 Add debugging output. Use warn_level.
21781 (add_bytes): Use warn_level.
21782 (pass_sprintf_length::compute_format_length): Add debugging output.
21783 (try_substitute_return_value): Same.
21784 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
21786 PR middle-end/78703
21787 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
21788 (struct fmtresult, format_integer, format_floating): Adjust.
21789 (fmtresult::fmtresult): Set max correctly in two argument ctor.
21790 (get_string_length, format_string,format_directive): Same.
21791 (pass_sprintf_length::compute_format_length): Same.
21792 (try_substitute_return_value): Simplify slightly.
21794 PR middle-end/78703
21795 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
21796 (fmtresult::operator+=): Outlined.
21797 (struct fmtresult): Add ctors.
21798 (struct conversion_spec): Rename...
21799 (struct directive): ...to this. Add and remove data members.
21800 (directive::set_width, directive::set_precision): New functions.
21801 (format_percent): Use fmtresult ctor.
21802 (get_width_and_precision): Remove.
21803 (format_integer): Make naming changes. Avoid computing width and
21805 (format_floating): Same. Adjust indentation.
21806 (format_character, format_none): New functions.
21807 (format_string): Moved character handling to format_character.
21808 (format_directive): Remove arguments, change return type.
21809 (parse_directive): New function.
21810 (pass_sprintf_length::compute_format_length): Move directive
21811 parsing to parse_directive.
21813 2017-01-23 Jakub Jelinek <jakub@redhat.com>
21815 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
21816 (assign_assembler_name_if_needed): ... this.
21817 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
21818 (assign_assembler_name_if_needed): ... this.
21819 (free_lang_data_in_cgraph): Adjust callers.
21820 * cgraphunit.c (cgraph_node::analyze): Likewise.
21821 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
21824 2017-01-23 Richard Biener <rguenther@suse.de>
21826 PR tree-optimization/79088
21827 PR tree-optimization/79188
21828 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
21829 resetting loop bounds after last path deletion. Reset loop
21830 bounds of the target loop, make code match the comments.
21831 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
21832 Make sure loops need no fixups.
21834 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
21836 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
21837 exponent support with double type for first argument.
21838 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
21839 type returned by __builtin_vec_extract_sig,
21840 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
21841 functions from "vector int" to "vector unsigned int" or from
21842 "vector long long int" to "vector unsigned long long int".
21843 Changed type returned by __builtin_vec_extract_exp,
21844 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
21845 functions from "vector int" to "vector unsigned int" or from
21846 "vector long long int" to "vector unsigned long long int".
21847 Changed return type of __builtin_vec_test_data_class,
21848 __builtin_vec_test_data_class_sp, and
21849 __builtin_vec_test_data_class_dp from "vector int" to
21850 "vector bool int" or from "vector long long int" to "vector bool
21851 long long int" and changed second argument type from "unsigned
21852 int" to "int". Added new overloaded function forms "vector float
21853 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
21854 "vector float __builtin_vec_insert_exp_sp (vector float, vector
21855 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
21856 double, vector unsigned long long int)" and "vector double
21857 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
21858 long int)". Changed return type of
21859 __builtin_scalar_test_data_class and
21860 __builtin_scalar_test_data_class_sp and
21861 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
21862 int" and changed second argument from "unsigned int" to "int".
21863 Changed type returned by __builtin_scalar_test_neg,
21864 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
21865 from "int" to "bool int". Added new overloaded function form
21866 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
21867 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
21868 exponent double-precision with floating point first argument.
21869 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
21870 documentation of scalar_test_data_class, scalar_test_neg,
21871 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
21872 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
21873 vec_test_data_class built-in functions to reflect refinements in
21874 their type signatures.
21876 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
21878 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
21880 (aarch64_elf_asm_destructor): Likewise.
21882 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
21884 PR rtl-optimization/78634
21885 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
21886 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
21887 * ifcvt.c (noce_try_cmove): Add missing cost check.
21889 PR rtl-optimization/71724
21890 * combine.c (if_then_else_cond): Look for situations where it is
21891 beneficial to undo the work of one of the recursive calls.
21893 2017-01-23 Bin Cheng <bin.cheng@arm.com>
21895 PR tree-optimization/70754
21896 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
21897 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
21898 combined stmt before it if not NULL.
21899 (combine_chains): Process refs reversely and compute dominance point
21902 2017-01-23 Martin Liska <mliska@suse.cz>
21904 PR tree-optimization/79196
21905 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
21906 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
21908 (strlen_optimize_stmt): Call the renamed function.
21910 2017-01-23 Michael Matz <matz@suse.de>
21912 PR tree-optimization/78384
21913 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
21915 2017-01-23 Richard Biener <rguenther@suse.de>
21917 PR tree-optimization/79186
21918 * tree-vrp.c (register_new_assert_for): Make sure we've seen
21919 both incoming edges before moving an assert.
21921 2017-01-23 Martin Jambor <mjambor@suse.cz>
21923 * ipa-prop.c (load_from_param_1): Removed.
21924 (load_from_unmodified_param): Bits from load_from_param_1 put back
21926 (load_from_param): Removed.
21927 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
21928 with stmt. Reverted back to use of load_from_unmodified_param.
21930 2017-01-23 Martin Jambor <mjambor@suse.cz>
21933 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
21934 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
21935 field a pointer to garbage collected vector, mark lattices and
21936 ipcp_orig_node with GTY((skip)).
21937 (ipa_get_param_count): Adjust to descriptors being a pointer.
21938 (ipa_get_param): Likewise.
21939 (ipa_get_type): Likewise.
21940 (ipa_get_param_move_cost): Likewise.
21941 (ipa_set_param_used): Likewise.
21942 (ipa_get_controlled_uses): Likewise.
21943 (ipa_set_controlled_uses): Likewise.
21944 (ipa_is_param_used): Likewise.
21945 (ipa_node_params_t): Move into garbage collector. New methods insert
21947 (ipa_node_params_sum): Annotate wth GTY(()).
21948 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
21950 (ipa_load_from_parm_agg): Adjust declaration.
21951 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
21952 * ipa-profile.c (ipa_profile): Likewise.
21953 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
21954 (ipa_populate_param_decls): Make descriptors parameter garbage
21956 (ipa_dump_param): Adjust to descriptors being a pointer.
21957 (ipa_alloc_node_params): Likewise.
21958 (ipa_initialize_node_params): Likewise.
21959 (load_from_param_1): Make descriptors parameter garbage collected.
21960 (load_from_unmodified_param): Likewise.
21961 (load_from_param): Likewise.
21962 (ipa_load_from_parm_agg): Likewise.
21963 (ipa_node_params::~ipa_node_params): Removed.
21964 (ipa_free_all_node_params): Remove call to delete operator.
21965 (ipa_node_params_t::insert): New.
21966 (ipa_node_params_t::remove): Likewise.
21967 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
21968 copy known_csts and known_contexts vectors.
21969 (ipa_read_node_info): Adjust to descriptors being a pointer.
21970 (ipcp_modif_dom_walker): Make m_descriptors field garbage
21972 (ipcp_transform_function): Make descriptors variable garbage
21975 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
21977 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
21978 * config/i386/avx512dqintrin.h: Ditto.
21979 * config/i386/avx512fintrin.h: Ditto.
21980 * config/i386/i386.c: Handle new builtins.
21981 * config/i386/i386-builtin.def: Add new builtins.
21982 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
21983 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
21985 2017-01-23 Jakub Jelinek <jakub@redhat.com>
21986 Martin Liska <mliska@suse.cz>
21988 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
21989 * asan.c (asan_expand_poison_ifn): Support stores and use
21990 appropriate ASAN report function.
21991 * internal-fn.c (expand_ASAN_POISON_USE): New function.
21992 * internal-fn.def (ASAN_POISON_USE): Declare.
21993 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
21994 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
21995 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
21996 ASAN_POISON calls w/o LHS.
21997 * tree-ssa.c (execute_update_addresses_taken): Create clobber
21998 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
21999 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
22000 * gimplify.c (asan_poison_variables): Add attribute
22001 use_after_scope_memory to variables that really needs to live
22003 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
22004 having the attribute.
22006 2017-01-23 Martin Liska <mliska@suse.cz>
22008 * asan.c (create_asan_shadow_var): New function.
22009 (asan_expand_poison_ifn): Likewise.
22010 * asan.h (asan_expand_poison_ifn): New declaration.
22011 * internal-fn.c (expand_ASAN_POISON): Likewise.
22012 * internal-fn.def (ASAN_POISON): New builtin.
22013 * sanopt.c (pass_sanopt::execute): Expand
22014 asan_expand_poison_ifn.
22015 * tree-inline.c (copy_decl_for_dup_finish): Make function
22017 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
22018 * tree-ssa.c (is_asan_mark_p): New function.
22019 (execute_update_addresses_taken): Rewrite local variables
22020 (identified just by use-after-scope as addressable) into SSA.
22022 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
22024 * doc/install.texi (Specific): opensource.apple.com uses https
22025 now. Remove trailing slash.
22027 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
22029 * README.Portability: Remove note on an Irix compatibility issue.
22031 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
22033 * gcov.c (INCLUDE_ALGORITHM): Define.
22034 (INCLUDE_VECTOR): Define.
22035 No longer include <vector> and <algorithm> directly.
22037 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
22039 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
22041 * doc/invoke.texi (Code Gen Options): Ditto.
22043 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
22046 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
22048 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
22050 rtl-optimization/79125
22051 * cprop.c (local_cprop_pass): Handle cases where we make an
22052 unconditional trap.
22054 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
22058 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
22059 read from, for big endian.
