1 2017-09-01 Jeff Law <law@redhat.com>
3 PR tree-optimization/82052
4 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
5 Always initialize the returned slot after a hash table miss
8 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
10 * config/s390/s390.md (mem_signal_fence): Remove.
11 * doc/md.texi (mem_signal_fence): Remove.
12 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
14 * target-insns.def (mem_signal_fence): Remove.
16 2017-09-01 Jakub Jelinek <jakub@redhat.com>
19 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
22 * asan.c (create_odr_indicator): Strip name encoding from assembler
23 name before appending it after __odr_asan_.
25 2017-09-01 Martin Liska <mliska@suse.cz>
27 PR tree-optimization/82059
28 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
29 frequency only when an edge is redirected.
31 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
33 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
34 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
35 (arc_conditional_register_usage): Remove ARC600 lp_count
37 (arc_file_start): Emit Tag_ARC_CPU_variation.
38 (arc_can_use_doloop_p): New conditions to use ZOLs.
39 (hwloop_fail): New function.
40 (hwloop_optimize): Likewise.
41 (hwloop_pattern_reg): Likewise.
42 (arc_doloop_hooks): New struct, to be used with reorg_loops.
43 (arc_reorg_loops): New function, calls reorg_loops.
44 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
45 (arc600_corereg_hazard): Remove ZOL checking, case handled by
47 (arc_loop_hazard): Remove function, functionality moved into
49 (arc_hazard): Remove arc_loop_hazard call.
50 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
52 (arc_label_align): Remove ZOL handling.
53 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
54 * config/arc/arc.md (doloop_begin): Remove pattern.
55 (doloop_begin_i): Likewise.
56 (doloop_end_i): Likewise.
57 (doloop_fallback): Likewise.
58 (doloop_fallback_m): Likewise.
59 (doloop_end): Reimplement expand.
60 (arc_lp): New pattern for LP instruction.
61 (loop_end): New pattern.
62 (loop_fail): Likewise.
63 (decrement_and_branch_until_zero): Likewise.
64 * config/arc/arc.opt (mlpc-width): New option.
65 * doc/invoke.texi (mlpc-width): Document option.
67 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
69 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
70 (arc_ccfsm_advance): Fix checking for delay slots.
71 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
73 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
75 * config/arc/arc.md (movqi_insn): Add stores to save constant long
77 (movhi_insn): Update store instruction constraint which are saving
78 6-bit short immediates.
79 (movsi_insn): Consider also short scaled load operations.
80 (zero_extendhisi2_i): Use Usd constraint instead of T.
81 (extendhisi2_i): Add q constraint.
82 (arc_clzsi2): Add type and length attributes.
83 (arc_ctzsi2): Likewise.
84 * config/arc/constraints.md (Usc): Update constraint, the
85 assembler can parse two relocations for a single instruction.
87 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
89 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
90 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
92 2017-08-31 Olivier Hainque <hainque@adacore.com>
94 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
95 match as powerpc-wrs-vxworks*.
97 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
99 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
100 register constraint for by-element operand.
101 (aarch64_mls_elt_merge<mode>): Likewise.
103 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
105 * config/arc/arc.c (arc_can_follow_jump): Check for short
108 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
110 * config.gcc: Use g.opt for arc.
111 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
112 functionality moved to ...
113 (legitimate_scaled_address_p): New function, ...here.
114 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
115 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
116 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
118 (arc_override_options): Handle G option.
119 (arc_output_pic_addr_const): Correct function definition.
120 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
121 (arc_decl_anon_ns_mem_p): Delete.
122 (arc_in_small_data_p): Overhaul this function to take into
123 consideration the value given via G option.
124 (arc_rewrite_small_data_1): Renamed and corrected old
125 arc_rewrite_small_data function.
126 (arc_rewrite_small_data): New function.
127 (small_data_pattern): Don't use pic_offset_table_rtx.
128 * config/arc/arc.h (CC1_SPEC): Recognize G option.
129 * config/arc/simdext.md (movmisalignv2hi): Use
130 prepare_move_operands function.
132 (movmisalign*): Likewise.
133 * doc/invoke.texi (ARC options): Document -G option.
135 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
137 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
139 * config/arc/arc.c (arc_print_operand): Output scalled address for
140 sdata whenever is possible.
141 (arc_in_small_data_p): Allow sdata for 64bit datum when double
142 load/stores are available.
143 (compact_sda_memory_operand): Check for the alignment required by
144 code density instructions.
145 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
147 * config/arc/constraints.md (Usd): Update constraint.
148 (Us0): New constraint.
149 (Usc): Update constraint.
151 2017-08-31 Richard Biener <rguenther@suse.de>
154 * dwarf2out.c (dwarf2out_early_global_decl): Process each
157 2017-08-31 Tamar Christina <tamar.christina@arm.com>
159 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
160 Resize type_signature.
162 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
163 Alan Hayward <alan.hayward@arm.com>
164 David Sherwood <david.sherwood@arm.com>
166 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
167 subregs whose inner modes can be stored in GPRs.
168 (aarch64_classify_index): Likewise.
170 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
171 Alan Hayward <alan.hayward@arm.com>
172 David Sherwood <david.sherwood@arm.com>
174 * config/aarch64/iterators.md (V_cmp_result): Rename to...
175 (V_INT_EQUIV): ...this.
176 (v_cmp_result): Rename to...
177 (v_int_equiv): ...this.
178 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
179 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
180 (copysign<mode>3): Likewise.
181 (aarch64_simd_bsl<mode>_internal): Likewise.
182 (aarch64_simd_bsl<mode>): Likewise.
183 (vec_cmp<mode><mode>): Likewise.
184 (vcond<mode><mode>): Likewise.
185 (vcond<v_cmp_mixed><mode>): Likewise.
186 (vcondu<mode><v_cmp_mixed>): Likewise.
187 (aarch64_cm<optab><mode>): Likewise.
188 (aarch64_cmtst<mode>): Likewise.
189 (aarch64_fac<optab><mode>): Likewise.
190 (vec_perm_const<mode>): Likewise.
191 (vcond_mask_<mode><v_cmp_result>): Rename to...
192 (vcond_mask_<mode><v_int_equiv>): ...this.
193 (vec_cmp<mode><v_cmp_result>): Rename to...
194 (vec_cmp<mode><v_int_equiv>): ...this.
196 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
197 Alan Hayward <alan.hayward@arm.com>
198 David Sherwood <david.sherwood@arm.com>
200 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
202 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
203 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
204 (UNSPEC_LD4_DREG): New unspecs.
205 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
206 (aarch64_ld2<mode>_dreg_be): Replace with...
207 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
209 (aarch64_ld3<mode>_dreg_le)
210 (aarch64_ld3<mode>_dreg_be): Replace with...
211 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
213 (aarch64_ld4<mode>_dreg_le)
214 (aarch64_ld4<mode>_dreg_be): Replace with...
215 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
218 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
220 PR tree-optimization/81987
221 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
222 insert an initializer in a location not dominated by the stride
225 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
227 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
228 on the entire header of the finally block in the fallthru case.
230 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
232 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
234 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
236 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
237 rs6000_emit_move_from_cr and call renamed function.
238 (rs6000_emit_prologue): Call renamed functions.
239 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
240 movesi_from_cr, remove volatile CRs.
242 2017-08-30 Jon Beniston <jon@beniston.com>
243 Richard Biener <rguenther@suse.de>
245 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
246 of VECTOR_MODE_P check.
247 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
248 element vector types.
250 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
252 * df.h (df_read_modify_subreg_p): Remove in favor of...
253 * rtl.h (read_modify_subreg_p): ...this new function. Take a
254 const_rtx instead of an rtx.
255 * cprop.c (local_cprop_find_used_regs): Update accordingly.
256 * df-problems.c (df_word_lr_mark_ref): Likewise.
257 * ira-lives.c (mark_pseudo_reg_live): Likewise.
258 (mark_pseudo_reg_dead): Likewise.
259 (mark_ref_dead): Likewise.
260 * reginfo.c (init_subregs_of_mode): Likewise.
261 * sched-deps.c (sched_analyze_1): Likewise.
262 * df-scan.c (df_def_record_1): Likewise.
263 (df_uses_record): Likewise.
264 (df_read_modify_subreg_p): Remove in favor of...
265 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
266 const_rtx instead of an rtx.
268 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
269 Alan Hayward <alan.hayward@arm.com>
270 David Sherwood <david.sherwood@arm.com>
272 * rtl.h (partial_subreg_p): New function.
273 * caller-save.c (save_call_clobbered_regs): Use it.
274 * calls.c (expand_call): Likewise.
275 * combine.c (combinable_i3pat): Likewise.
276 (simplify_set): Likewise.
277 (make_extraction): Likewise.
278 (make_compound_operation_int): Likewise.
279 (gen_lowpart_or_truncate): Likewise.
280 (force_to_mode): Likewise.
281 (make_field_assignment): Likewise.
282 (reg_truncated_to_mode): Likewise.
283 (record_truncated_value): Likewise.
284 (move_deaths): Likewise.
285 * cse.c (record_jump_cond): Likewise.
286 (cse_insn): Likewise.
287 * cselib.c (cselib_lookup_1): Likewise.
288 * expmed.c (extract_bit_field_using_extv): Likewise.
289 * function.c (assign_parm_setup_reg): Likewise.
290 * ifcvt.c (noce_convert_multiple_sets): Likewise.
291 * ira-build.c (create_insn_allocnos): Likewise.
292 * lra-coalesce.c (merge_pseudos): Likewise.
293 * lra-constraints.c (match_reload): Likewise.
294 (simplify_operand_subreg): Likewise.
295 (curr_insn_transform): Likewise.
296 * lra-lives.c (process_bb_lives): Likewise.
297 * lra.c (new_insn_reg): Likewise.
298 (lra_substitute_pseudo): Likewise.
299 * regcprop.c (mode_change_ok): Likewise.
300 (maybe_mode_change): Likewise.
301 (copyprop_hardreg_forward_1): Likewise.
302 * reload.c (push_reload): Likewise.
303 (find_reloads): Likewise.
304 (find_reloads_subreg_address): Likewise.
305 * reload1.c (alter_reg): Likewise.
306 (eliminate_regs_1): Likewise.
307 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
309 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
311 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
312 back to if statements, including unpack.
314 2017-08-30 Martin Liska <mliska@suse.cz>
317 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
319 (func_checker::compare_asm_inputs_outputs): ... this function.
320 (func_checker::compare_gimple_asm): Use the function to compare
322 * ipa-icf-gimple.h: Rename the function.
324 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
325 Alan Hayward <alan.hayward@arm.com>
326 David Sherwood <david.sherwood@arm.com>
328 * coretypes.h (complex_mode): New type.
329 * gdbhooks.py (build_pretty_printer): Handle it.
330 * machmode.h (complex_mode): New class.
331 (complex_mode::includes_p): New function.
332 (is_complex_int_mode): Likewise.
333 (is_complex_float_mode): Likewise.
334 * genmodes.c (get_mode_class): Handle complex mode classes.
335 * function.c (expand_function_end): Use is_complex_int_mode.
337 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
338 Alan Hayward <alan.hayward@arm.com>
339 David Sherwood <david.sherwood@arm.com>
341 * coretypes.h (scalar_mode_pod): New typedef.
342 * gdbhooks.py (build_pretty_printer): Handle it.
343 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
344 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
345 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
346 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
349 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
350 Alan Hayward <alan.hayward@arm.com>
351 David Sherwood <david.sherwood@arm.com>
353 * machmode.h (mode_for_vector): Take a scalar_mode instead
355 * stor-layout.c (mode_for_vector): Likewise.
356 * explow.c (promote_mode): Use as_a <scalar_mode>.
357 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
359 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
360 Alan Hayward <alan.hayward@arm.com>
361 David Sherwood <david.sherwood@arm.com>
363 * target.def (preferred_simd_mode): Take a scalar_mode
364 instead of a machine_mode.
365 * targhooks.h (default_preferred_simd_mode): Likewise.
366 * targhooks.c (default_preferred_simd_mode): Likewise.
367 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
368 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
369 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
370 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
371 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
372 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
373 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
374 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
376 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
377 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
378 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
379 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
380 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
381 * doc/tm.texi: Regenerate.
382 * optabs-query.c (can_vec_mask_load_store_p): Return false for
385 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
386 Alan Hayward <alan.hayward@arm.com>
387 David Sherwood <david.sherwood@arm.com>
389 * target.def (scalar_mode_supported_p): Take a scalar_mode
390 instead of a machine_mode.
391 * targhooks.h (default_scalar_mode_supported_p): Likewise.
392 * targhooks.c (default_scalar_mode_supported_p): Likewise.
393 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
394 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
395 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
396 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
397 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
398 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
399 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
400 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
401 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
402 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
403 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
404 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
406 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
407 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
408 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
409 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
410 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
412 * doc/tm.texi: Regenerate.
414 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
415 Alan Hayward <alan.hayward@arm.com>
416 David Sherwood <david.sherwood@arm.com>
418 * coretypes.h (opt_scalar_mode): New typedef.
419 * gdbhooks.py (build_pretty_printers): Handle it.
420 * machmode.h (mode_iterator::get_2xwider): Add overload for
422 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
424 * expr.c (convert_mode_scalar): Likewise.
425 * omp-low.c (omp_clause_aligned_alignment): Likewise.
426 * optabs.c (expand_float): Likewise.
427 (expand_fix): Likewise.
428 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
430 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
431 Alan Hayward <alan.hayward@arm.com>
432 David Sherwood <david.sherwood@arm.com>
434 * optabs.c (expand_float): Explicitly check for scalars before
435 using a branching expansion.
436 (expand_fix): Likewise.
438 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
439 Alan Hayward <alan.hayward@arm.com>
440 David Sherwood <david.sherwood@arm.com>
442 * expr.c (convert_mode): Split scalar handling out into...
443 (convert_mode_scalar): ...this new function. Treat the modes
446 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
447 Alan Hayward <alan.hayward@arm.com>
448 David Sherwood <david.sherwood@arm.com>
450 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
452 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
454 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
455 Alan Hayward <alan.hayward@arm.com>
456 David Sherwood <david.sherwood@arm.com>
458 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
459 rather than a machine_mode.
460 (fixed_from_string): Likewise.
461 (fixed_convert): Likewise.
462 (fixed_convert_from_int): Likewise.
463 (fixed_convert_from_real): Likewise.
464 (real_convert_from_fixed): Likewise.
465 * fixed-value.c (fixed_from_double_int): Likewise.
466 (fixed_from_string): Likewise.
467 (fixed_convert): Likewise.
468 (fixed_convert_from_int): Likewise.
469 (fixed_convert_from_real): Likewise.
470 (real_convert_from_fixed): Likewise.
471 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
473 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
474 Alan Hayward <alan.hayward@arm.com>
475 David Sherwood <david.sherwood@arm.com>
477 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
478 of separate mode class checks. Do not allow vector modes here.
479 (immed_wide_int_const): Use as_a <scalar_mode>.
480 * explow.c (trunc_int_for_mode): Likewise.
481 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
482 (wi::shwi): Likewise.
483 (wi::min_value): Likewise.
484 (wi::max_value): Likewise.
485 * dwarf2out.c (loc_descriptor): Likewise.
486 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
489 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
490 Alan Hayward <alan.hayward@arm.com>
491 David Sherwood <david.sherwood@arm.com>
493 * tree.h (SCALAR_TYPE_MODE): New macro.
494 * expr.c (expand_expr_addr_expr_1): Use it.
495 (expand_expr_real_2): Likewise.
496 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
497 (fold_convert_const_fixed_from_int): Likewise.
498 (fold_convert_const_fixed_from_real): Likewise.
499 (native_encode_fixed): Likewise
500 (native_encode_complex): Likewise
501 (native_encode_vector): Likewise.
502 (native_interpret_fixed): Likewise.
503 (native_interpret_real): Likewise.
504 (native_interpret_complex): Likewise.
505 (native_interpret_vector): Likewise.
506 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
507 (simd_clone_adjust_argument_types): Likewise.
508 (simd_clone_init_simd_arrays): Likewise.
509 (simd_clone_adjust): Likewise.
510 * stor-layout.c (layout_type): Likewise.
511 * tree.c (build_minus_one_cst): Likewise.
512 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
513 * tree-inline.c (estimate_move_cost): Likewise.
514 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
515 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
516 (vectorizable_reduction): Likewise.
517 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
518 (vect_recog_mixed_size_cond_pattern): Likewise.
519 (check_bool_pattern): Likewise.
520 (adjust_bool_pattern): Likewise.
521 (search_type_for_mask_1): Likewise.
522 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
523 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
524 (vectorizable_load): Likewise.
525 (vectorizable_store): Likewise.
526 * ubsan.c (ubsan_encode_value): Likewise.
527 * varasm.c (output_constant): Likewise.
529 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
530 Alan Hayward <alan.hayward@arm.com>
531 David Sherwood <david.sherwood@arm.com>
533 * coretypes.h (scalar_mode): New class.
534 * machmode.h (scalar_mode): Likewise.
535 (scalar_mode::includes_p): New function.
536 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
537 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
538 * genmodes.c (get_mode_class): Handle remaining scalar modes.
539 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
540 * expmed.c (store_bit_field_1): Likewise.
541 (extract_bit_field_1): Likewise.
542 * expr.c (write_complex_part): Likewise.
543 (read_complex_part): Likewise.
544 (emit_move_complex_push): Likewise.
545 (expand_expr_real_2): Likewise.
546 * function.c (assign_parm_setup_reg): Likewise.
547 (assign_parms_unsplit_complex): Likewise.
548 * optabs.c (expand_binop): Likewise.
549 * rtlanal.c (subreg_get_info): Likewise.
550 * simplify-rtx.c (simplify_immed_subreg): Likewise.
551 * varasm.c (output_constant_pool_2): Likewise.
553 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
554 Alan Hayward <alan.hayward@arm.com>
555 David Sherwood <david.sherwood@arm.com>
557 * expmed.c (extract_high_half): Use scalar_int_mode and remove
559 (expmed_mult_highpart_optab): Likewise.
560 (expmed_mult_highpart): Likewise.
562 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
563 Alan Hayward <alan.hayward@arm.com>
564 David Sherwood <david.sherwood@arm.com>
566 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
567 instead of a machine_mode.
568 (builtin_memset_read_str): Likewise.
569 * builtins.c (c_readstr): Likewise.
570 (builtin_memcpy_read_str): Likewise.
571 (builtin_strncpy_read_str): Likewise.
572 (builtin_memset_read_str): Likewise.
573 (builtin_memset_gen_str): Likewise.
574 (expand_builtin_signbit): Use scalar_int_mode for local variables.
575 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
576 instead of a machine_mode.
577 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
579 (make_extraction): Likewise.
580 (try_widen_shift_mode): Take and return scalar_int_modes instead
582 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
583 a scalar_int_mode instead of a machine_mode.
584 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
585 (avr_addr_space_pointer_mode): Likewise.
586 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
587 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
588 (msp430_unwind_word_mode): Likewise.
589 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
590 (spu_addr_space_pointer_mode): Likewise.
591 (spu_addr_space_address_mode): Likewise.
592 (spu_libgcc_cmp_return_mode): Likewise.
593 (spu_libgcc_shift_count_mode): Likewise.
594 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
595 (rl78_addr_space_pointer_mode): Likewise.
596 (fl78_unwind_word_mode): Likewise.
597 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
599 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
600 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
601 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
602 (mips_valid_pointer_mode): Likewise.
603 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
604 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
605 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
607 (ft32_addr_space_address_mode): Likewise.
608 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
609 scalar_int_mode instead of a machine_mode.
610 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
612 (m32c_addr_space_address_mode): Likewise.
613 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
614 (rs6000_eh_return_filter_mode): Likewise.
615 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
616 (rs6000_eh_return_filter_mode): Likewise.
617 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
618 (s390_libgcc_shift_count_mode): Likewise.
619 (s390_unwind_word_mode): Likewise.
620 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
622 * target.def (mode_rep_extended): Likewise.
623 (valid_pointer_mode): Likewise.
624 (addr_space.valid_pointer_mode): Likewise.
625 (eh_return_filter_mode): Return a scalar_int_mode rather than
627 (libgcc_cmp_return_mode): Likewise.
628 (libgcc_shift_count_mode): Likewise.
629 (unwind_word_mode): Likewise.
630 (addr_space.pointer_mode): Likewise.
631 (addr_space.address_mode): Likewise.
632 * doc/tm.texi: Regenerate.
633 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
635 (do_jump): Use scalar_int_mode for local variables.
636 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
637 rather than a machine_mode.
638 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
639 (scompare_loc_descriptor_wide): Likewise.
640 (scompare_loc_descriptor_narrow): Likewise.
641 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
643 * except.c (sjlj_emit_dispatch_table): Likewise.
644 (expand_builtin_eh_copy_values): Likewise.
645 * explow.c (convert_memory_address_addr_space_1): Likewise.
646 Take a scalar_int_mode rather than a machine_mode.
647 (convert_memory_address_addr_space): Take a scalar_int_mode rather
649 (memory_address_addr_space): Use scalar_int_mode for local variables.
650 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
651 rather than a machine_mode.
652 * expmed.c (mask_rtx): Likewise.
653 (init_expmed_one_conv): Likewise.
654 (expand_mult_highpart_adjust): Likewise.
655 (extract_high_half): Likewise.
656 (expmed_mult_highpart_optab): Likewise.
657 (expmed_mult_highpart): Likewise.
658 (expand_smod_pow2): Likewise.
659 (expand_sdiv_pow2): Likewise.
660 (emit_store_flag_int): Likewise.
661 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
663 (extract_low_bits): Likewise.
664 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
666 * expr.c (pieces_addr::adjust): Likewise.
667 (can_store_by_pieces): Likewise.
668 (store_by_pieces): Likewise.
669 (clear_by_pieces_1): Likewise.
670 (expand_expr_addr_expr_1): Likewise.
671 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
672 (expand_expr_real_1): Likewise.
673 (try_casesi): Likewise.
674 * final.c (shorten_branches): Likewise.
675 * fold-const.c (fold_convert_const_int_from_fixed): Change the
676 type of "mode" to machine_mode.
677 * internal-fn.c (expand_arith_overflow_result_store): Take a
678 scalar_int_mode rather than a machine_mode.
679 (expand_mul_overflow): Use scalar_int_mode for local variables.
680 * loop-doloop.c (doloop_modify): Likewise.
681 (doloop_optimize): Likewise.
682 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
684 (expand_doubleword_shift_condmove): Likewise.
685 (expand_doubleword_shift): Likewise.
686 (expand_doubleword_clz): Likewise.
687 (expand_doubleword_popcount): Likewise.
688 (expand_doubleword_parity): Likewise.
689 (expand_absneg_bit): Use scalar_int_mode for local variables.
690 (prepare_float_lib_cmp): Likewise.
691 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
692 rather than a machine_mode.
693 (convert_memory_address_addr_space): Likewise.
694 (get_mode_bounds): Likewise.
695 (get_address_mode): Return a scalar_int_mode rather than a
697 * rtlanal.c (get_address_mode): Likewise.
698 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
700 * targhooks.c (default_mode_rep_extended): Likewise.
701 (default_valid_pointer_mode): Likewise.
702 (default_addr_space_valid_pointer_mode): Likewise.
703 (default_eh_return_filter_mode): Return a scalar_int_mode rather
705 (default_libgcc_cmp_return_mode): Likewise.
706 (default_libgcc_shift_count_mode): Likewise.
707 (default_unwind_word_mode): Likewise.
708 (default_addr_space_pointer_mode): Likewise.
709 (default_addr_space_address_mode): Likewise.
710 * targhooks.h (default_eh_return_filter_mode): Likewise.
711 (default_libgcc_cmp_return_mode): Likewise.
712 (default_libgcc_shift_count_mode): Likewise.
713 (default_unwind_word_mode): Likewise.
714 (default_addr_space_pointer_mode): Likewise.
715 (default_addr_space_address_mode): Likewise.
716 (default_mode_rep_extended): Take a scalar_int_mode rather than
718 (default_valid_pointer_mode): Likewise.
719 (default_addr_space_valid_pointer_mode): Likewise.
720 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
722 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
723 rather than a machine_mode.
724 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
726 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
727 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
730 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
731 Alan Hayward <alan.hayward@arm.com>
732 David Sherwood <david.sherwood@arm.com>
734 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
735 the mode argument to scalar_int_mode.
736 (do_jump_by_parts_zero_rtx): Likewise.
737 (do_jump_by_parts_equality_rtx): Likewise.
738 (do_jump_by_parts_greater): Take a mode argument.
739 (do_jump_by_parts_equality): Likewise.
740 (do_jump_1): Update calls accordingly.
742 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
743 Alan Hayward <alan.hayward@arm.com>
744 David Sherwood <david.sherwood@arm.com>
746 * is-a.h (safe_dyn_cast): New function.
747 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
748 (jump_table_for_label): Likewise.
749 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
750 instead of an rtx_insn *.
751 (shorten_branches): Use dyn_cast instead of LABEL_P and
752 JUMP_TABLE_DATA_P. Use jump_table_for_label and
753 rtx_jump_table_data::get_data_mode.
754 (final_scan_insn): Likewise.
756 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
757 Alan Hayward <alan.hayward@arm.com>
758 David Sherwood <david.sherwood@arm.com>
760 * combine.c (try_combine): Use is_a <scalar_int_mode> when
761 trying to combine a full-register integer set with a subreg
764 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
765 Alan Hayward <alan.hayward@arm.com>
766 David Sherwood <david.sherwood@arm.com>
768 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
769 that is always either address_mode or pointer_mode.
771 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
772 Alan Hayward <alan.hayward@arm.com>
773 David Sherwood <david.sherwood@arm.com>
775 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
776 when the two are known to be equal.
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 * simplify-rtx.c (simplify_const_unary_operation): Use
783 is_a <scalar_int_mode> instead of checking for a nonzero
784 precision. Forcibly convert op_mode to a scalar_int_mode
785 in that case. More clearly differentiate the operand and
786 result modes and use the former when deciding what the value
787 of a count-bits operation should be. Use is_int_mode instead
788 of checking for a MODE_INT. Remove redundant check for whether
789 this mode has a zero precision.