22061 2017-01-20 Jiong Wang <jiong.wang@arm.com>
22063 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
22064 register pauth builtins for LP64 only.
22066 2017-01-20 Marek Polacek <polacek@redhat.com>
22069 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
22072 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
22074 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
22076 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
22077 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
22078 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
22080 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22083 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
22084 in big-endian mode when they are not a single duplicated value.
22086 2017-01-20 Richard Biener <rguenther@suse.de>
22088 * BASE-VER: Bump to 7.0.1.
22090 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
22092 * omp-low.c (omplow_simd_context): New struct. Use it...
22093 (lower_rec_simd_input_clauses): ...here and...
22094 (lower_rec_input_clauses): ...here to hold common data. Adjust all
22095 references to idx, lane, max_vf, is_simt.
22097 2017-01-20 Graham Markall <graham.markall@embecosm.com>
22099 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
22102 2017-01-20 Martin Jambor <mjambor@suse.cz>
22104 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
22105 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
22107 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
22108 (GTFILES): Rename hsa.c to hsa-common.c.
22109 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
22110 * hsa-dump.c: Likewise.
22111 * hsa-gen.c: Likewise.
22112 * hsa-regalloc.c: Likewise.
22113 * ipa-hsa.c: Likewise.
22114 * omp-expand.c: Likewise.
22115 * omp-low.c: Likewise.
22116 * toplev.c: Likewise.
22118 2017-01-20 Marek Polacek <polacek@redhat.com>
22121 * doc/invoke.texi: Document -Wduplicated-branches.
22122 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
22123 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
22124 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
22125 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
22126 return 0 only when not OEP_LEXICOGRAPHIC.
22127 (fold_build_cleanup_point_expr): Use the expression
22128 location when building CLEANUP_POINT_EXPR.
22129 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
22130 * tree.c (add_expr): Handle error_mark_node.
22132 2017-01-20 Martin Liska <mliska@suse.cz>
22135 * tree-profile.c (init_ic_make_global_vars): Do not call
22137 (gimple_init_gcov_profiler): Likewise.
22139 2017-01-20 Martin Liska <mliska@suse.cz>
22142 * cgraph.h (maybe_create_reference): Remove argument and
22144 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
22146 * ipa-cp.c (create_specialized_node): Likewise.
22147 * symtab.c (symtab_node::maybe_create_reference): Handle
22148 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
22150 2017-01-20 Martin Liska <mliska@suse.cz>
22152 * read-rtl-function.c (function_reader::create_function): Use
22153 build_decl instread of build_decl_stat.
22155 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
22157 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
22158 * config/i386/avx512dqintrin.h: Ditto.
22159 * config/i386/avx512fintrin.h: Ditto.
22160 * config/i386/i386-builtin-types.def: Add new types.
22161 * config/i386/i386.c: Handle new types.
22162 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
22163 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
22164 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
22165 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
22166 (__builtin_ia32_kshiftridi): New.
22167 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
22169 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
22173 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
22174 define to rs6000_init_stack_protect_guard.
22175 (rs6000_init_stack_protect_guard): New function.
22177 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
22178 Yunqiang Su <yunqiang.su@imgtec.com>
22180 * config.gcc (supported_defaults): Add madd4.
22181 (with_madd4): Add validation.
22182 (all_defaults): Add madd4.
22183 * config/mips/mips.opt (mmadd4): New option.
22184 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
22186 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
22188 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
22189 (ISA_HAS_FUSED_MADD4): Likewise.
22190 * doc/invoke.texi (-mmadd4): Document the new option.
22191 * doc/install.texi (--with-madd4): Document the new option.
22193 2017-01-19 Jiong Wang <jiong.wang@arm.com>
22195 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
22196 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
22197 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
22198 (aarch64_init_pauth_hint_builtins): New.
22199 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
22200 (aarch64_expand_builtin): Expand new builtins.
22202 2017-01-19 Jiong Wang <jiong.wang@arm.com>
22204 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
22205 * combine-stack-adj.c (no_unhandled_cfa): Handle
22206 REG_CFA_TOGGLE_RA_MANGLE.
22207 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
22208 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
22209 info for return address signing.
22210 (aarch64_expand_epilogue): Likewise.
22212 2017-01-19 Jiong Wang <jiong.wang@arm.com>
22214 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
22215 * config/aarch64/aarch64-protos.h
22216 (aarch64_return_address_signing_enabled): New declaration.
22217 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
22219 (aarch64_expand_prologue): Sign return address before it's pushed onto
22221 (aarch64_expand_epilogue): Authenticate return address fetched from
22223 (aarch64_override_options): Sanity check for ILP32 and ISA level.
22224 (aarch64_attributes): New function attributes for "sign-return-address".
22225 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
22226 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
22227 ("*do_return"): Generate combined instructions according to key index.
22228 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
22229 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
22231 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
22232 * config/aarch64/aarch64.opt (msign-return-address=): New.
22233 * doc/extend.texi (AArch64 Function Attributes): Documents
22234 "sign-return-address=".
22235 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
22237 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
22239 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
22240 overall option summary.
22242 2017-01-19 Jiong Wang <jiong.wang@arm.com>
22244 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
22245 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
22246 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
22247 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
22249 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
22251 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
22252 -mpower9-minmax by default for -mcpu=power9.
22253 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
22254 128-bit floating point.
22256 2017-01-20 Alan Modra <amodra@gmail.com>
22258 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
22259 optimizing for size.
22261 2017-01-20 Alan Modra <amodra@gmail.com>
22264 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
22265 for strcmp and strncmp from corresponding builtin decl.
22267 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
22269 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
22270 instead of i386/rtems-64.h.
22271 * config/i386/rtems-64.h: Remove.
22273 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
22277 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
22279 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
22281 2017-01-19 Tamar Christina <tamar.christina@arm.com>
22283 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
22284 Change int to HOST_WIDE_INT.
22285 * config/aarch64/aarch64-protos.h
22286 (aarch64_simd_gen_const_vector_dup): Likewise.
22287 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
22289 2017-01-19 David Malcolm <dmalcolm@redhat.com>
22291 * langhooks-def.h (lhd_type_for_size): New decl.
22292 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
22293 * langhooks.c (lhd_type_for_size): New function, taken from
22296 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
22298 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
22299 define_bypass for CR latency.
22300 (power9-cracked-alu): Update bypass latency and remove power9-branch.
22301 (power9-alu2): Add define_bypass for CR latency.
22303 (power9-mul): Update insn latency.
22304 (power9-mul-compare): Update insn latency, bypass latency and remove
22307 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22309 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
22311 * config/aarch64/aarch64.md
22312 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
22313 aarch64_nopcrelative_literal_loads.
22314 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
22316 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
22318 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
22319 TARGET_LOONGSON_3A.
22320 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
22322 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
22325 * config.gcc (supported_defaults): Add lxc1-sxc1.
22326 (with_lxc1_sxc1): Add validation.
22327 (all_defaults): Add lxc1-sxc1.
22328 * config/mips/mips.opt (mlxc1-sxc1): New option.
22329 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
22331 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
22332 __mips_no_lxc1_sxc1.
22333 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
22334 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
22335 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
22337 2017-01-19 Richard Biener <rguenther@suse.de>
22339 PR tree-optimization/72488
22340 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
22341 sure to restore SSA info.
22342 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
22344 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
22346 PR rtl-optimization/79121
22347 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
22348 of the inner type when shifting an extended value.
22350 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
22353 * symtab.c (symtab_node::equal_address_to): Fix comparing of
22354 interposable aliases.
22356 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
22359 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
22360 Use the evmergelohi instruction.
22361 (mov_si<mode>_e500_subreg4_2_le): Likewise.
22362 (mov_sitf_e500_subreg8_2_be): Likewise.
22363 (mov_sitf_e500_subreg12_2_le): Likewise.
22364 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
22365 (mov_si<mode>_e500_subreg4_2_be): Likewise.
22366 (mov_sitf_e500_subreg8_2_le): Likewise.
22367 (mov_sitf_e500_subreg12_2_be): Likewise.
22369 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22371 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
22372 attribute from vecsimple to vecperm.
22373 (altivec_vbpermq2): Likewise.
22375 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22378 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
22380 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22381 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
22382 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
22383 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
22384 case where N arg is SIZE_MAX.
22385 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
22386 (cmpstrsi): Add pattern.
22388 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22390 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22391 __builtin_vec_revb builtins.
22392 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
22393 built-in functions to support generation of the ISA 3.0 XXBR<x>
22394 vector byte reverse instructions.
22395 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
22396 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
22397 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
22398 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
22399 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
22400 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
22401 (P9V_BUILTIN_VEC_REVB): Likewise.
22402 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
22403 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
22404 (p9_xxbrq_v16qi): Likewise.
22405 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
22406 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
22407 (p9_xxbrh_v8hi): Likewise.
22408 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
22409 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
22410 vec_revb built-in functions.
22412 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
22414 PR rtl-optimization/78952
22415 * config/i386/i386.md (any_extract): New code iterator.
22416 (*insvqi_2): Use any_extract for source operand.
22417 (*insvqi_3): Use any_shiftrt for source operand.
22419 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
22421 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
22423 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
22425 2017-01-18 Matthias Klose <doko@ubuntu.com>
22427 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
22429 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22431 * config/rs6000/altivec.h (vec_bperm): Change #define.
22432 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
22433 (altivec_vbpermq2): New define_insn.
22434 (altivec_vbpermd): Likewise.
22435 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
22436 function interface.
22437 (VBPERMD): Likewise.
22438 (VBPERM): New polymorphic function interface.
22439 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
22440 Add entries for P9V_BUILTIN_VEC_VBPERM.
22441 * doc/extend.texi: Add interfaces for vec_bperm.
22443 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22445 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
22446 first letter of error messages.
22447 (s390_resolve_overloaded_builtin): Likewise.
22448 * config/s390/s390.c (s390_expand_builtin): Likewise.
22449 (s390_invalid_arg_for_unprototyped_fn): Likewise.
22450 (s390_valid_target_attribute_inner_p): Likewise.
22451 * config/s390/s390.md ("tabort"): Likewise.
22453 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
22455 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
22456 (ISA_AVOID_DIV_HILO): New macro.
22457 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
22458 (ISA_HAS_DDIV): Likewise.
22460 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
22462 * doc/invoke.texi (fabi-version): Correct number of occurrences.
22464 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
22466 * doc/invoke.texi (fabi-version): Spelling fix.
22468 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
22471 * doc/invoke.texi (fabi-version): Mention mangling fix for
22474 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
22477 * doc/invoke.texi (fabi-version): Document discriminator mangling.
22479 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
22482 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
22483 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
22485 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
22487 (stack_protect_test): Ditto.
22488 * config/rs6000/rs6000.opt (mstack-protector-guard=,
22489 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
22491 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
22492 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
22493 -mstack-protector-guard-offset=.
22494 (RS/6000 and PowerPC Options): Ditto.
22496 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
22498 * config/i386/i386.h (MASK_CLASS_P): New define.
22499 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
22500 there are no registers from different register sets also when
22501 mask registers are used. Update function comment.
22502 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
22503 to (*k/*r) and (*k/*km) alternatives.
22505 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
22507 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
22508 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
22509 (EH_RETURN_HANDLER_RTX): New define.
22510 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
22511 Force frame pointer in EH return functions.
22512 (aarch64_expand_epilogue): Add barrier for eh_return.
22513 (aarch64_final_eh_return_addr): Remove.
22514 (aarch64_eh_return_handler_rtx): New function.
22515 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
22517 (aarch64_eh_return_handler_rtx): New prototype.
22519 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22521 * config/rs6000/altivec.h (vec_rlmi): New #define.
22522 (vec_vrlnm): Likewise.
22523 (vec_rlnm): Likewise.
22524 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
22525 (UNSPEC_VRLNM): Likewise.
22526 (VIlong): New mode iterator.
22527 (altivec_vrl<VI_char>mi): New define_insn.
22528 (altivec_vrl<VI_char>nm): Likewise.
22529 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
22531 (VRLDNM): Likewise.
22532 (RLNM): New polymorphic function entry.
22533 (VRLWMI): New monomorphic function entry.
22534 (VRLDMI): Likewise.
22535 (RLMI): New polymorphic function entry.
22536 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
22537 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
22538 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
22541 2017-01-17 Jakub Jelinek <jakub@redhat.com>
22544 * dwarf2out.c (field_byte_offset): Restore the
22545 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
22546 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
22547 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
22550 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
22553 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
22555 2017-01-17 Jakub Jelinek <jakub@redhat.com>
22558 * dwarf2out.c (add_data_member_location_attribute): For constant
22559 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
22560 instead of DW_AT_data_member_location, DW_AT_bit_offset and
22561 DW_AT_byte_size attributes.
22563 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
22565 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
22566 after forcing to constant memory when the code model is medium.
22568 2017-01-17 Julia Koval <julia.koval@intel.com>
22571 * config/i386/avx512fintrin.h
22572 (_mm512_i32gather_ps): Change __addr type to void const*.
22573 (_mm512_mask_i32gather_ps): Ditto.
22574 (_mm512_i32gather_pd): Ditto.
22575 (_mm512_mask_i32gather_pd): Ditto.
22576 (_mm512_i64gather_ps): Ditto.
22577 (_mm512_mask_i64gather_ps): Ditto.
22578 (_mm512_i64gather_pd): Ditto.
22579 (_mm512_mask_i64gather_pd): Ditto.
22580 (_mm512_i32gather_epi32): Ditto.
22581 (_mm512_mask_i32gather_epi32): Ditto.
22582 (_mm512_i32gather_epi64): Ditto.
22583 (_mm512_mask_i32gather_epi64): Ditto.
22584 (_mm512_i64gather_epi32): Ditto.
22585 (_mm512_mask_i64gather_epi32): Ditto.
22586 (_mm512_i64gather_epi64): Ditto.
22587 (_mm512_mask_i64gather_epi64): Ditto.
22588 (_mm512_i32scatter_ps): Change __addr type to void*.
22589 (_mm512_mask_i32scatter_ps): Ditto.
22590 (_mm512_i32scatter_pd): Ditto.
22591 (_mm512_mask_i32scatter_pd): Ditto.
22592 (_mm512_i64scatter_ps): Ditto.
22593 (_mm512_mask_i64scatter_ps): Ditto.
22594 (_mm512_i64scatter_pd): Ditto.
22595 (_mm512_mask_i64scatter_pd): Ditto.
22596 (_mm512_i32scatter_epi32): Ditto.
22597 (_mm512_mask_i32scatter_epi32): Ditto.
22598 (_mm512_i32scatter_epi64): Ditto.
22599 (_mm512_mask_i32scatter_epi64): Ditto.
22600 (_mm512_i64scatter_epi32): Ditto.
22601 (_mm512_mask_i64scatter_epi32): Ditto.
22602 (_mm512_i64scatter_epi64): Ditto.
22603 (_mm512_mask_i64scatter_epi64): Ditto.
22604 * config/i386/avx512pfintrin.h
22605 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
22606 (_mm512_mask_prefetch_i32gather_ps): Ditto.
22607 (_mm512_mask_prefetch_i64gather_pd): Ditto.
22608 (_mm512_mask_prefetch_i64gather_ps): Ditto.
22609 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
22610 (_mm512_prefetch_i32scatter_ps): Ditto.
22611 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
22612 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
22613 (_mm512_prefetch_i64scatter_pd): Ditto.
22614 (_mm512_prefetch_i64scatter_ps): Ditto.
22615 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
22616 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
22617 * config/i386/avx512vlintrin.h
22618 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
22619 (_mm_mmask_i32gather_ps): Ditto.
22620 (_mm256_mmask_i32gather_pd): Ditto.
22621 (_mm_mmask_i32gather_pd): Ditto.
22622 (_mm256_mmask_i64gather_ps): Ditto.
22623 (_mm_mmask_i64gather_ps): Ditto.
22624 (_mm256_mmask_i64gather_pd): Ditto.
22625 (_mm_mmask_i64gather_pd): Ditto.
22626 (_mm256_mmask_i32gather_epi32): Ditto.
22627 (_mm_mmask_i32gather_epi32): Ditto.
22628 (_mm256_mmask_i32gather_epi64): Ditto.
22629 (_mm_mmask_i32gather_epi64): Ditto.
22630 (_mm256_mmask_i64gather_epi32): Ditto.
22631 (_mm_mmask_i64gather_epi32): Ditto.
22632 (_mm256_mmask_i64gather_epi64): Ditto.
22633 (_mm_mmask_i64gather_epi64): Ditto.
22634 (_mm256_i32scatter_ps): Change __addr type to void*.
22635 (_mm256_mask_i32scatter_ps): Ditto.
22636 (_mm_i32scatter_ps): Ditto.
22637 (_mm_mask_i32scatter_ps): Ditto.
22638 (_mm256_i32scatter_pd): Ditto.
22639 (_mm256_mask_i32scatter_pd): Ditto.
22640 (_mm_i32scatter_pd): Ditto.
22641 (_mm_mask_i32scatter_pd): Ditto.
22642 (_mm256_i64scatter_ps): Ditto.
22643 (_mm256_mask_i64scatter_ps): Ditto.
22644 (_mm_i64scatter_ps): Ditto.
22645 (_mm_mask_i64scatter_ps): Ditto.
22646 (_mm256_i64scatter_pd): Ditto.
22647 (_mm256_mask_i64scatter_pd): Ditto.
22648 (_mm_i64scatter_pd): Ditto.
22649 (_mm_mask_i64scatter_pd): Ditto.
22650 (_mm256_i32scatter_epi32): Ditto.
22651 (_mm256_mask_i32scatter_epi32): Ditto.
22652 (_mm_i32scatter_epi32): Ditto.
22653 (_mm_mask_i32scatter_epi32): Ditto.
22654 (_mm256_i32scatter_epi64): Ditto.
22655 (_mm256_mask_i32scatter_epi64): Ditto.
22656 (_mm_i32scatter_epi64): Ditto.
22657 (_mm_mask_i32scatter_epi64): Ditto.
22658 (_mm256_i64scatter_epi32): Ditto.
22659 (_mm256_mask_i64scatter_epi32): Ditto.
22660 (_mm_i64scatter_epi32): Ditto.
22661 (_mm_mask_i64scatter_epi32): Ditto.
22662 (_mm256_i64scatter_epi64): Ditto.
22663 (_mm256_mask_i64scatter_epi64): Ditto.
22664 (_mm_i64scatter_epi64): Ditto.
22665 (_mm_mask_i64scatter_epi64): Ditto.
22666 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
22667 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
22668 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
22669 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
22670 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
22671 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
22672 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
22673 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
22674 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
22675 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
22676 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
22677 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
22678 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
22679 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
22680 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
22681 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
22682 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
22683 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
22684 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
22685 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
22686 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
22687 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
22688 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
22689 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
22690 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
22691 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
22692 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
22693 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
22694 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
22695 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
22696 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
22697 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
22698 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
22699 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
22700 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
22701 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
22702 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
22703 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
22704 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
22705 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
22706 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
22707 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
22708 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
22709 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
22710 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
22711 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
22712 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
22713 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
22714 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
22715 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
22716 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
22717 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
22718 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
22719 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
22720 definitions accordingly.