791 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
792 Alan Hayward <alan.hayward@arm.com>
793 David Sherwood <david.sherwood@arm.com>
795 * optabs.c (widen_leading): Change the type of the mode argument
796 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
797 (widen_bswap): Likewise.
798 (expand_parity): Likewise.
799 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
800 (expand_ffs): Likewise.
801 (epand_unop): Check for scalar integer modes before calling the
804 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
805 Alan Hayward <alan.hayward@arm.com>
806 David Sherwood <david.sherwood@arm.com>
808 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
810 (expand_expr_real_1): Update call accordingly.
812 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
813 Alan Hayward <alan.hayward@arm.com>
814 David Sherwood <david.sherwood@arm.com>
816 * expmed.c (store_bit_field_using_insv): Add op0_mode and
817 value_mode arguments. Use scalar_int_mode internally.
818 (store_bit_field_1): Rename the new integer mode from imode
819 to op0_mode and use it instead of GET_MODE (op0). Update calls
820 to store_split_bit_field, store_bit_field_using_insv and
821 store_fixed_bit_field.
822 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
823 Use scalar_int_mode internally. Use a bit count rather than a mode
824 when calculating the largest bit size for get_best_mode.
825 Update calls to store_split_bit_field and store_fixed_bit_field_1.
826 (store_fixed_bit_field_1): Add mode and value_mode arguments.
827 Remove assertion that OP0 has a scalar integer mode.
828 (store_split_bit_field): Add op0_mode and value_mode arguments.
829 Update calls to extract_fixed_bit_field.
830 (extract_bit_field_using_extv): Add an op0_mode argument.
831 Use scalar_int_mode internally.
832 (extract_bit_field_1): Rename the new integer mode from imode to
833 op0_mode and use it instead of GET_MODE (op0). Update calls to
834 extract_split_bit_field, extract_bit_field_using_extv and
835 extract_fixed_bit_field.
836 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
837 to extract_split_bit_field and extract_fixed_bit_field_1.
838 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
839 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
841 (extract_split_bit_field): Add an op0_mode argument. Update call
842 to extract_fixed_bit_field.
844 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
845 Alan Hayward <alan.hayward@arm.com>
846 David Sherwood <david.sherwood@arm.com>
848 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
849 * explow.c (hard_function_value): Likewise.
850 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
851 convert_to_mode call outside the loop.
852 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
853 for the mode iterator. Require the mode specified by max_pieces
855 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
857 (copy_blkmode_to_reg): Likewise.
858 (set_storage_via_setmem): Likewise.
859 * optabs.c (prepare_cmp_insn): Likewise.
860 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
861 * stor-layout.c (finish_bitfield_representative): Likewise.
863 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
864 Alan Hayward <alan.hayward@arm.com>
865 David Sherwood <david.sherwood@arm.com>
867 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
868 * expr.c (convert_move): Use them.
869 (convert_modes): Likewise.
870 (store_expr_with_bounds): Likewise.
872 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
873 Alan Hayward <alan.hayward@arm.com>
874 David Sherwood <david.sherwood@arm.com>
876 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
877 parameter for the mode of "x". Remove the "known_x", "known_mode"
878 and "known_ret" arguments. Change the type of the mode argument
880 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
881 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
882 (reg_num_sign_bit_copies_for_combine): Likewise.
883 * rtlanal.c (nonzero_bits1): Likewise.
884 (num_sign_bit_copies1): Likewise.
885 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
886 (reg_num_sign_bit_copies_general): Likewise.
887 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
888 (reg_nonzero_bits_general): Likewise.
890 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
891 Alan Hayward <alan.hayward@arm.com>
892 David Sherwood <david.sherwood@arm.com>
894 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
895 than in subroutines. Return 1 for non-integer modes.
896 (cached_num_sign_bit_copies): Change the type of the mode parameter
898 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
899 classes. Handle CONST_INT_P first and then check whether X also
900 has a scalar integer mode. Check the same thing for inner registers
901 of a SUBREG and for values that are being extended or truncated.
903 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
904 Alan Hayward <alan.hayward@arm.com>
905 David Sherwood <david.sherwood@arm.com>
907 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
908 in subroutines. Return the mode mask for non-integer modes.
909 (cached_nonzero_bits): Change the type of the mode parameter
911 (nonzero_bits1): Likewise. Remove early exit for other mode
912 classes. Handle CONST_INT_P first and then check whether X
913 also has a scalar integer mode.
915 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
916 Alan Hayward <alan.hayward@arm.com>
917 David Sherwood <david.sherwood@arm.com>
919 * expr.c (widest_int_mode_for_size): Make the comment match the code.
920 Return a scalar_int_mode and assert that the size is greater than
922 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
923 (op_by_pieces_d::op_by_pieces_d): Likewise.
924 (op_by_pieces_d::run): Likewise.
925 (can_store_by_pieces): Likewise.
927 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
928 Alan Hayward <alan.hayward@arm.com>
929 David Sherwood <david.sherwood@arm.com>
931 * combine.c (extract_left_shift): Add a mode argument and update
933 (make_compound_operation_int): Change the type of the mode parameter
934 to scalar_int_mode and update the call to extract_left_shift.
936 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
937 Alan Hayward <alan.hayward@arm.com>
938 David Sherwood <david.sherwood@arm.com>
940 * combine.c (simplify_and_const_int): Change the type of the mode
941 parameter to scalar_int_mode.
942 (simplify_and_const_int_1): Likewise. Update recursive call.
944 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
945 Alan Hayward <alan.hayward@arm.com>
946 David Sherwood <david.sherwood@arm.com>
948 * combine.c (simplify_compare_const): Check that the mode is a
949 scalar_int_mode (rather than VOIDmode) before testing its
951 (simplify_comparison): Move COMPARISON_P handling out of the
952 loop and restrict the latter part of the loop to scalar_int_modes.
953 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
954 and when considering SUBREG_REGs. Use is_int_mode instead of
955 checking GET_MODE_CLASS against MODE_INT.
957 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
958 Alan Hayward <alan.hayward@arm.com>
959 David Sherwood <david.sherwood@arm.com>
961 * combine.c (try_widen_shift_mode): Move check for equal modes to...
962 (simplify_shift_const_1): ...here. Use scalar_int_mode for
963 shift_unit_mode and for modes involved in scalar shifts.
965 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
966 Alan Hayward <alan.hayward@arm.com>
967 David Sherwood <david.sherwood@arm.com>
969 * combine.c (force_int_to_mode): New function, split out from...
970 (force_to_mode): ...here. Keep xmode up-to-date and use it
971 instead of GET_MODE (x).
973 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
974 Alan Hayward <alan.hayward@arm.com>
975 David Sherwood <david.sherwood@arm.com>
977 * optabs-query.h (extraction_insn::struct_mode): Change type to
978 opt_scalar_int_mode and update comment.
979 (extraction_insn::field_mode): Change type to scalar_int_mode.
980 (extraction_insn::pos_mode): Likewise.
981 * combine.c (make_extraction): Update accordingly.
982 * optabs-query.c (get_traditional_extraction_insn): Likewise.
983 (get_optab_extraction_insn): Likewise.
984 * recog.c (simplify_while_replacing): Likewise.
985 * expmed.c (narrow_bit_field_mem): Change the type of the mode
986 parameter to opt_scalar_int_mode.
988 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
989 Alan Hayward <alan.hayward@arm.com>
990 David Sherwood <david.sherwood@arm.com>
992 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
993 to a scalar_int_mode instead of a machine_mode.
994 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
995 (get_best_mode): Return a boolean and use a pointer argument to store
996 the selected mode. Replace the limit mode parameter with a bit limit.
997 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
998 for the values returned by bit_field_mode_iterator::next_mode.
999 (store_bit_field): Update call to get_best_mode.
1000 (store_fixed_bit_field): Likewise.
1001 (extract_fixed_bit_field): Likewise.
1002 * expr.c (optimize_bitfield_assignment_op): Likewise.
1003 * fold-const.c (optimize_bit_field_compare): Likewise.
1004 (fold_truth_andor_1): Likewise.
1005 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
1006 Update for new type of m_mode.
1007 (get_best_mode): As above.
1009 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1010 Alan Hayward <alan.hayward@arm.com>
1011 David Sherwood <david.sherwood@arm.com>
1013 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
1014 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
1015 (store_bit_field): Check is_a <scalar_int_mode> before calling
1016 strict_volatile_bitfield_p.
1017 (extract_bit_field): Likewise.
1019 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1020 Alan Hayward <alan.hayward@arm.com>
1021 David Sherwood <david.sherwood@arm.com>
1023 * target.def (cstore_mode): Return a scalar_int_mode.
1024 * doc/tm.texi: Regenerate.
1025 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
1026 * targhooks.h (default_cstore_mode): Likewise.
1027 * targhooks.c (default_cstore_mode): Likewise, using a forced
1029 * expmed.c (emit_cstore): Expect the target of the cstore to be
1032 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1033 Alan Hayward <alan.hayward@arm.com>
1034 David Sherwood <david.sherwood@arm.com>
1036 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
1038 (niter_desc): Likewise mode.
1039 (iv_analyze): Add a mode parameter.
1041 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
1042 and change its type to scalar_int_mode.
1043 * loop-iv.c: Update commentary at head of file.
1044 (iv_constant): Pass the mode paraeter before the rtx it describes
1045 and change its type to scalar_int_mode. Remove VOIDmode handling.
1046 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
1047 (iv_extend): Likewise.
1048 (shorten_into_mode): Likewise.
1049 (iv_add): Use scalar_int_mode.
1050 (iv_mult): Likewise.
1051 (iv_shift): Likewise.
1052 (canonicalize_iv_subregs): Likewise.
1053 (get_biv_step_1): Pass the outer_mode parameter before the rtx
1054 it describes and change its mode to scalar_int_mode. Also change
1055 the type of the returned inner_mode to scalar_int_mode.
1056 (get_biv_step): Likewise, turning outer_mode from a pointer
1057 into a direct parameter. Update call to get_biv_step_1.
1058 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
1059 iv_constant and get_biv_step.
1060 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
1061 and change its type to scalar_int_mode. Don't initialise iv->mode
1062 to VOIDmode and remove later checks for its still being VOIDmode.
1063 Update calls to iv_analyze_op and iv_analyze_expr. Check
1064 is_a <scalar_int_mode> when changing the mode under consideration.
1065 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
1066 Update call to iv_analyze_expr.
1067 (iv_analyze_op): Add a mode parameter. Reject subregs whose
1068 inner register is not also a scalar_int_mode. Update call to
1070 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
1071 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
1072 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
1073 separate mode class checks. Update calls to iv_analyze. Remove
1074 fix-up of VOIDmodes after iv_analyze_biv.
1075 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
1076 don't have a scalar_int_mode. Update call to biv_p.
1078 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1079 Alan Hayward <alan.hayward@arm.com>
1080 David Sherwood <david.sherwood@arm.com>
1082 * cfgexpand.c (convert_debug_memory_address): Use
1083 as_a <scalar_int_mode>.
1084 * combine.c (expand_compound_operation): Likewise.
1085 (make_extraction): Likewise.
1086 (change_zero_ext): Likewise.
1087 (simplify_comparison): Likewise.
1088 * cse.c (cse_insn): Likewise.
1089 * dwarf2out.c (minmax_loc_descriptor): Likewise.
1090 (mem_loc_descriptor): Likewise.
1091 (loc_descriptor): Likewise.
1092 * expmed.c (init_expmed_one_mode): Likewise.
1093 (synth_mult): Likewise.
1094 (emit_store_flag_1): Likewise.
1095 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
1096 of a comparison with size.
1097 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
1098 (reduce_to_bit_field_precision): Likewise.
1099 * function.c (expand_function_end): Likewise.
1100 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
1101 * loop-doloop.c (doloop_modify): Likewise.
1102 * optabs.c (expand_binop): Likewise.
1103 (expand_unop): Likewise.
1104 (expand_copysign_absneg): Likewise.
1105 (prepare_cmp_insn): Likewise.
1106 (maybe_legitimize_operand): Likewise.
1107 * recog.c (const_scalar_int_operand): Likewise.
1108 * rtlanal.c (get_address_mode): Likewise.
1109 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1110 (simplify_cond_clz_ctz): Likewise.
1111 * tree-nested.c (get_nl_goto_field): Likewise.
1112 * tree.c (build_vector_type_for_mode): Likewise.
1113 * var-tracking.c (use_narrower_mode): Likewise.
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 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
1120 * builtins.c (expand_builtin_signbit): Use it.
1121 * cfgexpand.c (expand_debug_expr): Likewise.
1122 * dojump.c (do_jump): Likewise.
1123 (do_compare_and_jump): Likewise.
1124 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
1125 * expmed.c (make_tree): Likewise.
1126 * expr.c (expand_expr_real_2): Likewise.
1127 (expand_expr_real_1): Likewise.
1128 (try_casesi): Likewise.
1129 * fold-const-call.c (fold_const_call_ss): Likewise.
1130 * fold-const.c (unextend): Likewise.
1131 (extract_muldiv_1): Likewise.
1132 (fold_single_bit_test): Likewise.
1133 (native_encode_int): Likewise.
1134 (native_encode_string): Likewise.
1135 (native_interpret_int): Likewise.
1136 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
1137 * internal-fn.c (expand_addsub_overflow): Likewise.
1138 (expand_neg_overflow): Likewise.