22722 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
22723 Kuan-Lin Chen <kuanlinchentw@gmail.com>
22726 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
22729 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
22732 * ira-conflicts.c (ira_build_conflicts): Update total conflict
22733 hard regs for inner regno.
22735 2017-01-17 Martin Liska <mliska@suse.cz>
22738 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
22739 assumption and add comment.
22741 2017-01-17 Nathan Sidwell <nathan@acm.org>
22743 * ipa-visibility.c (localize_node): New function, broken out of ...
22744 (function_and_variable_visibility): ... here. Call it.
22746 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
22748 PR middle-end/77445
22749 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
22750 correctly set frequency of oudgoing edge.
22751 (duplicate_thread_path): Fix profile updating.
22753 2017-01-17 Jakub Jelinek <jakub@redhat.com>
22756 * configure.ac: Add GCC_BASE_VER.
22757 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
22758 version from BASE-VER file.
22759 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
22760 (gcc.o): Depend on $(BASEVER).
22761 * common.opt (dumpfullversion): New option.
22762 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
22763 * doc/invoke.texi: Document -dumpfullversion.
22764 * doc/install.texi: Document --with-gcc-major-version-only.
22765 * configure: Regenerated.
22767 2017-01-17 Richard Biener <rguenther@suse.de>
22769 PR tree-optimization/71433
22770 * tree-vrp.c (register_new_assert_for): Merge same asserts
22771 on all incoming edges.
22772 (process_assert_insertions_for): Handle insertions at the
22775 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
22777 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
22778 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
22780 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
22783 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
22786 2017-01-17 Alan Modra <amodra@gmail.com>
22789 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
22790 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
22791 symbolic stack limit when pic.
22793 2017-01-16 Martin Sebor <msebor@redhat.com>
22795 PR tree-optimization/78608
22796 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
22798 2017-01-16 Jeff Law <law@redhat.com>
22801 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
22802 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
22803 for several include directories that may be relative to sysroot.
22804 * config/i386/x-mingw32 (gplus_includedir): Define.
22805 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
22806 (native_system_includedir): Likewise.
22807 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
22808 override if TARGET_SYSTEM_ROOT is defined.
22809 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
22811 PR tree-optimization/79090
22812 PR tree-optimization/33562
22813 PR tree-optimization/61912
22814 PR tree-optimization/77485
22815 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
22816 and computed trims into the dump file.
22818 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
22820 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
22822 2017-01-16 Jakub Jelinek <jakub@redhat.com>
22825 * gimplify.c (gimplify_init_constructor): If want_value and
22826 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
22830 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
22831 sequence. Formatting fixes.
22832 (doloop_optimize): Formatting fixes.
22835 * gcc.c (debug_level_greater_than_spec_func): New function.
22836 (static_spec_functions): Add debug-level-gt spec function.
22837 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
22839 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
22840 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
22841 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
22842 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
22843 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
22844 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
22846 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
22848 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
22849 QImode fixups to general and mask registers only.
22851 2017-01-16 Carl Love <cel@us.ibm.com>
22853 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
22854 for built-in functions
22855 vector signed char vec_nabs (vector signed char)
22856 vector signed short vec_nabs (vector signed short)
22857 vector signed int vec_nabs (vector signed int)
22858 vector signed long long vec_nabs (vector signed long long)
22859 vector float vec_nabs (vector float)
22860 vector double vec_nabs (vector double)
22861 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
22863 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
22864 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
22865 * doc/extend.texi: Update the documentation file for the new built-in
22868 2017-01-16 Martin Sebor <msebor@redhat.com>
22870 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
22873 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22875 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
22876 UNSPEC_VSX__XXSPLTD to require special splat handling.
22878 2017-01-16 David Malcolm <dmalcolm@redhat.com>
22881 * system.h: Poison strndup.
22883 2017-01-16 Alan Modra <amodra@gmail.com>
22886 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
22889 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
22891 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
22893 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
22895 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
22896 call recog here. Assert that INSN_CODE (insn) is non-negative.
22898 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
22901 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
22903 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
22904 in the currently scheduled RTL fragment.
22906 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
22908 PR rtl-optimization/78751
22909 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
22912 2017-01-14 Jeff Law <law@redhat.com>
22914 PR tree-optimization/79090
22915 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
22916 variable length stores.
22917 (compute_trims): Delete dead assignment to *trim_tail.
22918 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
22921 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
22923 PR rtl-optimization/78626
22924 PR rtl-optimization/78727
22925 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
22926 of a block, and split such blocks after everything else is finished.
22928 2017-01-14 Alan Modra <amodra@gmail.com>
22931 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
22932 target legitimate_combined_insn.
22933 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
22934 (rs6000_legitimate_combined_insn): New function.
22935 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
22937 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
22938 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
22939 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
22941 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
22943 * doc/frontends.texi (G++ and GCC): Remove references to Java.
22945 2017-01-13 Jeff Law <law@redhat.com>
22947 PR tree-optimization/33562
22948 PR tree-optimization/61912
22949 PR tree-optimization/77485
22950 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
22952 (delete_dead_assignment): Likewise.
22953 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
22954 statement to delete_dead_call and delete_dead_assignment.
22956 2017-01-13 David Malcolm <dmalcolm@redhat.com>
22959 * substring-locations.c (format_warning_va): Strengthen case 1 so
22960 that both endpoints of the substring must be within the format
22961 range for just the substring to be printed.
22963 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
22965 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
22966 * config/i386/i386.c (ix86_target_string): Add missing options
22967 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
22968 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
22969 flags_other and ix86_target_other to flags2_other. Display unknown
22971 (ix86_valid_target_attribute_inner_p): Add missing options and
22972 reorder options by implied ISAs, as in ix86_target_string.
22974 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
22976 * hash-table.h (hash_table::too_empty_p): New function.
22977 (hash_table::expand): Use it.
22978 (hash_table::traverse): Likewise.
22979 (hash_table::empty_slot): Use sizeof (value_type) instead of
22980 sizeof (PTR) to convert bytes to elements. Shrink the table
22981 if the current size is excessive for the current number of
22984 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
22986 * ira-costs.c (record_reg_classes): Break from the inner loop
22987 early once alt_fail is known to be true. Update outer loop
22988 handling accordingly.
22990 2017-01-13 Jeff Law <law@redhat.com>
22992 * tree-ssa-dse.c (decrement_count): New function.
22993 (increment_start_addr, maybe_trim_memstar_call): Likewise.
22994 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
22995 when we know the partially dead statement is a mem* function.
22997 PR tree-optimization/61912
22998 PR tree-optimization/77485
22999 * tree-ssa-dse.c: Include expr.h.
23000 (maybe_trim_constructor_store): New function.
23001 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
23003 PR tree-optimization/33562
23004 PR tree-optimization/61912
23005 PR tree-optimization/77485
23006 * doc/invoke.texi: Document new dse-max-object-size param.
23007 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
23008 * tree-ssa-dse.c: Include params.h.
23009 (dse_store_status): New enum.
23010 (initialize_ao_ref_for_dse): New, partially extracted from
23012 (valid_ao_ref_for_dse, normalize_ref): New.
23013 (setup_live_bytes_from_ref, compute_trims): Likewise.
23014 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
23015 (maybe_trim_partially_dead_store): Likewise.
23016 (maybe_trim_complex_store): Likewise.
23017 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
23018 Track what bytes live from the original store. Return tri-state
23019 for dead, partially dead or live.
23020 (dse_dom_walker): Add constructor, destructor and new private members.
23021 (delete_dead_call, delete_dead_assignment): New extracted from
23023 (dse_optimize_stmt): Make a member of dse_dom_walker.
23024 Use initialize_ao_ref_for_dse.
23026 PR tree-optimization/33562
23027 PR tree-optimization/61912
23028 PR tree-optimization/77485
23029 * sbitmap.h (bitmap_count_bits): Prototype.
23030 (bitmap_clear_range, bitmap_set_range): Likewise.
23031 * sbitmap.c (bitmap_clear_range): New function.
23032 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
23034 2017-01-13 Martin Liska <mliska@suse.cz>
23037 * function.c (set_cfun): Add new argument force.
23038 * function.h (set_cfun): Likewise.
23039 * ipa-inline-transform.c (inline_call): Use the function when
23040 strict alising from is dropped for function we inline to.
23042 2017-01-13 Richard Biener <rguenther@suse.de>
23044 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
23045 for dumping GIMPLE INTEGER_CSTs.
23047 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23049 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
23050 to 201112L since C++17.
23052 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
23055 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
23056 if -fsanitize=kernel-address is present.
23058 2017-01-13 Richard Biener <rguenther@suse.de>
23060 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
23061 as _Literal ( type ) number in case usual suffixes do not
23062 preserve all information.
23064 2017-01-13 Richard Biener <rguenther@suse.de>
23066 PR tree-optimization/77283
23067 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
23068 and ssa-iterators.h.
23069 (is_feasible_trace): Implement a cost model based on joiner
23072 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
23075 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
23076 char or short to __float128/_Float128 directly.
23078 2017-01-12 Martin Sebor <msebor@redhat.com>
23080 to -Wformat-overflow.
23081 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
23082 (min_bytes_remaining): Same.
23083 (get_string_length): Same.
23084 (format_string): Same.
23085 (format_directive): Same.
23087 (pass_sprintf_length::handle_gimple_call): Same.
23089 2017-01-12 Jakub Jelinek <jakub@redhat.com>
23091 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
23092 info.nowrite calls with no lhs that can't throw. Return bool
23093 whether gsi_remove has been called or not.