1139 (expand_mul_overflow): Likewise.
1140 (expand_arith_overflow): Likewise.
1141 * match.pd: Likewise.
1142 * stor-layout.c (layout_type): Likewise.
1143 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
1144 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
1145 * tree-ssanames.c (get_range_info): Likewise.
1146 * tree-switch-conversion.c (array_value_type) Likewise.
1147 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
1148 (vect_recog_divmod_pattern): Likewise.
1149 (vect_recog_mixed_size_cond_pattern): Likewise.
1150 * tree-vrp.c (extract_range_basic): Likewise.
1151 (simplify_float_conversion_using_ranges): Likewise.
1152 * tree.c (int_fits_type_p): Likewise.
1153 * ubsan.c (instrument_bool_enum_load): Likewise.
1154 * varasm.c (mergeable_string_section): Likewise.
1155 (narrowing_initializer_constant_valid_p): Likewise.
1156 (output_constant): Likewise.
1158 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1159 Alan Hayward <alan.hayward@arm.com>
1160 David Sherwood <david.sherwood@arm.com>
1162 * machmode.h (NARROWEST_INT_MODE): New macro.
1163 * expr.c (alignment_for_piecewise_move): Use it instead of
1164 GET_CLASS_NARROWEST_MODE (MODE_INT).
1165 (push_block): Likewise.
1166 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
1168 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1170 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1171 Alan Hayward <alan.hayward@arm.com>
1172 David Sherwood <david.sherwood@arm.com>
1174 * postreload.c (move2add_valid_value_p): Change the type of the
1175 mode parameter to scalar_int_mode.
1176 (move2add_use_add2_insn): Add a mode parameter and use it instead
1178 (move2add_use_add3_insn): Likewise.
1179 (reload_cse_move2add): Update accordingly.
1181 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1182 Alan Hayward <alan.hayward@arm.com>
1183 David Sherwood <david.sherwood@arm.com>
1185 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
1187 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
1188 * optabs.c (expand_unop): Likewise.
1190 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1191 Alan Hayward <alan.hayward@arm.com>
1192 David Sherwood <david.sherwood@arm.com>
1194 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
1195 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
1196 (popcount_loc_descriptor): Likewise.
1197 (bswap_loc_descriptor): Likewise.
1198 (rotate_loc_descriptor): Likewise.
1199 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
1200 calling the functions above.
1202 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1203 Alan Hayward <alan.hayward@arm.com>
1204 David Sherwood <david.sherwood@arm.com>
1206 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
1208 (try_combine): Likewise.
1209 (simplify_if_then_else): Likewise.
1210 * cse.c (cse_insn): Likewise.
1211 * dwarf2out.c (mem_loc_descriptor): Likewise.
1212 * emit-rtl.c (gen_lowpart_common): Likewise.
1213 * simplify-rtx.c (simplify_truncation): Likewise.
1214 (simplify_binary_operation_1): Likewise.
1215 (simplify_const_relational_operation): Likewise.
1216 (simplify_ternary_operation): Likewise.
1217 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
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 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
1224 * reload.c (push_reload): Likewise.
1225 (find_reloads): Likewise.
1227 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1228 Alan Hayward <alan.hayward@arm.com>
1229 David Sherwood <david.sherwood@arm.com>
1231 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
1232 (make_compound_operation_int): Likewise.
1233 (change_zero_ext): Likewise.
1234 * expr.c (convert_move): Likewise.
1235 (convert_modes): Likewise.
1236 * fwprop.c (forward_propagate_subreg): Likewise.
1237 * loop-iv.c (get_biv_step_1): Likewise.
1238 * optabs.c (widen_operand): Likewise.
1239 * postreload.c (move2add_valid_value_p): Likewise.
1240 * recog.c (simplify_while_replacing): Likewise.
1241 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1242 (simplify_binary_operation_1): Likewise. Remove redundant
1243 mode equality check.
1245 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1246 Alan Hayward <alan.hayward@arm.com>
1247 David Sherwood <david.sherwood@arm.com>
1249 * combine.c (combine_simplify_rtx): Add checks for
1250 is_a <scalar_int_mode>.
1251 (simplify_if_then_else): Likewise.
1252 (make_field_assignment): Likewise.
1253 (simplify_comparison): Likewise.
1254 * ifcvt.c (noce_try_bitop): Likewise.
1255 * loop-invariant.c (canonicalize_address_mult): Likewise.
1256 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1258 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1259 Alan Hayward <alan.hayward@arm.com>
1260 David Sherwood <david.sherwood@arm.com>
1262 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
1263 is_a <scalar_int_mode> instead of != BLKmode.
1265 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1266 Alan Hayward <alan.hayward@arm.com>
1267 David Sherwood <david.sherwood@arm.com>
1269 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
1270 instead of != VOIDmode.
1271 * combine.c (if_then_else_cond): Likewise.
1272 (change_zero_ext): Likewise.
1273 * dwarf2out.c (mem_loc_descriptor): Likewise.
1274 (loc_descriptor): Likewise.
1275 * rtlanal.c (canonicalize_condition): Likewise.
1276 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
1278 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1279 Alan Hayward <alan.hayward@arm.com>
1280 David Sherwood <david.sherwood@arm.com>
1282 * simplify-rtx.c (simplify_binary_operation_1): Use
1283 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
1285 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1286 Alan Hayward <alan.hayward@arm.com>
1287 David Sherwood <david.sherwood@arm.com>
1289 * wide-int.h (int_traits<unsigned char>) New class.
1290 (int_traits<unsigned short>) Likewise.
1291 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
1292 Use GET_MODE_UNIT_PRECISION and remove redundant test for
1294 * combine.c (set_nonzero_bits_and_sign_copies): Use
1295 is_a <scalar_int_mode>.
1296 (find_split_point): Likewise.
1297 (combine_simplify_rtx): Likewise.
1298 (simplify_logical): Likewise.
1299 (expand_compound_operation): Likewise.
1300 (expand_field_assignment): Likewise.
1301 (make_compound_operation): Likewise.
1302 (extended_count): Likewise.
1303 (change_zero_ext): Likewise.
1304 (simplify_comparison): Likewise.
1305 * dwarf2out.c (scompare_loc_descriptor): Likewise.
1306 (ucompare_loc_descriptor): Likewise.
1307 (minmax_loc_descriptor): Likewise.
1308 (mem_loc_descriptor): Likewise.
1309 (loc_descriptor): Likewise.
1310 * expmed.c (init_expmed_one_mode): Likewise.
1311 * lra-constraints.c (lra_constraint_offset): Likewise.
1312 * optabs.c (prepare_libcall_arg): Likewise.
1313 * postreload.c (move2add_note_store): Likewise.
1314 * reload.c (operands_match_p): Likewise.
1315 * rtl.h (load_extend_op): Likewise.
1316 * rtlhooks.c (gen_lowpart_general): Likewise.
1317 * simplify-rtx.c (simplify_truncation): Likewise.
1318 (simplify_unary_operation_1): Likewise.
1319 (simplify_binary_operation_1): Likewise.
1320 (simplify_const_binary_operation): Likewise.
1321 (simplify_const_relational_operation): Likewise.
1322 (simplify_subreg): Likewise.
1323 * stor-layout.c (bitwise_mode_for_mode): Likewise.
1324 * var-tracking.c (adjust_mems): Likewise.
1325 (prepare_call_arguments): Likewise.
1327 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1328 Alan Hayward <alan.hayward@arm.com>
1329 David Sherwood <david.sherwood@arm.com>
1331 * machmode.h (is_int_mode): New fuction.
1332 * combine.c (find_split_point): Use it.
1333 (combine_simplify_rtx): Likewise.
1334 (simplify_if_then_else): Likewise.
1335 (simplify_set): Likewise.
1336 (simplify_shift_const_1): Likewise.
1337 (simplify_comparison): Likewise.
1338 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
1339 * cse.c (notreg_cost): Likewise.
1340 (cse_insn): Likewise.
1341 * cselib.c (cselib_lookup_1): Likewise.
1342 * dojump.c (do_jump_1): Likewise.
1343 (do_compare_rtx_and_jump): Likewise.
1344 * dse.c (get_call_args): Likewise.
1345 * dwarf2out.c (rtl_for_decl_init): Likewise.
1346 (native_encode_initializer): Likewise.
1347 * expmed.c (emit_store_flag_1): Likewise.
1348 (emit_store_flag): Likewise.
1349 * expr.c (convert_modes): Likewise.
1350 (store_field): Likewise.
1351 (expand_expr_real_1): Likewise.
1352 * fold-const.c (fold_read_from_constant_string): Likewise.
1353 * gimple-ssa-sprintf.c (get_format_string): Likewise.
1354 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
1355 * optabs.c (expand_binop): Likewise.
1356 (expand_unop): Likewise.
1357 (expand_abs_nojump): Likewise.
1358 (expand_one_cmpl_abs_nojump): Likewise.
1359 * simplify-rtx.c (mode_signbit_p): Likewise.
1360 (val_signbit_p): Likewise.
1361 (val_signbit_known_set_p): Likewise.
1362 (val_signbit_known_clear_p): Likewise.
1363 (simplify_relational_operation_1): Likewise.
1364 * tree.c (vector_type_mode): Likewise.
1366 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1367 Alan Hayward <alan.hayward@arm.com>
1368 David Sherwood <david.sherwood@arm.com>
1370 * machmode.h (smallest_mode_for_size): Fix formatting.
1371 (smallest_int_mode_for_size): New function.
1372 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
1373 instead of smallest_mode_for_size.
1374 * combine.c (make_extraction): Likewise.
1375 * config/arc/arc.c (arc_expand_movmem): Likewise.
1376 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
1377 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
1378 * config/s390/s390.c (s390_expand_insv): Likewise.
1379 * config/sparc/sparc.c (assign_int_registers): Likewise.
1380 * config/spu/spu.c (spu_function_value): Likewise.
1381 (spu_function_arg): Likewise.
1382 * coverage.c (get_gcov_type): Likewise.
1383 (get_gcov_unsigned_t): Likewise.
1384 * dse.c (find_shift_sequence): Likewise.
1385 * expmed.c (store_bit_field_1): Likewise.
1386 * expr.c (convert_move): Likewise.
1387 (store_field): Likewise.
1388 * internal-fn.c (expand_arith_overflow): Likewise.
1389 * optabs-query.c (get_best_extraction_insn): Likewise.
1390 * optabs.c (expand_twoval_binop_libfunc): Likewise.
1391 * stor-layout.c (layout_type): Likewise.
1392 (initialize_sizetypes): Likewise.
1393 * targhooks.c (default_get_mask_mode): Likewise.
1394 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
1396 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1397 Alan Hayward <alan.hayward@arm.com>
1398 David Sherwood <david.sherwood@arm.com>
1400 * machmode.h (opt_mode::else_blk): New function.
1401 (int_mode_for_mode): Declare.
1402 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
1403 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
1405 * cfgexpand.c (expand_debug_expr): Likewise.
1406 * combine.c (gen_lowpart_or_truncate): Likewise.
1407 (gen_lowpart_for_combine): Likewise.
1408 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
1409 * config/avr/avr.c (avr_to_int_mode): Likewise.
1410 (avr_out_plus_1): Likewise.
1411 (avr_out_plus): Likewise.
1412 (avr_out_round): Likewise.
1413 * config/i386/i386.c (ix86_split_to_parts): Likewise.
1414 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
1415 (s390_expand_vcond): Likewise.
1416 * config/spu/spu.c (spu_split_immediate): Likewise.
1417 (spu_expand_mov): Likewise.
1418 * dse.c (get_stored_val): Likewise.
1419 * expmed.c (store_bit_field_1): Likewise.
1420 (convert_extracted_bit_field): Use int_mode_for_mode instead of
1422 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
1423 (extract_low_bits): Likewise.
1424 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
1425 handling rather than repeating the check.
1426 (emit_group_store): Likewise.
1427 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
1428 * optabs.c (expand_absneg_bit): Likewise.
1429 (expand_copysign_absneg): Likewise.
1430 (expand_copysign_bit): Likewise.
1431 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
1432 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
1433 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
1434 * var-tracking.c (prepare_call_arguments): Likewise.
1435 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
1436 int_mode_for_mode instead of mode_for_size.
1437 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
1439 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1440 Alan Hayward <alan.hayward@arm.com>
1441 David Sherwood <david.sherwood@arm.com>
1443 * machmode.h (int_mode_for_size): New function.
1444 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
1445 instead of mode_for_size.
1446 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
1448 * combine.c (expand_field_assignment): Use int_mode_for_size
1449 instead of mode_for_size.
1450 (make_extraction): Likewise.
1451 (simplify_shift_const_1): Likewise.
1452 (simplify_comparison): Likewise.
1453 * dojump.c (do_jump): Likewise.
1454 * dwarf2out.c (mem_loc_descriptor): Likewise.
1455 * emit-rtl.c (init_derived_machine_modes): Likewise.
1456 * expmed.c (flip_storage_order): Likewise.
1457 (convert_extracted_bit_field): Likewise.
1458 * expr.c (copy_blkmode_from_reg): Likewise.
1459 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
1460 * internal-fn.c (expand_mul_overflow): Likewise.
1461 * lower-subreg.c (simple_move): Likewise.
1462 * optabs-libfuncs.c (init_optabs): Likewise.
1463 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1464 * tree.c (vector_type_mode): Likewise.
1465 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
1466 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
1467 * tree-vect-generic.c (expand_vector_parallel): Likewise.
1468 * tree-vect-stmts.c (vectorizable_load): Likewise.
1469 (vectorizable_store): Likewise.
1471 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1472 Alan Hayward <alan.hayward@arm.com>
1473 David Sherwood <david.sherwood@arm.com>
1475 * coretypes.h (pod_mode): New type.
1476 (scalar_int_mode_pod): New typedef.
1477 * machmode.h (pod_mode): New class.
1478 (int_n_data_t::m): Change type to scalar_int_mode_pod.
1479 * genmodes.c (emit_mode_int_n): Update accordingly.
1480 * lower-subreg.h (target_lower_subreg): Change type to
1481 scalar_int_mode_pod.
1482 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
1483 scalar_int_mode_pod.
1485 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1486 Alan Hayward <alan.hayward@arm.com>
1487 David Sherwood <david.sherwood@arm.com>
1489 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
1490 machine_mode to scalar_int_mode.
1491 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
1492 (rs6000_option_override_internal): Remove cast to int.
1493 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
1494 machine_mode to scalar_int_mode.
1495 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
1496 (rs6000_option_override_internal): Remove cast to int.
1497 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
1498 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
1500 * config/s390/s390.c (s390_expand_builtin): Likewise.
1501 * coretypes.h (scalar_int_mode): New type.
1502 (opt_scalar_int_mode): New typedef.
1503 * machmode.h (scalar_int_mode): New class.
1504 (scalar_int_mode::includes_p): New function.
1505 (byte_mode): Change type to scalar_int_mode.
1506 (word_mode): Likewise.
1507 (ptr_mode): Likewise.
1508 * emit-rtl.c (byte_mode): Likewise.
1509 (word_mode): Likewise.
1510 (ptr_mode): Likewise.
1511 (init_derived_machine_modes): Update accordingly.
1512 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
1513 and MODE_PARTIAL_INT.
1514 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
1515 opt_scalar_int_mode.
1517 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1518 Alan Hayward <alan.hayward@arm.com>
1519 David Sherwood <david.sherwood@arm.com>
1521 * target.def (libgcc_floating_mode_supported_p): Take a
1523 * doc/tm.texi: Regenerate.
1524 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
1526 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
1527 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
1530 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1531 Alan Hayward <alan.hayward@arm.com>
1532 David Sherwood <david.sherwood@arm.com>
1534 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
1535 * doc/tm.texi: Regenerate.
1536 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
1537 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
1538 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
1539 * targhooks.h (default_floatn_mode): Likewise.
1540 * targhooks.c (default_floatn_mode): Likewise.
1541 * tree.c (build_common_tree_nodes): Update accordingly.
1543 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1544 Alan Hayward <alan.hayward@arm.com>
1545 David Sherwood <david.sherwood@arm.com>
1547 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
1548 (mode_iterator::iterate_p): Likewise.
1549 (mode_iterator::get_wider): Likewise.
1550 * expr.c (init_expr_target): Use opt_scalar_float_mode.
1552 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1553 Alan Hayward <alan.hayward@arm.com>
1554 David Sherwood <david.sherwood@arm.com>
1556 * coretypes.h (opt_scalar_float_mode): New typedef.
1557 * machmode.h (float_mode_for_size): New function.
1558 * emit-rtl.c (double_mode): Delete.
1559 (init_emit_once): Use float_mode_for_size.
1560 * stor-layout.c (layout_type): Likewise.
1561 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
1563 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1564 Alan Hayward <alan.hayward@arm.com>
1565 David Sherwood <david.sherwood@arm.com>
1567 * output.h (assemble_real): Take a scalar_float_mode.
1568 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
1569 * config/arm/arm.md (consttable_4): Likewise.
1570 (consttable_8): Likewise.
1571 (consttable_16): Likewise.
1572 * config/mips/mips.md (consttable_float): Likewise.
1573 * config/s390/s390.c (s390_output_pool_entry): Likewise.
1574 * varasm.c (assemble_real): Take a scalar_float_mode.
1575 (output_constant_pool_2): Update accordingly.
1576 (output_constant): Likewise.
1578 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1579 Alan Hayward <alan.hayward@arm.com>
1580 David Sherwood <david.sherwood@arm.com>
1582 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
1583 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
1584 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
1585 (native_encode_real): Likewise.
1586 (native_interpret_real): Likewise.
1587 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
1588 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1590 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1591 Alan Hayward <alan.hayward@arm.com>
1592 David Sherwood <david.sherwood@arm.com>
1594 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
1595 <scalar_float_mode>. Simplify.
1596 (gen_extend_conv_libfunc): Likewise.
1598 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1599 Alan Hayward <alan.hayward@arm.com>
1600 David Sherwood <david.sherwood@arm.com>
1602 * coretypes.h (scalar_float_mode): New type.
1603 * machmode.h (mode_traits::from_int): Use machine_mode if
1604 USE_ENUM_MODES is defined.
1605 (is_a): New function.
1607 (dyn_cast): Likewise.
1608 (scalar_float_mode): New class.
1609 (scalar_float_mode::includes_p): New function.
1610 (is_float_mode): Likewise.
1611 * gdbhooks.py (MachineModePrinter): New class.
1612 (build_pretty_printer): Use it for scalar_float_mode.
1613 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
1614 (format_helper::format_helper): Turn into a template.
1615 * genmodes.c (get_mode_class): New function.
1616 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
1617 or machine_mode if none.
1618 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
1619 as_a <scalar_float_mode>.
1620 * dwarf2out.c (mem_loc_descriptor): Likewise.
1621 (insert_float): Likewise.
1622 (add_const_value_attribute): Likewise.
1623 * simplify-rtx.c (simplify_immed_subreg): Likewise.
1624 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
1625 (expand_unop): Update accordingly.
1626 (expand_abs_nojump): Likewise.
1627 (expand_copysign_absneg): Take a scalar_float_mode.
1628 (expand_copysign_bit): Likewise.
1629 (expand_copysign): Update accordingly.
1631 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1632 Alan Hayward <alan.hayward@arm.com>
1633 David Sherwood <david.sherwood@arm.com>
1635 * coretypes.h (opt_mode): New class.
1636 * machmode.h (opt_mode): Likewise.
1637 (opt_mode::else_void): New function.
1638 (opt_mode::require): Likewise.
1639 (opt_mode::exists): Likewise.
1640 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
1641 (GET_MODE_2XWIDER_MODE): Likewise.
1642 (mode_iterator::get_wider): Update accordingly.
1643 (mode_iterator::get_2xwider): Likewise.
1644 (mode_iterator::get_known_wider): Likewise, turning into a template.
1645 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
1646 forcing a wider mode to exist.
1647 * config/cr16/cr16.h (LONG_REG_P): Likewise.
1648 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
1649 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
1650 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
1651 * lower-subreg.c (init_lower_subreg): Likewise.
1652 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
1653 on the final iteration.
1654 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
1655 a wider mode exists before asking for a move pattern.
1656 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
1657 forcing a wider mode to exist.
1658 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
1659 returning false if no such mode exists.
1660 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
1661 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
1662 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
1663 Avoid checking for a MODE_INT if we already know the mode is not a
1665 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
1666 forcing a wider mode to exist.
1667 (expmed_mult_highpart_optab): Likewise.
1668 (expmed_mult_highpart): Likewise.
1669 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
1671 * lto-streamer-in.c (lto_input_mode_table): Likewise.
1672 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
1673 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
1674 * internal-fn.c (expand_mul_overflow): Update use of
1675 GET_MODE_2XWIDER_MODE.
1676 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1677 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
1678 GET_MODE_WIDER_MODE.
1679 (convert_plusminus_to_widen): Likewise.
1680 * tree-switch-conversion.c (array_value_type): Likewise.
1681 * var-tracking.c (emit_note_insn_var_location): Likewise.
1682 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1683 Return false inside rather than outside the loop if no wider mode
1685 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
1686 and GET_MODE_2XWIDER_MODE
1687 (can_compare_p): Use else_void.
1688 * gdbhooks.py (OptMachineModePrinter): New class.
1689 (build_pretty_printer): Use it for opt_mode.
1691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1692 Alan Hayward <alan.hayward@arm.com>
1693 David Sherwood <david.sherwood@arm.com>
1695 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
1696 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
1698 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1699 Alan Hayward <alan.hayward@arm.com>
1700 David Sherwood <david.sherwood@arm.com>
1702 * machmode.h (mode_traits): New structure.
1703 (get_narrowest_mode): New function.
1704 (mode_iterator::start): Likewise.
1705 (mode_iterator::iterate_p): Likewise.
1706 (mode_iterator::get_wider): Likewise.
1707 (mode_iterator::get_known_wider): Likewise.
1708 (mode_iterator::get_2xwider): Likewise.
1709 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
1710 (FOR_EACH_MODE): Likewise.
1711 (FOR_EACH_MODE_FROM): Likewise.
1712 (FOR_EACH_MODE_UNTIL): Likewise.
1713 (FOR_EACH_WIDER_MODE): Likewise.
1714 (FOR_EACH_2XWIDER_MODE): Likewise.
1715 * builtins.c (expand_builtin_strlen): Use new mode iterators.
1716 * combine.c (simplify_comparison): Likewise
1717 * config/i386/i386.c (type_natural_mode): Likewise.
1718 * cse.c (cse_insn): Likewise.
1719 * dse.c (find_shift_sequence): Likewise.
1720 * emit-rtl.c (init_derived_machine_modes): Likewise.
1721 (init_emit_once): Likewise.
1722 * explow.c (hard_function_value): Likewise.
1723 * expmed.c (extract_fixed_bit_field_1): Likewise.
1724 (extract_bit_field_1): Likewise.
1725 (expand_divmod): Likewise.
1726 (emit_store_flag_1): Likewise.
1727 * expr.c (init_expr_target): Likewise.
1728 (convert_move): Likewise.
1729 (alignment_for_piecewise_move): Likewise.
1730 (widest_int_mode_for_size): Likewise.
1731 (emit_block_move_via_movmem): Likewise.
1732 (copy_blkmode_to_reg): Likewise.
1733 (set_storage_via_setmem): Likewise.
1734 (compress_float_constant): Likewise.
1735 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1736 * optabs-query.c (get_best_extraction_insn): Likewise.
1737 * optabs.c (expand_binop): Likewise.
1738 (expand_twoval_unop): Likewise.
1739 (expand_twoval_binop): Likewise.
1740 (widen_leading): Likewise.
1741 (widen_bswap): Likewise.
1742 (expand_parity): Likewise.
1743 (expand_unop): Likewise.
1744 (prepare_cmp_insn): Likewise.
1745 (prepare_float_lib_cmp): Likewise.
1746 (expand_float): Likewise.
1747 (expand_fix): Likewise.
1748 (expand_sfix_optab): Likewise.
1749 * postreload.c (move2add_use_add2_insn): Likewise.
1750 * reg-stack.c (reg_to_stack): Likewise.
1751 * reginfo.c (choose_hard_reg_mode): Likewise.
1752 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
1753 * stor-layout.c (mode_for_size): Likewise.
1754 (smallest_mode_for_size): Likewise.
1755 (mode_for_vector): Likewise.
1756 (finish_bitfield_representative): Likewise.
1757 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
1758 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
1759 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1760 * var-tracking.c (prepare_call_arguments): Likewise.
1762 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1763 Alan Hayward <alan.hayward@arm.com>
1764 David Sherwood <david.sherwood@arm.com>
1766 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
1767 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
1768 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
1769 * machmode.h (mode_size): Move earlier in file.
1770 (mode_precision): Likewise.
1771 (mode_inner): Likewise.
1772 (mode_nunits): Likewise.
1773 (mode_unit_size): Likewise.
1774 (unit_unit_precision): Likewise.
1775 (mode_wider): Likewise.
1776 (mode_2xwider): Likewise.
1777 (machine_mode): New class.
1778 (mode_to_bytes): New function.
1779 (mode_to_bits): Likewise.
1780 (mode_to_precision): Likewise.
1781 (mode_to_inner): Likewise.
1782 (mode_to_unit_size): Likewise.
1783 (mode_to_unit_precision): Likewise.
1784 (mode_to_nunits): Likewise.
1785 (GET_MODE_SIZE): Use mode_to_bytes.
1786 (GET_MODE_BITSIZE): Use mode_to_bits.
1787 (GET_MODE_PRECISION): Use mode_to_precision.
1788 (GET_MODE_INNER): Use mode_to_inner.
1789 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
1790 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
1791 (GET_MODE_NUNITS): Use mode_to_nunits.
1792 * system.h (ALWAYS_INLINE): New macro.
1793 * config/powerpcspe/powerpcspe-c.c
1794 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
1795 int for arg1_mode and arg2_mode.
1797 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1798 Alan Hayward <alan.hayward@arm.com>
1799 David Sherwood <david.sherwood@arm.com>
1801 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
1802 Prefix mode names with E_ in case statements.
1803 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1804 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
1805 (aarch64_split_simd_move): Likewise.
1806 (aarch64_gen_storewb_pair): Likewise.
1807 (aarch64_gen_loadwb_pair): Likewise.