23094 (pass_sprintf_length::handle_gimple_call): Return bool whether
23095 try_substitute_return_value called gsi_remove. Formatting fix.
23096 (pass_sprintf_length::execute): Don't use gsi_remove if
23097 handle_gimple_call returned true.
23100 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
23101 be removed due to side-effects, don't remove following barrier nor
23102 turn the successor edge into fallthru edge.
23104 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23107 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
23108 element-reversing loads and stores as not swappable.
23110 2017-01-12 Nathan Sidwell <nathan@acm.org>
23111 Nicolai Stange <nicstange@gmail.com>
23113 * combine.c (try_combine): Don't ignore result of overlap checking
23114 loop. Combine overlap & asm check into single loop.
23116 2017-01-12 Richard Biener <rguenther@suse.de>
23118 * tree-pretty-print.c (dump_generic_node): Provide -gimple
23119 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
23121 2017-01-12 Richard Biener <rguenther@suse.de>
23123 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
23124 and TS_TARGET_OPTION directly derive from TS_BASE.
23125 * tree-core.h (tree_optimization_option): Derive from tree_base.
23126 (tree_target_option): Likewise.
23128 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
23130 * config/i386/i386.c (memory_address_length): Increase len
23131 only when rip_relative_addr_p returns false.
23133 2017-01-11 Julia Koval <julia.koval@intel.com>
23135 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
23136 (OPTION_MASK_ISA_SGX_SET): New.
23137 (ix86_handle_option): Handle OPT_msgx.
23138 * config.gcc: Added sgxintrin.h.
23139 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
23140 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
23141 * config/i386/i386.c (ix86_target_string): Add -msgx.
23143 (ix86_option_override_internal): Handle new options.
23144 (ix86_valid_target_attribute_inner_p): Add sgx.
23145 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
23146 * config/i386/i386.opt: Add msgx.
23147 * config/i386/sgxintrin.h: New file.
23148 * config/i386/x86intrin.h: Add sgxintrin.h.
23150 2017-01-11 Jakub Jelinek <jakub@redhat.com>
23153 * fold-const.c (maybe_nonzero_address): Return 1 for function
23155 (tree_single_nonzero_warnv_p): Don't handle function local objects
23159 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
23162 2017-01-11 David Malcolm <dmalcolm@redhat.com>
23165 * opts.c: Include "spellcheck.h"
23166 (struct string_fragment): New struct.
23167 (struct edit_distance_traits<const string_fragment &>): New
23169 (get_closest_sanitizer_option): New function.
23170 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
23172 2017-01-11 Jakub Jelinek <jakub@redhat.com>
23174 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
23176 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
23177 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
23178 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
23179 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
23180 for initial die_offset if dwarf_split_debug_info.
23181 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
23182 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
23184 (output_skeleton_debug_sections): Formatting fix. Use
23185 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
23186 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
23188 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
23190 * config/arm/cortex-a53.md: Add bypasses for
23191 cortex_a53_r2f_cvt.
23192 (cortex_a53_r2f): Only use for transfers.
23193 (cortex_a53_f2r): Likewise.
23194 (cortex_a53_r2f_cvt): Add reservation for conversions.
23195 (cortex_a53_f2r_cvt): Likewise.
23197 2017-01-11 Tamar Christina <tamar.christina@arm.com>
23199 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
23200 to all inlined functions, change static to extern.
23202 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
23205 * config/arm/arm.c (legitimize_pic_address): Handle reference to
23207 (arm_assemble_integer): Likewise.
23209 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
23211 * config.gcc: Use new awk script to check CPU, FPU and architecture
23212 parameters for --with-... options.
23213 * config/arm/parsecpu.awk: New file
23214 * config/arm/arm-cpus.in: New file.
23215 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
23217 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
23219 * config/arm/t-arm: Update dependency rules.
23220 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
23221 of processing .def files.
23222 * config/arm/genopt.sh: Deleted.
23223 * config/arm/gentune.sh: Deleted.
23224 * config/arm/arm-cores.def: Deleted.
23225 * config/arm/arm-arches.def: Deleted.
23226 * config/arm/arm-fpus.def: Deleted.
23227 * config/arm/arm-tune.md: Regenerated.
23228 * config/arm/arm-tables.opt: Regenerated.
23229 * config/arm/arm-cpu.h: New generated file.
23230 * config/arm/arm-cpu-data.h: New generated file.
23231 * config/arm/arm-cpu-cdata.h: New generated file.
23233 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
23236 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
23238 (input_varpool_node): Unpack dynamically_initialized bit.
23240 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
23242 PR rtl-optimization/79032
23243 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
23244 the alignment of the adjusted memory reference against that of MODE,
23245 instead of the alignment of the original memory reference.
23247 2017-01-11 Martin Jambor <mjambor@suse.cz>
23249 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
23251 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
23252 decorated functions.
23254 2017-01-11 Richard Biener <rguenther@suse.de>
23256 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
23257 set range/nonnull info for PHI results. Do not set it on
23258 stmts marked for removal.
23260 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
23262 * expr.c (store_field): In the bitfield case, fetch the return value
23263 from the registers before applying a single big-endian adjustment.
23264 Always do a final load for a BLKmode value not larger than a word.
23266 2017-01-10 David Malcolm <dmalcolm@redhat.com>
23269 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
23270 that we correctly handle column numbers greater than
23271 LINE_MAP_MAX_COLUMN_NUMBER.
23273 2017-01-10 Martin Sebor <msebor@redhat.com>
23275 PR middle-end/78245
23276 * gimple-ssa-sprintf.c (get_destination_size): Call
23277 {init,fini}object_sizes.
23278 * tree-object-size.c (addr_object_size): Adjust.
23279 (pass_through_call): Adjust.
23280 (pass_object_sizes::execute): Adjust.
23281 * tree-object-size.h (fini_object_sizes): Declare.
23283 2017-01-10 Martin Sebor <msebor@redhat.com>
23285 PR tree-optimization/78775
23286 * builtins.c (get_size_range): Move...
23287 * calls.c: ...to here.
23288 (alloc_max_size): Accept zero argument.
23289 (operand_signed_p): Remove.
23290 (maybe_warn_alloc_args_overflow): Call get_size_range.
23291 * calls.h (get_size_range): Declare.
23293 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
23295 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
23296 from TI's devices.csv file as of September 2016.
23297 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
23299 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
23301 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
23302 * doc/invoke.texi: Likewise.
23303 * doc/md.texi: Likewise.
23304 * doc/objc.texi: Likewise.
23306 2017-01-10 Joshua Conner <joshconner@google.com>
23308 * config/arm/fuchsia-elf.h: New file.
23309 * config/fuchsia.h: New file.
23310 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
23311 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
23313 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
23315 2016-01-10 Richard Biener <rguenther@suse.de>
23317 PR tree-optimization/79034
23318 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
23319 Propagate out degenerate PHIs in the joiner.
23321 2017-01-10 Martin Liska <mliska@suse.cz>
23323 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
23324 (sort_congruence_classes_by_decl_uid): Likewise.
23325 (sort_congruence_class_groups_by_decl_uid): Likewise.
23326 (sem_item_optimizer::merge_classes): Sort class, groups in these
23327 classes and members in the groups by DECL_UID of declarations.
23328 This would make merge operations stable.
23330 2017-01-10 Martin Liska <mliska@suse.cz>
23332 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
23333 usage of m_classes_vec.
23334 (sem_item_optimizer::~sem_item_optimizer): Likewise.
23335 (sem_item_optimizer::get_group_by_hash): Likewise.
23336 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
23337 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
23338 (sem_item_optimizer::verify_classes): Likewise.
23339 (sem_item_optimizer::process_cong_reduction): Likewise.
23340 (sem_item_optimizer::dump_cong_classes): Likewise.
23341 (sem_item_optimizer::merge_classes): Likewise.
23342 * ipa-icf.h (congruence_class_hash): Rename from
23343 congruence_class_group_hash. Remove declaration of m_classes_vec.
23345 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
23347 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
23348 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
23349 * config.gcc: Add avx512vpopcntdqintrin.h.
23350 * config/i386/avx512vpopcntdqintrin.h: New.
23351 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
23352 * config/i386/i386-builtin-types.def: Add new types.
23353 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
23354 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
23355 __builtin_ia32_vpopcountq_v8di_mask): New.
23356 * config/i386/i386-c.c (ix86_target_macros_internal): Define
23357 __AVX512VPOPCNTDQ__.
23358 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
23359 (PTA_AVX512VPOPCNTDQ): Define.
23360 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
23361 TARGET_AVX512VPOPCNTDQ_P): Define.
23362 * config/i386/i386.opt: Add mavx512vpopcntdq.
23363 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
23364 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
23366 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23368 PR middle-end/77484
23369 * predict.def (PRED_CALL): Set to 67.
23371 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
23373 * expr.c (store_field): In the bitfield case, if the value comes from
23374 a function call and is of an aggregate type returned in registers, do
23375 not modify the field mode; extract the value in all cases if the mode
23376 is BLKmode and the size is not larger than a word.
23378 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
23381 * config/i386/cpuid.h: Fix undefined behavior.
23383 2017-01-04 Jeff Law <law@redhat.com>
23385 PR tree-optimization/79007
23386 PR tree-optimization/67955
23387 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
23388 conservative for pt.null when flag_non_call_exceptions is on.
23390 2017-01-09 Jakub Jelinek <jakub@redhat.com>
23392 PR translation/79019
23393 PR translation/79020
23394 * params.def (PARAM_INLINE_MIN_SPEEDUP,
23395 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
23396 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
23398 * config/avr/avr.opt (maccumulate-args): Likewise.
23399 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
23400 * common.opt (freport-bug): Likewise.
23401 * cif-code.def (CIF_FINAL_ERROR): Likewise.
23402 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
23403 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
23404 translatable string.
23405 * config/i386/i386.c (function_value_32): Likewise.
23406 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
23407 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
23409 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
23410 * common/config/msp430/msp430-common.c (msp430_handle_option):
23412 * symtab.c (symtab_node::verify_base): Likewise.
23413 * opts.c (set_debug_level): Likewise.
23414 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
23415 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
23416 missing whitespace to translatable strings.
23417 * config/avr/avr.md (bswapsi2): Fix typo in comment.
23418 * config/sh/superh.h: Likewise.
23419 * config/i386/xopintrin.h: Likewise.
23420 * config/i386/znver1.md: Likewise.
23421 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
23422 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
23423 * double-int.h (struct double_int): Likewise.
23424 * double-int.c (div_and_round_double): Likewise.
23425 * wide-int.cc: Likewise.
23426 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
23427 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
23428 * cfgcleanup.c (crossjumps_occured): Renamed to ...
23429 (crossjumps_occurred): ... this.
23430 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
23433 PR tree-optimization/78899
23434 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
23435 returning bool return struct loop *, NULL for failure and the new
23437 (versionable_outer_loop_p): Don't version outer loop if it has
23438 dont_vectorized bit set.
23439 (tree_if_conversion): When versioning outer loop, ensure
23440 tree_if_conversion is performed also on the inner loop of the
23441 non-vectorizable outer loop copy.
23442 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
23443 LOOP_VECTORIZED in inner loop of the scalar outer loop and
23444 prevent vectorization of it.
23445 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
23446 the outer loop vectorization of the non-scalar version is attempted
23447 before vectorization of the inner loop in scalar version. If
23448 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
23449 vectorization of its inner loop.
23450 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
23451 has 2 inner loops, rename also on edges from bb whose single pred
23452 is outer_loop->header. Fix typo in function comment.
23454 2017-01-09 Martin Sebor <msebor@redhat.com>
23457 * asan.c (asan_emit_stack_protection): Increase local buffer size
23458 to avoid snprintf truncation warning.
23460 2017-01-09 Andrew Pinski <apinski@cavium.com>
23462 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
23463 to reference thunderx2t99 for the tuning structure
23464 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
23466 (thunderx2t99_extra_costs): This.
23467 * config/aarch64/aarch64-tune.md: Regenerate.
23468 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
23469 (vulcan_addrcost_table): This.
23470 (vulcan_regmove_cost): Rename to ...
23471 (thunderx2t99_regmove_cost): This.
23472 (vulcan_vector_cost): Rename to ...
23473 (thunderx2t99_vector_cost): this.
23474 (vulcan_branch_cost): Rename to ...
23475 (thunderx2t99_branch_cost): This.
23476 (vulcan_tunings): Rename to ...
23477 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
23478 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
23480 2017-01-09 Martin Jambor <mjambor@suse.cz>
23484 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
23485 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
23486 (propagate_vr_accross_jump_function): Use the above function for all
23487 value range computations for pass-through jump functions and type
23488 converasion from explicit value range values.
23489 (ipcp_propagate_stage): Do not attempt to deduce types of formal
23490 parameters from TYPE_ARG_TYPES.
23491 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
23492 (ipa_write_node_info): Stream type of the actual argument.
23493 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
23495 2017-01-09 Martin Liska <mliska@suse.cz>
23498 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
23499 (lookup_compiler): Do not show error message with have_E.
23501 2017-01-09 Jakub Jelinek <jakub@redhat.com>
23503 PR tree-optimization/78938
23504 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
23505 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
23506 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
23507 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
23510 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23512 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
23515 2017-01-09 Richard Biener <rguenther@suse.de>
23517 PR tree-optimization/78997
23518 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
23519 name condition properly.
23521 2017-01-09 Richard Biener <rguenther@suse.de>
23524 * dwarf2out.c (is_cxx): New overload with context.
23525 (is_naming_typedef_decl): Use it.
23527 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
23529 * invoke.texi (Option Summary): Correct spacing in option lists
23530 and add line breaks to fix over-long lines.
23532 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
23534 PR middle-end/17660
23536 * extend.texi (Common Variable Attributes): Add xref to GCC
23537 Internals manual to explain mode attribute keywords.
23539 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
23542 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
23543 and Preprocessor Options.
23544 (Options for Linking): Document -pthread here....
23545 (RS/6000 and PowerPC Options): ...not here.
23546 (Solaris 2 Options): ...or here.
23547 * doc/cppopts.texi: Document -pthread.
23549 2017-01-08 Martin Sebor <msebor@redhat.com>
23551 PR middle-end/77708
23552 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
23553 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
23554 New member functions.
23555 (format_directive): Used them.
23557 (pass_sprintf_length::handle_gimple_call): Same.
23558 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
23559 to avoid truncation for any argument.
23560 (extract_affine_mul): Same.
23561 * tree.c (get_file_function_name): Same.
23563 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23565 PR middle-end/77484
23566 * predict.def (PRED_INDIR_CALL): Set to 86.
23568 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
23570 PR preprocessor/54124
23571 * doc/cppopts.texi: Reformat -d subtable to list the full name
23572 of the options. Add cross-reference to the docs for the general
23573 compiler -d options.
23574 * doc/invoke.texi (Developer Options): Add cross-reference to the
23575 preprocessor-specific -d option documentation.
23577 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
23579 PR preprocessor/13498
23580 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
23581 redudant material, and reflect new command-line options.
23582 (System Headers): Likewise.
23584 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
23586 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
23587 -isystem, and -idirafter. Copy-edit.
23588 * doc/cppopts.texi: Copy-edit. Remove contradiction about
23589 default for -ftrack-macro-expansion. Delete obsolete and
23590 badly-formatted implementation details about -fdebug-cpp output.
23591 * doc/cppwarnopts.texi: Copy-edit.
23593 2017-01-07 David Malcolm <dmalcolm@redhat.com>
23596 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
23597 that the transition from a max line width >= 1<<10 to narrower
23598 lines works correctly.
23600 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
23602 * doc/options.texi (PerFunction): New.
23603 * opt-functions.awk (switch_flags): Map both Optimization and
23604 PerFunction to CL_OPTIMIZATION.
23605 * opth-gen.awk: Test for PerFunction flag along with
23607 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
23608 it only when the latter is present. Skip those that don't in
23609 the hash function generator.
23610 * common.opt (fvar-tracking): Mark as PerFunction instead of
23612 (fvar-tracking-assignments): Likewise.
23613 (fvar-tracking-assignments-toggle): Likewise.
23614 (fvar-tracking-uninit): Likewise.
23616 2017-01-07 Jakub Jelinek <jakub@redhat.com>
23618 PR translation/79018
23619 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
23622 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
23625 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
23626 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
23627 TARGET_LONG_JUMP_TABLE_OFFSETS.
23628 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
23629 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
23630 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
23631 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
23632 * config/m68k/m68k.md (tablejump expander): Likewise.
23633 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
23634 TARGET_LONG_JUMP_TABLE_OFFSETS.
23635 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
23636 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
23638 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
23639 David Holsgrove <david.holsgrove@xilinx.com>
23641 * common/config/microblaze/microblaze-common.c
23642 (TARGET_EXCEPT_UNWIND_INFO): Remove.
23643 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
23645 * config/microblaze/microblaze.c (microblaze_must_save_register)
23646 (microblaze_expand_epilogue, microblaze_return_addr): Handle
23648 (microblaze_eh_return): New function.
23649 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
23650 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
23651 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
23652 * config/microblaze/microblaze.md (eh_return): New pattern.
23654 2017-01-06 Jakub Jelinek <jakub@redhat.com>
23656 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
23657 GCC_DIAGNOSTIC_STRINGIFY): Define.
23659 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
23661 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23663 * config/arm/arm.md (<mcrr>): New.
23665 * config/arm/arm.c (arm_arch5te): New.
23666 (arm_option_override): Set arm_arch5te.
23667 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
23669 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
23670 (arm_mcrr_qualifiers): ... this. New.
23671 (MRRC_QUALIFIERS): Define to...
23672 (arm_mrrc_qualifiers): ... this. New.
23673 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
23675 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
23676 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
23677 (MRRCI, mrrc, MRRC): New.
23678 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
23679 VUNSPEC_MRRC2): New.
23681 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23683 * config/arm/arm.md (<mcr>): New.
23685 * config/arm/arm.c (arm_coproc_builtin_available): Add
23686 support for mcr, mrc, mcr2 and mrc2.
23687 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
23688 (arm_mcr_qualifiers): ... this. New.
23689 (MRC_QUALIFIERS): Define to ...
23690 (arm_mrc_qualifiers): ... this. New.
23691 (MCR_QUALIFIERS): Define to ...
23692 (arm_mcr_qualifiers): ... this. New.
23693 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
23695 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
23696 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
23697 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
23698 VUNSPEC_MRC2): New.
23700 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23702 * config/arm/arm.md (*ldc): New.
23706 * config/arm/arm.c (arm_coproc_builtin_available): Add
23707 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
23708 (arm_coproc_ldc_stc_legitimate_address): New.
23709 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
23710 'qualifier_const_pointer'.
23711 (LDC_QUALIFIERS): Define to...
23712 (arm_ldc_qualifiers): ... this. New.
23713 (STC_QUALIFIERS): Define to...
23714 (arm_stc_qualifiers): ... this. New.
23715 * config/arm/arm-protos.h
23716 (arm_coproc_ldc_stc_legitimate_address): New.