1808 (aarch64_gen_store_pair): Likewise.
1809 (aarch64_gen_load_pair): Likewise.
1810 (aarch64_get_condition_code_1): Likewise.
1811 (aarch64_constant_pool_reload_icode): Likewise.
1812 (get_rsqrte_type): Likewise.
1813 (get_rsqrts_type): Likewise.
1814 (get_recpe_type): Likewise.
1815 (get_recps_type): Likewise.
1816 (aarch64_gimplify_va_arg_expr): Likewise.
1817 (aarch64_simd_container_mode): Likewise.
1818 (aarch64_emit_load_exclusive): Likewise.
1819 (aarch64_emit_store_exclusive): Likewise.
1820 (aarch64_expand_compare_and_swap): Likewise.
1821 (aarch64_gen_atomic_cas): Likewise.
1822 (aarch64_emit_bic): Likewise.
1823 (aarch64_emit_atomic_swap): Likewise.
1824 (aarch64_emit_atomic_load_op): Likewise.
1825 (aarch64_evpc_trn): Likewise.
1826 (aarch64_evpc_uzp): Likewise.
1827 (aarch64_evpc_zip): Likewise.
1828 (aarch64_evpc_ext): Likewise.
1829 (aarch64_evpc_rev): Likewise.
1830 (aarch64_evpc_dup): Likewise.
1831 (aarch64_gen_ccmp_first): Likewise.
1832 (aarch64_gen_ccmp_next): Likewise.
1833 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
1834 (alpha_emit_xfloating_libcall): Likewise.
1835 (emit_insxl): Likewise.
1836 (alpha_arg_type): Likewise.
1837 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
1838 (arc_preferred_simd_mode): Likewise.
1839 (arc_secondary_reload): Likewise.
1840 (get_arc_condition_code): Likewise.
1841 (arc_print_operand): Likewise.
1842 (arc_legitimate_constant_p): Likewise.
1843 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1844 * config/arc/arc.md (casesi_load): Likewise.
1845 (casesi_compact_jump): Likewise.
1846 * config/arc/predicates.md (proper_comparison_operator): Likewise.
1847 (cc_use_register): Likewise.
1848 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1849 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
1850 (arm_init_iwmmxt_builtins): Likewise.
1851 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
1852 (neon_expand_vector_init): Likewise.
1853 (arm_attr_length_move_neon): Likewise.
1854 (maybe_get_arm_condition_code): Likewise.
1855 (arm_emit_vector_const): Likewise.
1856 (arm_preferred_simd_mode): Likewise.
1857 (arm_output_iwmmxt_tinsr): Likewise.
1858 (thumb1_output_casesi): Likewise.
1859 (thumb2_output_casesi): Likewise.
1860 (arm_emit_load_exclusive): Likewise.
1861 (arm_emit_store_exclusive): Likewise.
1862 (arm_expand_compare_and_swap): Likewise.
1863 (arm_evpc_neon_vuzp): Likewise.
1864 (arm_evpc_neon_vzip): Likewise.
1865 (arm_evpc_neon_vrev): Likewise.
1866 (arm_evpc_neon_vtrn): Likewise.
1867 (arm_evpc_neon_vext): Likewise.
1868 (arm_validize_comparison): Likewise.
1869 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
1870 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
1871 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
1872 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
1873 (c6x_preferred_simd_mode): Likewise.
1874 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
1875 (epiphany_rtx_costs): Likewise.
1876 * config/epiphany/predicates.md (proper_comparison_operator):
1878 * config/frv/frv.c (condexec_memory_operand): Likewise.
1879 (frv_emit_move): Likewise.
1880 (output_move_single): Likewise.
1881 (output_condmove_single): Likewise.
1882 (frv_hard_regno_mode_ok): Likewise.
1883 (frv_matching_accg_mode): Likewise.
1884 * config/h8300/h8300.c (split_adds_subs): Likewise.
1885 (h8300_rtx_costs): Likewise.
1886 (h8300_print_operand): Likewise.
1887 (compute_mov_length): Likewise.
1888 (output_logical_op): Likewise.
1889 (compute_logical_op_length): Likewise.
1890 (compute_logical_op_cc): Likewise.
1891 (h8300_shift_needs_scratch_p): Likewise.
1892 (output_a_shift): Likewise.
1893 (compute_a_shift_length): Likewise.
1894 (compute_a_shift_cc): Likewise.
1895 (expand_a_rotate): Likewise.
1896 (output_a_rotate): Likewise.
1897 * config/i386/i386.c (classify_argument): Likewise.
1898 (function_arg_advance_32): Likewise.
1899 (function_arg_32): Likewise.
1900 (function_arg_64): Likewise.
1901 (function_value_64): Likewise.
1902 (ix86_gimplify_va_arg): Likewise.
1903 (ix86_legitimate_constant_p): Likewise.
1904 (put_condition_code): Likewise.
1905 (split_double_mode): Likewise.
1906 (ix86_avx256_split_vector_move_misalign): Likewise.
1907 (ix86_expand_vector_logical_operator): Likewise.
1908 (ix86_split_idivmod): Likewise.
1909 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
1910 (ix86_build_const_vector): Likewise.
1911 (ix86_build_signbit_mask): Likewise.
1912 (ix86_match_ccmode): Likewise.
1913 (ix86_cc_modes_compatible): Likewise.
1914 (ix86_expand_branch): Likewise.
1915 (ix86_expand_sse_cmp): Likewise.
1916 (ix86_expand_sse_movcc): Likewise.
1917 (ix86_expand_int_sse_cmp): Likewise.
1918 (ix86_expand_vec_perm_vpermi2): Likewise.
1919 (ix86_expand_vec_perm): Likewise.
1920 (ix86_expand_sse_unpack): Likewise.
1921 (ix86_expand_int_addcc): Likewise.
1922 (ix86_split_to_parts): Likewise.
1923 (ix86_vectorize_builtin_gather): Likewise.
1924 (ix86_vectorize_builtin_scatter): Likewise.
1925 (avx_vpermilp_parallel): Likewise.
1926 (inline_memory_move_cost): Likewise.
1927 (ix86_tieable_integer_mode_p): Likewise.
1928 (x86_maybe_negate_const_int): Likewise.
1929 (ix86_expand_vector_init_duplicate): Likewise.
1930 (ix86_expand_vector_init_one_nonzero): Likewise.
1931 (ix86_expand_vector_init_one_var): Likewise.
1932 (ix86_expand_vector_init_concat): Likewise.
1933 (ix86_expand_vector_init_interleave): Likewise.
1934 (ix86_expand_vector_init_general): Likewise.
1935 (ix86_expand_vector_set): Likewise.
1936 (ix86_expand_vector_extract): Likewise.
1937 (emit_reduc_half): Likewise.
1938 (ix86_emit_i387_round): Likewise.
1939 (ix86_mangle_type): Likewise.
1940 (ix86_expand_round_sse4): Likewise.
1941 (expand_vec_perm_blend): Likewise.
1942 (canonicalize_vector_int_perm): Likewise.
1943 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
1944 (expand_vec_perm_1): Likewise.
1945 (expand_vec_perm_interleave3): Likewise.
1946 (expand_vec_perm_even_odd_pack): Likewise.
1947 (expand_vec_perm_even_odd_1): Likewise.
1948 (expand_vec_perm_broadcast_1): Likewise.
1949 (ix86_vectorize_vec_perm_const_ok): Likewise.
1950 (ix86_expand_vecop_qihi): Likewise.
1951 (ix86_expand_mul_widen_hilo): Likewise.
1952 (ix86_expand_sse2_abs): Likewise.
1953 (ix86_expand_pextr): Likewise.
1954 (ix86_expand_pinsr): Likewise.
1955 (ix86_preferred_simd_mode): Likewise.
1956 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
1957 * config/i386/sse.md (*andnot<mode>3): Likewise.
1958 (<mask_codefor><code><mode>3<mask_name>): Likewise.
1959 (*<code><mode>3): Likewise.
1960 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
1961 (ia64_expand_atomic_op): Likewise.
1962 (ia64_arg_type): Likewise.
1963 (ia64_mode_to_int): Likewise.
1964 (ia64_scalar_mode_supported_p): Likewise.
1965 (ia64_vector_mode_supported_p): Likewise.
1966 (expand_vec_perm_broadcast): Likewise.
1967 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
1968 (iq2000_function_arg_advance): Likewise.
1969 (iq2000_function_arg): Likewise.
1970 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
1971 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
1972 (m68k_libcall_value): Likewise.
1973 (m68k_function_value): Likewise.
1974 (sched_attr_op_type): Likewise.
1975 * config/mcore/mcore.c (mcore_output_move): Likewise.
1976 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
1978 (microblaze_function_arg): Likewise.
1979 * config/mips/mips.c (mips16_build_call_stub): Likewise.
1980 (mips_print_operand): Likewise.
1981 (mips_mode_ok_for_mov_fmt_p): Likewise.
1982 (mips_vector_mode_supported_p): Likewise.
1983 (mips_preferred_simd_mode): Likewise.
1984 (mips_expand_vpc_loongson_even_odd): Likewise.
1985 (mips_expand_vec_unpack): Likewise.
1986 (mips_expand_vi_broadcast): Likewise.
1987 (mips_expand_vector_init): Likewise.
1988 (mips_expand_vec_reduc): Likewise.
1989 (mips_expand_msa_cmp): Likewise.
1990 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
1991 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
1992 (cc_flags_for_mode): Likewise.
1993 * config/msp430/msp430.c (msp430_print_operand): Likewise.
1994 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
1995 (nds32_output_casesi_pc_relative): Likewise.
1996 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1997 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
1998 (nvptx_gen_unpack): Likewise.
1999 (nvptx_gen_pack): Likewise.
2000 (nvptx_gen_shuffle): Likewise.
2001 (nvptx_gen_wcast): Likewise.
2002 (nvptx_preferred_simd_mode): Likewise.
2003 * config/pa/pa.c (pa_secondary_reload): Likewise.
2004 * config/pa/predicates.md (base14_operand): Likewise.
2005 * config/powerpcspe/powerpcspe-c.c
2006 (altivec_resolve_overloaded_builtin): Likewise.
2007 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
2009 (rs6000_preferred_simd_mode): Likewise.
2010 (output_vec_const_move): Likewise.
2011 (rs6000_expand_vector_extract): Likewise.
2012 (rs6000_split_vec_extract_var): Likewise.
2013 (reg_offset_addressing_ok_p): Likewise.
2014 (rs6000_legitimate_offset_address_p): Likewise.
2015 (rs6000_legitimize_address): Likewise.
2016 (rs6000_emit_set_const): Likewise.
2017 (rs6000_const_vec): Likewise.
2018 (rs6000_emit_move): Likewise.
2019 (spe_build_register_parallel): Likewise.
2020 (rs6000_darwin64_record_arg_recurse): Likewise.
2021 (swap_selector_for_mode): Likewise.
2022 (spe_init_builtins): Likewise.
2023 (paired_init_builtins): Likewise.
2024 (altivec_init_builtins): Likewise.
2025 (do_load_for_compare): Likewise.
2026 (rs6000_generate_compare): Likewise.
2027 (rs6000_expand_float128_convert): Likewise.
2028 (emit_load_locked): Likewise.
2029 (emit_store_conditional): Likewise.
2030 (rs6000_output_function_epilogue): Likewise.
2031 (rs6000_handle_altivec_attribute): Likewise.
2032 (rs6000_function_value): Likewise.
2033 (emit_fusion_gpr_load): Likewise.
2034 (emit_fusion_p9_load): Likewise.
2035 (emit_fusion_p9_store): Likewise.
2036 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
2037 (fusion_gpr_mem_load): Likewise.
2038 (fusion_addis_mem_combo_load): Likewise.
2039 (fusion_addis_mem_combo_store): Likewise.
2040 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
2041 (fusion_gpr_mem_load): Likewise.
2042 (fusion_addis_mem_combo_load): Likewise.
2043 (fusion_addis_mem_combo_store): Likewise.
2044 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2046 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
2047 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
2048 (rs6000_preferred_simd_mode): Likewise.
2049 (output_vec_const_move): Likewise.
2050 (rs6000_expand_vector_extract): Likewise.
2051 (rs6000_split_vec_extract_var): Likewise.
2052 (reg_offset_addressing_ok_p): Likewise.
2053 (rs6000_legitimate_offset_address_p): Likewise.
2054 (rs6000_legitimize_address): Likewise.
2055 (rs6000_emit_set_const): Likewise.
2056 (rs6000_const_vec): Likewise.
2057 (rs6000_emit_move): Likewise.
2058 (rs6000_darwin64_record_arg_recurse): Likewise.
2059 (swap_selector_for_mode): Likewise.
2060 (paired_init_builtins): Likewise.
2061 (altivec_init_builtins): Likewise.
2062 (rs6000_expand_float128_convert): Likewise.
2063 (emit_load_locked): Likewise.
2064 (emit_store_conditional): Likewise.
2065 (rs6000_output_function_epilogue): Likewise.
2066 (rs6000_handle_altivec_attribute): Likewise.
2067 (rs6000_function_value): Likewise.
2068 (emit_fusion_gpr_load): Likewise.
2069 (emit_fusion_p9_load): Likewise.
2070 (emit_fusion_p9_store): Likewise.