23717 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
23718 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
23719 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
23720 stc2, stcl, stc2l): New.
23721 * config/arm/constraints.md (Uz): New.
23722 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
23723 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
23724 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
23725 VUNSPEC_STC2L): New.
23727 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23729 * config/arm/arm.md (<cdp>): New.
23730 * config/arm/arm.c (neon_const_bounds): Rename this ...
23731 (arm_const_bounds): ... this.
23732 (arm_coproc_builtin_available): New.
23733 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
23734 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
23735 (CDP_QUALIFIERS): Define to...
23736 (arm_cdp_qualifiers): ... this. New.
23738 (arm_expand_builtin_args): Add case for 6 arguments.
23739 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
23740 (arm_const_bounds): ... this.
23741 (arm_coproc_builtin_available): New.
23742 * config/arm/arm_acle.h (__arm_cdp): New.
23744 * config/arm/arm_acle_builtins.def (cdp): New.
23746 * config/arm/iterators.md (CDPI,CDP,cdp): New.
23747 * config/arm/neon.md: Rename all 'neon_const_bounds' to
23748 'arm_const_bounds'.
23749 * config/arm/types.md (coproc): New.
23750 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
23751 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
23752 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
23753 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
23755 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23757 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
23758 (UBINOP_QUALIFIERS): New.
23760 (acle_builtin_data): New. Change comment.
23761 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
23762 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
23763 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
23764 arm_acle_builtins.def.
23765 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
23766 (arm_init_acle_builtins): New.
23767 (CRC32_BUILTIN): Remove.
23768 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
23769 crc32cb, crc32ch and crc32cw.
23770 (arm_init_crc32_builtins): Remove.
23771 (arm_init_builtins): Use arm_init_acle_builtins rather
23772 than arm_init_crc32_builtins.
23773 (arm_expand_acle_builtin): New.
23774 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
23775 * config/arm/arm_acle_builtins.def: New.
23777 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23779 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
23780 (arm_builtin_datum): ... this.
23781 (arm_init_neon_builtin): Rename to ...
23782 (arm_init_builtin): ... this. Add a new parameters PREFIX
23783 and USE_SIG_IN_NAME.
23784 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
23785 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
23786 'arm_builtin_datum'.
23787 (arm_init_vfp_builtins): Likewise.
23788 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
23789 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
23790 (arm_expand_neon_args): Rename to ...
23791 (arm_expand_builtin_args): ... this. Rename builtin_arg
23792 enum values and differentiate between ARG_BUILTIN_MEMORY
23793 and ARG_BUILTIN_NEON_MEMORY.
23794 (arm_expand_neon_builtin_1): Rename to ...
23795 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
23796 values, arm_expand_builtin_args and add bool parameter NEON.
23797 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
23798 (arm_expand_vfp_builtin): Likewise.
23799 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
23801 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23803 PR middle-end/77484
23804 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
23805 * predict.c (tree_estimate_probability_bb): Reverse direction of
23806 polymorphic call predictor.
23808 2017-01-06 David Malcolm <dmalcolm@redhat.com>
23810 * passes.c (execute_one_pass): Split out pass-skipping logic into...
23811 (determine_pass_name_match): ...this new function and...
23812 (should_skip_pass_p): ...this new function.
23814 2017-01-06 Nathan Sidwell <nathan@acm.org>
23816 * ipa-visibility.c (function_and_variable_visibility): Reformat
23817 comments and long lines. Remove extrneous if.
23818 * symtab.c (symtab_node::make_decl_local): Fix code format.
23819 (symtab_node::set_section_for_node): Fix comment typo.
23821 2017-01-06 Martin Liska <mliska@suse.cz>
23824 * lra-constraints.c: Rename invariant to lra_invariant.
23825 * predict.c (set_even_probabilities): Initialize e to NULL.
23827 2017-01-05 Martin Sebor <msebor@redhat.com>
23829 PR tree-optimization/78910
23830 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
23831 (format_integer): Correct off-by-one error in the handling
23832 of precision with negative numbers in signed conversions..
23834 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
23836 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
23838 2017-01-05 Jakub Jelinek <jakub@redhat.com>
23840 PR tree-optimization/71016
23841 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
23842 factor_out_conditional_conversion. Formatting fix.
23843 (factor_out_conditional_conversion): Add cond_stmt argument.
23844 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
23845 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
23848 2017-01-05 David Malcolm <dmalcolm@redhat.com>
23850 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
23851 read-rtl-function.o, and selftest-rtl.o.
23852 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
23853 (selftest::aarch64_test_loading_full_dump): New function.
23854 (selftest::aarch64_run_selftests): New function.
23855 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
23856 selftest::aarch64_run_selftests.
23857 * config/i386/i386.c
23858 (selftest::ix86_test_loading_dump_fragment_1): New function.
23859 (selftest::ix86_test_loading_call_insn): New function.
23860 (selftest::ix86_test_loading_full_dump): New function.
23861 (selftest::ix86_test_loading_unspec): New function.
23862 (selftest::ix86_run_selftests): Call the new functions.
23863 * emit-rtl.c (maybe_set_max_label_num): New function.
23864 * emit-rtl.h (maybe_set_max_label_num): New decl.
23865 * function.c (instantiate_decls): Guard call to
23866 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
23867 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
23869 * gensupport.c (gen_reader::gen_reader): Pass "false"
23870 for new "compact" param of rtx_reader.
23871 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
23872 rather than an empty string for NULL strings.
23873 * read-md.c: Potentially include config.h rather than bconfig.h.
23874 Wrap include of errors.h with #ifdef GENERATOR_FILE.
23875 (have_error): New global, copied from errors.c.
23876 (md_reader::read_name): Rename to...
23877 (md_reader::read_name_1): ...this, adding "out_loc" param,
23878 and converting "missing name or number" to returning false, rather
23880 (md_reader::read_name): Reimplement in terms of read_name_1.
23881 (md_reader::read_name_or_nil): New function.
23882 (md_reader::read_string): Handle "(nil)" by returning NULL.
23883 (md_reader::md_reader): Add new param "compact".
23884 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
23885 (md_reader::read_file): New method.
23886 * read-md.h (md_reader::md_reader): Add new param "compact".
23887 (md_reader::read_file): New method.
23888 (md_reader::is_compact): New accessor.
23889 (md_reader::read_name): Convert return type from void to file_location.
23890 (md_reader::read_name_or_nil): New decl.
23891 (md_reader::read_name_1): New decl.
23892 (md_reader::m_compact): New field.
23893 (noop_reader::noop_reader): Pass "false" for new "compact" param
23895 (rtx_reader::rtx_reader): Add new "compact" param.
23896 (rtx_reader::read_rtx_operand): Make virtual and convert return
23897 type from void to rtx.
23898 (rtx_reader::read_until): New decl.
23899 (rtx_reader::handle_any_trailing_information): New virtual function.
23900 (rtx_reader::postprocess): New virtual function.
23901 (rtx_reader::finalize_string): New virtual function.
23902 (rtx_reader::m_in_call_function_usage): New field.
23903 (rtx_reader::m_reuse_rtx_by_id): New field.
23904 * read-rtl-function.c: New file.
23905 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
23906 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
23907 (selftest::verify_three_block_rtl_cfg): New decl.
23908 * read-rtl-function.h: New file.
23909 * read-rtl.c: Potentially include config.h rather than bconfig.h.
23910 For host, include function.h, memmodel.h, and emit-rtl.h.
23911 (one_time_initialization): New function.
23912 (struct compact_insn_name): New struct.
23913 (compact_insn_names): New array.
23914 (find_code): Handle insn codes in compact dumps.
23915 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
23916 (bind_subst_iter_and_attr): Likewise.
23917 (add_condition_to_string): Likewise.
23918 (add_condition_to_rtx): Likewise.
23919 (apply_attribute_uses): Likewise.
23920 (add_current_iterators): Likewise.
23921 (apply_iterators): Likewise.
23922 (initialize_iterators): Guard usage of apply_subst_iterator with
23923 #ifdef GENERATOR_FILE.
23924 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
23925 (md_reader::read_mapping): Likewise.
23926 (add_define_attr_for_define_subst): Likewise.
23927 (add_define_subst_attr): Likewise.
23928 (read_subst_mapping): Likewise.
23929 (check_code_iterator): Likewise.
23930 (rtx_reader::read_rtx): Likewise. Move one-time initialization
23932 (one_time_initialization): New function.
23933 (rtx_reader::read_until): New method.
23934 (read_flags): New function.
23935 (parse_reg_note_name): New function.
23936 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
23937 Handle reuse_rtx ids.
23938 Wrap iterator lookup within #ifdef GENERATOR_FILE.
23939 Add parsing support for RTL dumps, mirroring the special-cases in
23940 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
23941 values, and calling handle_any_trailing_information.
23942 (rtx_reader::read_rtx_operand): Convert return type from void
23943 to rtx, returning return_rtx. Handle case 'e'. Call
23944 finalize_string on XSTR and XTMPL fields.
23945 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
23946 "(nil)" values were omitted. Call the postprocess vfunc on the
23948 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
23949 class ctor. Initialize m_in_call_function_usage. Call
23950 one_time_initialization.
23951 * rtl-tests.c (selftest::test_uncond_jump): Call
23952 set_new_first_and_last_insn.
23953 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
23954 * selftest-rtl.c: New file.
23955 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
23956 (selftest::get_insn_by_uid): New decl.
23957 * selftest-run-tests.c (selftest::run_tests): Call
23958 read_rtl_function_c_tests.
23959 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
23960 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
23963 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
23965 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
23966 operands in a special way. Assert that pos+len <= mode precision.