2071 * config/rx/rx.c (rx_gen_move_template): Likewise.
2072 (flags_from_mode): Likewise.
2073 * config/s390/predicates.md (s390_alc_comparison): Likewise.
2074 (s390_slb_comparison): Likewise.
2075 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
2076 (s390_vector_mode_supported_p): Likewise.
2077 (s390_cc_modes_compatible): Likewise.
2078 (s390_match_ccmode_set): Likewise.
2079 (s390_canonicalize_comparison): Likewise.
2080 (s390_emit_compare_and_swap): Likewise.
2081 (s390_branch_condition_mask): Likewise.
2082 (s390_rtx_costs): Likewise.
2083 (s390_secondary_reload): Likewise.
2084 (__SECONDARY_RELOAD_CASE): Likewise.
2085 (s390_expand_cs): Likewise.
2086 (s390_preferred_simd_mode): Likewise.
2087 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
2088 * config/sh/sh.c (sh_print_operand): Likewise.
2089 (dump_table): Likewise.
2090 (sh_secondary_reload): Likewise.
2091 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2092 * config/sh/sh.md (casesi_worker_1): Likewise.
2093 (casesi_worker_2): Likewise.
2094 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
2095 (fcc_comparison_operator): Likewise.
2096 * config/sparc/sparc.c (sparc_expand_move): Likewise.
2097 (emit_soft_tfmode_cvt): Likewise.
2098 (sparc_preferred_simd_mode): Likewise.
2099 (output_cbranch): Likewise.
2100 (sparc_print_operand): Likewise.
2101 (sparc_expand_vec_perm_bmask): Likewise.
2102 (vector_init_bshuffle): Likewise.
2103 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
2104 (spu_vector_mode_supported_p): Likewise.
2105 (spu_expand_insv): Likewise.
2106 (spu_emit_branch_or_set): Likewise.
2107 (spu_handle_vector_attribute): Likewise.
2108 (spu_builtin_splats): Likewise.
2109 (spu_builtin_extract): Likewise.
2110 (spu_builtin_promote): Likewise.
2111 (spu_expand_sign_extend): Likewise.
2112 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
2113 (tilegx_simd_int): Likewise.
2114 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
2115 (tilepro_simd_int): Likewise.
2116 * config/v850/v850.c (const_double_split): Likewise.
2117 (v850_print_operand): Likewise.
2118 (ep_memory_offset): Likewise.
2119 * config/vax/vax.c (vax_rtx_costs): Likewise.
2120 (vax_output_int_move): Likewise.
2121 (vax_output_int_add): Likewise.
2122 (vax_output_int_subtract): Likewise.
2123 * config/visium/predicates.md (visium_branch_operator): Likewise.
2124 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
2125 (visium_print_operand_address): Likewise.
2126 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2127 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
2128 (xtensa_expand_conditional_branch): Likewise.
2129 (xtensa_copy_incoming_a7): Likewise.
2130 (xtensa_output_literal): Likewise.
2131 * dfp.c (decimal_real_maxval): Likewise.
2132 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
2134 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2135 Alan Hayward <alan.hayward@arm.com>
2136 David Sherwood <david.sherwood@arm.com>
2138 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
2139 (mode_nunits_inline): Likewise.
2140 (mode_inner_inline): Likewise.
2141 (mode_unit_size_inline): Likewise.
2142 (mode_unit_precision_inline): Likewise.
2143 (emit_insn_modes_h): Likewise. Also emit a #define of the
2145 (emit_mode_wider): Add an E_ prefix to mode names.
2146 (emit_mode_complex): Likewise.
2147 (emit_mode_inner): Likewise.
2148 (emit_mode_adjustments): Likewise.
2149 (emit_mode_int_n): Likewise.
2150 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
2151 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
2152 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
2153 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
2154 (CRC32_BUILTIN, ENTRY): Likewise.
2155 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
2156 (aarch64_pop_regs): Likewise.
2157 (aarch64_process_components): Likewise.
2158 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
2159 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
2160 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
2161 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
2162 * config/arm/arm.c (arm_init_libfuncs): Likewise.
2163 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
2165 * config/i386/i386-builtin.def (pcmpestr): Likewise.
2166 (pcmpistr): Likewise.
2167 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
2168 * config/mmix/mmix.c (mmix_output_condition): Likewise.
2169 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
2171 * config/rl78/rl78.c (mduc_regs): Likewise.
2172 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2173 (htm_expand_builtin): Likewise.
2174 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
2175 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
2176 * config/xtensa/xtensa.c (print_operand): Likewise.
2177 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
2178 (NUM_MODE_VECTOR_INT): Likewise.
2179 * genoutput.c (null_operand): Likewise.
2180 (output_operand_data): Likewise.
2181 * genrecog.c (print_parameter_value): Likewise.
2182 * lra.c (debug_operand_data): Likewise.
2184 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2186 * dwarf2out.c (scompare_loc_descriptor_wide)
2187 (scompare_loc_descriptor_narrow): New functions, split out from...
2188 (scompare_loc_descriptor): ...here.
2189 * expmed.c (emit_store_flag_int): New function, split out from...
2190 (emit_store_flag): ...here.
2192 2017-08-30 Richard Biener <rguenther@suse.de>
2194 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
2195 (dwarf2out_early_finish): Move setting of AT_pubnames from
2196 early debug output to early finish.
2198 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
2200 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
2201 and -mdata-region to the assembler.
2203 2017-08-30 Richard Biener <rguenther@suse.de>
2205 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
2207 (gen_subprogram_die): Add DW_AT_object_pointer only early.
2208 (dwarf2out_early_global_decl): Only generate a DIE for the
2209 abstract origin if it doesn't already exist or is a declaration DIE.
2210 (resolve_addr): Do not add the linkage name twice when
2211 generating a stub DIE for the DW_TAG_GNU_call_site target.
2213 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2215 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2216 Use machine_mode rather than int for arg1_mode.
2218 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
2221 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
2222 that the second argument of the built-in functions to unpack
2223 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
2224 switch statement instead a lot of if statements.
2225 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
2226 Allow 64-bit values to be in Altivec registers as well as
2227 traditional floating point registers.
2228 (pack<mode>, FMOVE128_VSX iterator): Likewise.
2230 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
2232 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
2233 MAX_REGS_PER_ADDRESS == 1.
2235 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
2237 * config/i386/i386.opt (flag_fentry): Do not init to -1.
2238 * config/i386/i386.c (ix86_option_override_internal): Simplify
2239 setting of opts->x_flag_entry.
2241 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2242 Jakub Jelinek <jakub@redhat.com>
2243 Richard Biener <rguenther@suse.de>
2245 PR tree-optimization/81503
2246 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
2247 folded constant fits in the target type; reorder tests for clarity.
2249 2017-08-29 Martin Liska <mliska@suse.cz>
2251 * passes.def: Include pass_lower_switch.
2252 * stmt.c (dump_case_nodes): Remove and move to
2253 tree-switch-conversion.
2254 (case_values_threshold): Likewise.
2255 (expand_switch_as_decision_tree_p): Likewise.
2256 (emit_case_decision_tree): Likewise.
2257 (expand_case): Likewise.
2258 (balance_case_nodes): Likewise.
2259 (node_has_low_bound): Likewise.
2260 (node_has_high_bound): Likewise.
2261 (node_is_bounded): Likewise.
2262 (emit_case_nodes): Likewise.
2263 (struct simple_case_node): New struct.
2264 (add_case_node): Remove.
2265 (emit_case_dispatch_table): Use vector instead of case_list.
2266 (reset_out_edges_aux): Remove.
2267 (compute_cases_per_edge): Likewise.
2268 (expand_case): Build list of simple_case_node.
2269 (expand_sjlj_dispatch_table): Use it.
2270 * tree-switch-conversion.c (struct case_node): Moved from
2271 stmt.c and adjusted.
2272 (emit_case_nodes): Likewise.
2273 (node_has_low_bound): Likewise.
2274 (node_has_high_bound): Likewise.
2275 (node_is_bounded): Likewise.
2276 (case_values_threshold): Likewise.
2277 (reset_out_edges_aux): Likewise.
2278 (compute_cases_per_edge): Likewise.
2279 (add_case_node): Likewise.
2280 (dump_case_nodes): Likewise.
2281 (balance_case_nodes): Likewise.
2282 (expand_switch_as_decision_tree_p): Likewise.
2283 (emit_jump): Likewise.
2284 (emit_case_decision_tree): Likewise.
2285 (try_switch_expansion): Likewise.
2286 (do_jump_if_equal): Likewise.
2287 (emit_cmp_and_jump_insns): Likewise.
2288 (fix_phi_operands_for_edge): New function.
2289 (record_phi_operand_mapping): Likewise.
2290 (class pass_lower_switch): New pass.
2291 (pass_lower_switch::execute): New function.
2292 (make_pass_lower_switch): Likewise.
2293 (conditional_probability):
2294 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
2295 * tree-pass.h: Add make_pass_lower_switch.
2297 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
2300 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
2303 2017-08-29 Richard Biener <rguenther@suse.de>
2305 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
2306 we do not add a DW_AT_inline attribute twice.
2307 (gen_subprogram_die): Remove code setting DW_AT_inline on
2308 DECL_ABSTRACT_P nodes.
2310 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
2312 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
2313 calls to internal functions.
2314 (gimplify_modify_expr): Likewise.
2315 * tree-call-cdce.c (use_internal_fn): Likewise.
2316 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
2317 (convert_to_divmod): Set the nothrow flag.
2318 * tree-if-conv.c (predicate_mem_writes): Likewise.
2319 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
2320 (vectorizable_call): Likewise.
2321 (vectorizable_store): Likewise.
2322 (vectorizable_load): Likewise.
2323 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
2324 (vect_recog_mask_conversion_pattern): Likewise.
2326 2017-08-29 Martin Liska <mliska@suse.cz>
2329 * gcc.c (driver_handle_option): Add new argument.
2330 * opts-common.c (handle_option): Pass
2331 target_option_override_hook.
2332 * opts-global.c (lang_handle_option): Add new option.
2333 (set_default_handlers): Add new argument.
2334 (decode_options): Likewise.
2335 * opts.c (target_handle_option): Likewise.
2336 (common_handle_option): Call target_option_override_hook.
2337 * opts.h (struct cl_option_handler_func): Add hook for
2338 target option override.
2339 (struct cl_option_handlers): Likewise.
2340 (set_default_handlers): Add new argument.
2341 (decode_options): Likewise.
2342 (common_handle_option): Likewise.
2343 (target_handle_option): Likewise.
2344 * toplev.c (toplev::main): Pass targetm.target_option.override
2347 2017-08-29 Richard Biener <rguenther@suse.de>
2348 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2350 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
2351 life to the active subtree.
2353 2017-08-28 Jeff Law <law@redhat.com>
2355 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
2356 derive_equivalences.
2357 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
2359 (edge_info::derive_equivalences): New private member function
2361 * tree-ssa-dom.c (class edge_info): Changed from a struct
2362 to a class. Add ctor/dtor, methods and data members.
2363 (edge_info::edge_info): Renamed from allocate_edge_info.
2364 Initialize additional members.
2365 (edge_info::~edge_info): New.
2366 (free_dom_edge_info): Delete the edge info.
2367 (record_edge_info): Use new class & associated member functions.
2368 Tighten forms for testing for edge equivalences.
2369 (record_temporary_equivalences): Iterate over the simple
2370 equivalences rather than assuming there's only one per edge.
2371 (cprop_into_successor_phis): Iterate over the simple
2372 equivalences rather than assuming there's only one per edge.
2373 (optimize_stmt): Use operand_equal_p rather than pointer
2374 equality for mini-DSE code.
2376 2017-08-28 Nathan Sidwell <nathan@acm.org>
2378 * gcc.c (execute): Fold SIGPIPE handling into switch
2379 statement. Adjust internal error message.
2381 2017-08-28 Richard Biener <rguenther@suse.de>
2384 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
2385 Do nothing for removed DIEs.
2387 2017-08-28 Richard Biener <rguenther@suse.de>
2389 PR tree-optimization/81977
2390 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
2393 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
2396 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
2397 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
2398 using targetm.gen_mem_thread_fence.
2400 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
2403 * config/i386/i386.md (*<btsc><mode>): Change operand 2
2404 predicate to register_operand. Reorder operands.
2405 (*btr<mode>): Ditto.
2406 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
2407 (*btr<mode>_mask): Ditto.
2409 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
2411 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
2412 * config/rs6000/xmmintrin.h: New file.
2413 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
2415 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2418 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
2419 parameter and_insn and return it.
2420 (recombine_lvx_pattern): Insert a copy to ensure availability of
2421 the base register of the copied masking operation at the point of
2422 the instruction replacement.
2423 (recombine_stvx_pattern): Likewise.
2425 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
2427 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
2428 undocumented switches.
2429 (-mpower9-dform-vector): Likewise.
2430 (-mpower9-dform): Likewise.
2431 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
2432 comments to delete references to -mpower9-dform* switches.
2433 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
2434 Delete reference to -mpower9-dform* switches, test for
2435 -mpower9-vector instead.
2436 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
2437 (OTHER_P9_VECTOR_MASKS): Likewise.
2438 (POWERPC_MASKS): Likewise.
2439 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
2440 tests against -mpower9-dform* to -mpower9-vector. Delete code
2441 that checked for -mpower9-dform* consistancy with other options.