23968 2017-01-05 Jakub Jelinek <jakub@redhat.com>
23970 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
23971 3 argument Alias with unlimited for the negative form.
23972 (fno-vect-cost-model): Removed.
23974 2017-01-05 Martin Liska <mliska@suse.cz>
23976 * hsa-gen.c (gen_hsa_divmod): New function.
23977 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
23979 2017-01-05 Martin Liska <mliska@suse.cz>
23982 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
23985 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23987 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
23988 small constant length operands.
23990 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23992 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
23993 between loop iterations.
23995 2017-01-05 Martin Liska <mliska@suse.cz>
23998 * gimplify.c (gimplify_decl_expr): Compare to
23999 asan_poisoned_variables instread of checking flags.
24000 (gimplify_target_expr): Likewise.
24001 (gimplify_expr): Likewise.
24002 (gimplify_function_tree): Conditionally initialize
24003 asan_poisoned_variables.
24005 2017-01-04 Jeff Law <law@redhat.com>
24007 PR tree-optimizatin/78812
24008 * rtl.h (contains_mem_rtx_p): Prototype.
24009 * ifcvt.c (containts_mem_rtx_p): Move from here to...
24010 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
24011 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
24012 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
24013 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
24015 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
24017 * input.c (assert_char_at_range): Default-initialize actual_range.
24019 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
24021 * df-scan.c (df_ref_create_structure): Make regno unsigned,
24022 to match the caller.
24024 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
24026 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
24027 insns after final jump in test to emit dummy move.
24029 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
24031 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
24032 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
24034 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
24036 * multiple_target.c (create_dispatcher_calls): Init e_next.
24037 * tree-ssa-loop-split.c (split_loop): Init border.
24038 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
24041 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
24046 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
24047 (altivec_register_operand): Do not return true if the operand
24048 contains a SUBREG mixing SImode and SFmode.
24049 (vsx_register_operand): Likewise.
24050 (vsx_reg_sfsubreg_ok): New predicate.
24051 (vfloat_operand): Do not return true if the operand contains a
24052 SUBREG mixing SImode and SFmode.
24053 (vint_operand): Likewise.
24054 (vlogical_operand): Likewise.
24055 (gpc_reg_operand): Likewise.
24056 (int_reg_operand): Likewise.
24057 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
24058 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
24059 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
24061 (rs6000_emit_move_si_sf_subreg): New helper function.
24062 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
24063 fixup SUBREGs involving SImode and SFmode.
24064 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
24065 numbers for the new peephole2 optimization.
24066 (peephole2 for SFmode unions): New peephole2 to optimize cases in
24067 the GLIBC math library that do AND/IOR/XOR operations on single
24068 precision floating point.
24069 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
24070 target macros to say whether we need to avoid SUBREGs mixing
24072 (TARGET_ALLOW_SF_SUBREG): Likewise.
24073 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
24074 (UNSPEC_SI_FROM_SF): Likewise.
24075 (iorxor): Change spacing.
24076 (and_ior_xor): New iterator for AND, IOR, and XOR.
24077 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
24078 (movdi_from_sf_zero_ext): Likewise.
24079 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
24080 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
24081 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
24082 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
24083 (fms<mode>4): Likewise.
24084 (fnma<mode>4): Likewise.
24085 (fnms<mode>4): Likewise.
24086 (nfma<mode>4): Likewise.
24087 (nfms<mode>4): Likewise.
24089 2017-01-04 Marek Polacek <polacek@redhat.com>
24092 * doc/invoke.texi: Document -Wpointer-compare.
24094 2017-01-04 Jakub Jelinek <jakub@redhat.com>
24096 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
24099 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
24100 descriptions for -gdwarf-5 and emit them as uleb128 instead of
24103 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
24106 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
24107 documentation of the powerpc_popcntb_ok attribute.
24108 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
24109 code to issue warning messages if a requested CPU configuration is
24110 not supported by the binary (assembler and loader) toolchain.
24111 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
24112 made to define a built-in function that has been disabled.
24113 (paired_init_builtins): Add assertion to prevent ICE if attempt is
24114 made to define a built-in function that has been disabled.
24115 (altivec_init_builtins): Add comment explaining why definition
24116 of the DST built-in functions is not preceded by an assertion
24117 check. Add assertions to prevent ICE if attempts are made to
24118 define an altivec predicate or an abs* built-in function that has
24120 (htm_init_builtins): Add comment explaining why definition of the
24121 htm built-in functions is not preceded by an assertion check.
24123 2017-01-04 Jeff Law <law@redhat.com>
24125 PR tree-optimizatin/67955
24126 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
24127 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
24128 the points-to solution does not include pt_null. Use DECL_PT_UID
24131 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
24133 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
24134 Use gen_int_mode instead of gen_lopwart for const_int operands.
24136 2017-01-04 Jakub Jelinek <jakub@redhat.com>
24138 PR tree-optimization/71563
24139 * match.pd: Simplify X << Y into X if Y is known to be 0 or
24140 out of range value - has low bits known to be zero.
24142 2017-01-04 Alan Modra <amodra@gmail.com>
24144 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
24145 * configure: Regenerate.
24146 * config.in: Regenerate.
24148 2017-01-04 Jakub Jelinek <jakub@redhat.com>
24151 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
24152 a substring of the message, but strcmp with the whole message. Ifdef
24153 ENABLE_NLS, translate the message first using dgettext.
24155 2017-01-03 Jeff Law <law@redhat.com>
24157 PR tree-optimizatin/78856
24158 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
24159 (mark_threaded_blocks): Remove code to truncate thread paths that
24160 cross multiple loop headers. Instead invalidate the cached loop
24161 iteration information and handle case of a thread path walking
24162 into an irreducible region.
24164 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
24167 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
24168 assertions. Add support for doing the signbit if the IEEE 128-bit
24169 floating point value is in a GPR.
24170 * config/rs6000/rs6000.md (Fsignbit): Delete.
24171 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
24172 Update the length attribute if the value is in a GPR.
24173 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
24174 the sign or zero extension instruction, since the value is always 0/1.
24175 (signbit<mode>2_dm2): Delete using <Fsignbit>.
24178 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
24179 extracting SImode to a GPR register so that we can generate a
24180 store, limit the vector to be in a traditional Altivec register
24181 for the vextuwrx instruction.
24183 2017-01-03 Ian Lance Taylor <iant@google.com>
24185 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
24187 2017-01-03 Martin Sebor <msebor@redhat.com>
24189 PR tree-optimization/78696
24190 * gimple-ssa-sprintf.c (format_floating): Correct handling of
24191 precision. Use MPFR for %f for greater fidelity. Correct handling
24193 (pass_sprintf_length::compute_format_length): Set width and precision
24194 specified by asrerisk to void_node for vararg functions.
24195 (try_substitute_return_value): Adjust dump output.
24197 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
24199 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
24201 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
24203 * doc/invoke.texi (SPARC options): Document -mlra as the default.
24204 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
24205 -mlra/-mno-lra was passed to the compiler.
24207 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
24209 PR rtl-optimization/65618
24210 * emit-rtl.c (try_split): Move initialization of "before" and
24211 "after" to just before the call to emit_insn_after_setloc.
24213 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
24215 * doc/md.texi (Standard Names): Remove reference to Java frontend.
24217 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
24219 * dwarf2out.c (gen_enumeration_type_die): When
24220 -gno-strict-dwarf, add a DW_AT_encoding attribute.
24222 2017-01-03 Jakub Jelinek <jakub@redhat.com>
24224 PR tree-optimization/78965
24225 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
24226 Change first argument from const call_info & to call_info &. For %n
24227 set info.nowrite to false.
24229 PR middle-end/78901
24230 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
24231 possibly throwing calls.
24233 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
24234 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
24235 and fns handling, rather than in a separate case SSA_NAME.
24237 2017-01-02 Jeff Law <law@redhat.com>
24239 * config/darwin-driver.c (darwin_driver_init): Const-correctness
24240 fixes for first_period and second_period variables.
24242 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
24245 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
24246 (*insvqi_1): New insn pattern.
24247 (*insvqi_1_mem_rex64): Ditto.
24248 (*insvqi_2): Ditto.
24249 (*insvqi_3): Rename from *insvqi.
24251 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
24253 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
24255 * doc/cfg.texi (Edges): Remove reference to Java.
24256 (Maintaining the CFG): Ditto.
24258 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24260 PR middle-end/77674
24261 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
24262 transparent aliases.
24264 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24266 PR middle-end/77484
24267 * predict.def (PRED_CALL): Update hitrate.
24268 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
24269 * predict.c (tree_estimate_probability_bb): Split CALL predictor
24270 into direct/indirect/polymorphic variants.
24272 2017-01-01 Jakub Jelinek <jakub@redhat.com>
24274 Update copyright years.
24276 * gcc.c (process_command): Update copyright notice dates.
24277 * gcov-dump.c (print_version): Ditto.
24278 * gcov.c (print_version): Ditto.
24279 * gcov-tool.c (print_version): Ditto.
24280 * gengtype.c (create_file): Ditto.
24281 * doc/cpp.texi: Bump @copying's copyright year.
24282 * doc/cppinternals.texi: Ditto.
24283 * doc/gcc.texi: Ditto.
24284 * doc/gccint.texi: Ditto.
24285 * doc/gcov.texi: Ditto.
24286 * doc/install.texi: Ditto.
24287 * doc/invoke.texi: Ditto.
24289 Copyright (C) 2017 Free Software Foundation, Inc.
24291 Copying and distribution of this file, with or without modification,
24292 are permitted in any medium without royalty provided the copyright
24293 notice and this notice are preserved.