2442 Add test for -mpower9-misc to enable other power9 switches.
2443 (rs6000_init_hard_regno_mode_ok): Likewise.
2444 (rs6000_option_override_internal): Likewise.
2445 (rs6000_emit_prologue): Likewise.
2446 (rs6000_emit_epilogue): Likewise.
2447 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
2448 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
2449 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
2451 (emit_fusion_p9_store): Likewise.
2452 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
2453 resetting these macros if the assembler does not support ISA 3.0
2455 (TARGET_P9_DFORM_VECTOR): Likewise.
2456 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
2457 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
2459 2017-08-25 Alan Modra <amodra@gmail.com>
2461 PR rtl-optimization/81747
2462 * cse.c (cse_extended_basic_block): Don't attempt to record
2463 equivalences for degenerate conditional branches that branch
2464 to their fall-through.
2466 2017-08-24 Martin Sebor <msebor@redhat.com>
2469 * gimple-fold.c (size_must_be_zero_p): New function.
2470 (gimple_fold_builtin_memory_op): Call it.
2472 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
2474 * config/rs6000/mm_malloc.h: New file.
2476 2017-08-24 Bin Cheng <bin.cheng@arm.com>
2478 PR tree-optimization/81913
2479 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
2480 analysis when either IVs in condition can wrap.
2482 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
2484 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
2485 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
2487 2017-08-24 Richard Biener <rguenther@suse.de>
2490 * targhooks.c (default_target_can_inline_p): Properly
2491 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
2492 is present and always compare.
2493 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
2494 infer -mfpmath=sse from TARGET_SSE_P.
2495 (ix86_can_inline_p): Properly use target_option_default_node when
2496 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
2498 2017-08-24 Richard Biener <rguenther@suse.de>
2501 * dwarf2out.c (output_die): Handle flag_generate_offload like
2503 (output_comp_unit): Likewise.
2504 (gen_array_type_die): Likewise.
2505 (dwarf2out_early_finish): Likewise.
2506 (note_variable_value_in_expr): Likewise.
2507 (dwarf2out_finish): Likewise. Adjust assert.
2508 * cgraphunit.c (symbol_table::compile): Move setting of
2509 flag_generate_offload earlier ...
2510 (symbol_table::finalize_compilation_unit): ... here, before
2511 early debug finalization.
2513 2017-08-24 Richard Biener <rguenther@suse.de>
2515 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
2516 and ipa-fnsummary.h.
2517 (ix86_can_inline_p): When ix86_fpmath flags do not match
2518 check whether the callee uses FP math at all.
2520 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
2523 * tree-ssanames.c (get_nonzero_bits): Use element_precision
2524 instead of TYPE_PRECISION.
2526 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
2527 Alan Hayward <alan.hayward@arm.com>
2528 David Sherwood <david.sherwood@arm.com>
2530 * combine.c (make_extraction): Use subreg_offset_from_lsb.
2532 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
2534 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
2536 (ix86_frame::stack_realign_allocate): New field.
2537 (struct machine_frame_state): Modify comments.
2538 (machine_frame_state::sp_realigned_fp_end): New field.
2539 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
2541 (sp_valid_at): Add assertion to assure no attempt to access invalid
2542 offset of a realigned stack.
2543 (fp_valid_at): Likewise.
2544 (choose_baseaddr): Modify comments.
2545 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
2546 ix86_expand_prologue.
2547 (ix86_expand_prologue): Modify stack realignment and allocation.
2548 (ix86_expand_epilogue): Modify comments.
2549 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
2550 avx2_runtime, avx512f, and avx512f_runtime.
2552 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
2554 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
2555 (mstackrealign): Do not init to -1.
2556 * config/i386/i386.c (ix86_option_override_internal):
2557 Check opts_set, not opts when setting default value of
2558 opts->x_ix86_force_align_arg_pointer.
2560 2017-08-23 Richard Biener <rguenther@suse.de>
2562 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
2563 lang_hooks.decl_printable_name.
2564 * print-rtl-function.c (print_rtx_function): Likewise.
2565 * tree-pretty-print.c (dump_function_header): Likewise.
2567 2017-08-23 Richard Biener <rguenther@suse.de>
2570 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
2571 -g0 at compile-time.
2573 2017-08-23 Tamar Christina <tamar.christina@arm.com>
2576 * doc/sourcebuild.texi (Other hardware attributes):
2579 2017-08-23 Tamar Christina <tamar.christina@arm.com>
2582 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
2583 Fix single-use check.
2585 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
2587 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
2589 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
2591 * doc/install.texi: Modify to add more details on running selected
2594 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
2596 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
2597 is combined with -mabi=ms.
2598 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
2601 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2603 PR tree-optimization/81488
2604 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
2605 and cached_basis fields.
2606 (MAX_SPREAD): New constant.
2607 (alloc_cand_and_find_basis): Initialize new fields.
2608 (clear_visited): New function.
2609 (create_phi_basis_1): Rename from create_phi_basis, set visited
2610 and cached_basis fields.
2611 (create_phi_basis): New wrapper function.
2612 (phi_add_costs_1): Rename from phi_add_costs, add spread
2613 parameter, set visited field, short-circuit when limits reached.
2614 (phi_add_costs): New wrapper function.
2615 (record_phi_increments_1): Rename from record_phi_increments, set
2617 (record_phi_increments): New wrapper function.
2618 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
2619 (phi_incr_cost): New wrapper function.
2620 (all_phi_incrs_profitable_1): Rename from
2621 all_phi_incrs_profitable, set visited field.
2622 (all_phi_incrs_profitable): New wrapper function.
2624 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
2625 Alan Hayward <alan.hayward@arm.com>
2626 David Sherwood <david.sherwood@arm.com>
2628 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
2629 that takes the outer and inner modes.
2630 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
2631 comparison as the canonical test for a paradoxical subreg.
2632 * combine.c (simplify_set): Use paradoxical_subreg_p.
2633 (make_extraction): Likewise.
2634 (force_to_mode): Likewise.
2635 (rtx_equal_for_field_assignment_p): Likewise.
2636 (gen_lowpart_for_combine): Likewise.
2637 (simplify_comparison): Likewise.
2638 * cse.c (equiv_constant): Likewise.
2639 * expmed.c (store_bit_field_1): Likewise.
2640 * final.c (alter_subreg): Likewise.
2641 * fwprop.c (propagate_rtx): Likewise.
2642 (forward_propagate_subreg): Likewise.
2643 * ira-conflicts.c (ira_build_conflicts): Likewise.
2644 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
2645 * lra-constraints.c (curr_insn_transform): Likewise.
2646 (split_reg): Likewise.
2647 * lra-eliminations.c (move_plus_up): Likewise.
2648 (lra_eliminate_regs_1): Likewise.
2649 * recog.c (general_operand): Likewise.
2650 * ree.c (combine_reaching_defs): Likewise.
2651 * reload.c (push_reload): Likewise.
2652 (find_reloads): Likewise.
2653 * reload1.c (elimination_effects): Likewise.
2654 (compute_reload_subreg_offset): Likewise.
2655 (choose_reload_regs): Likewise.
2656 * rtlanal.c (subreg_lsb_1): Likewise.
2657 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2658 (simplify_subreg): Likewise.
2659 * var-tracking.c (track_loc_p): Likewise.
2660 * emit-rtl.c (byte_lowpart_offset): Likewise.
2661 (paradoxical_subreg_p): Delete out-of-line definition.
2663 2017-08-22 Jeff Law <law@redhat.com>
2665 PR tree-optimization/81741
2666 PR tree-optimization/71947
2667 * tree-ssa-dom.c: Include tree-inline.h.
2668 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
2669 equivalences if one is more expensive to compute than the other.
2670 * tree-ssa-scopedtables.h (class const_or_copies): Make
2671 record_const_or_copy_raw method private.
2672 (class avail_exprs_stack): New method simplify_binary_operation.
2673 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
2674 avail_exprs_stack::simplify_binary_operation as needed.
2675 (avail_exprs_stack::simplify_binary_operation): New function.
2677 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
2679 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
2680 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
2681 (DOT_SYMBOLS): Likewise.
2682 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
2683 (RELOCATABLE_NEEDS_FIXUP): Likewise.
2684 (RS6000_ABI_NAME): Likewise.
2685 (TARGET_CMODEL): Likewise.
2686 (TOC_SECTION_ASM_OP): Likewise.
2687 (SET_CMODEL): New macro.
2688 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
2690 2017-08-22 Richard Biener <rguenther@suse.de>
2692 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
2693 to allow for free-lang-data replacements similar to verify_type_variant.
2695 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
2696 Alan Hayward <alan.hayward@arm.com>
2697 David Sherwood <david.sherwood@arm.com>
2699 * config/aarch64/aarch64.md (casesi): Use DImode rather than
2700 VOIDmode for the LABEL_REF.
2702 2017-08-22 Richard Biener <rguenther@suse.de>
2704 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
2706 2017-08-22 Richard Biener <rguenther@suse.de>
2708 * common.opt (feliminate-dwarf2-dups): Ignore.
2709 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
2710 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
2711 same_die_p_wrap, compute_section_prefix,
2712 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
2713 (comdat_symbol_id, comdat_symbol_number): Likewise.
2714 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
2716 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
2717 (output_die): Mark unreachable path unreachable.
2718 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
2719 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
2720 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
2721 (dwarf2out_early_finish): Likewise.
2723 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
2725 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
2727 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
2730 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
2731 not VAR_P. Filter attribute warnings with OPT_Wattributes.
2732 (avr_attribute_table) <io, io_low, address>: Initialize
2733 .decl_required with true.
2735 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
2737 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
2738 undocumented debugging options.
2739 (-mvsx-scalar-double): Likewise.
2740 (-mallow-df-permute): Likewise.
2741 (-mvectorize-builtins): Likewise.
2742 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2743 (rs6000_builtin_vectorized_function): Likewise.
2744 (rs6000_builtin_md_vectorized_function): Likewise.
2745 (rs6000_opt_vars): Likewise.
2747 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
2750 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
2751 (*btrq_imm): Rename from *btrq.
2752 (*btcq_imm): Rename from *btcq.
2753 (btsc): New code attribute.
2754 (*<btsc><mode>): New insn pattern.
2755 (*btr<mode>): Ditto.
2756 (*<btsc><mode>_mask): New insn_and_split pattern.
2757 (*btr<mode>_mask): Ditto.
2759 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2760 Alan Hayward <alan.hayward@arm.com>
2761 David Sherwood <david.sherwood@arm.com>
2763 * function.c (pad_below): Simplify padding calculation.
2765 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2766 Alan Hayward <alan.hayward@arm.com>
2767 David Sherwood <david.sherwood@arm.com>
2769 * target.def (function_prologue): Remove frame size argument.
2770 (function_epilogue): Likewise.
2771 * doc/tm.texi: Regenerate.
2772 * final.c (final_start_function): Update call to function_prologue.
2773 (final_end_function): Update call to function_epilogue.
2774 (default_function_pro_epilogue): Remove frame size argument.
2775 * output.h (default_function_pro_epilogue): Likewise.
2776 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
2777 (arm_output_function_prologue): Likewise.
2778 * config/frv/frv.c (frv_function_prologue): Likewise.
2779 (frv_function_epilogue): Likewise.
2780 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
2781 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
2782 (ia64_output_function_epilogue): Likewise.
2783 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
2784 (m32r_output_function_epilogue): Likewise.
2785 * config/microblaze/microblaze.c (microblaze_function_prologue)
2786 (microblaze_function_epilogue): Likewise.
2787 * config/mips/mips.c (mips_output_function_prologue): Likewise.
2788 (mips_output_function_epilogue): Likewise.
2789 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
2790 (mmix_target_asm_function_epilogue): Likewise.
2791 * config/msp430/msp430.c (msp430_start_function): Likewise.
2792 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
2793 (nds32_asm_function_epilogue): Likewise.
2794 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
2795 * config/pa/pa.c (pa_output_function_prologue): Likewise.
2796 (pa_output_function_epilogue): Likewise.
2797 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
2798 (rs6000_output_function_epilogue): Likewise.
2799 * config/rl78/rl78.c (rl78_start_function): Likewise.
2800 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
2801 (rs6000_output_function_epilogue): Likewise.
2802 * config/rx/rx.c (rx_output_function_prologue): Likewise.
2803 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
2804 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
2805 (sparc_asm_function_epilogue): Likewise.
2807 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2809 * tree.h (type_has_mode_precision_p): New function.
2810 * convert.c (convert_to_integer_1): Use it.
2811 * expr.c (expand_expr_real_2): Likewise.
2812 (expand_expr_real_1): Likewise.
2813 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
2814 * match.pd: Likewise.
2815 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
2816 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
2817 * tree-tailcall.c (process_assignment): Likewise.
2818 * tree-vect-loop.c (vectorizable_reduction): Likewise.
2819 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
2820 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
2821 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2822 (vectorizable_assignment): Likewise.
2823 (vectorizable_shift): Likewise.
2824 (vectorizable_operation): Likewise.
2825 * tree-vrp.c (register_edge_assert_for_2): Likewise.
2827 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
2829 * match.pd: Add pow (C, x) simplification.
2831 2017-08-21 Richard Biener <rguenther@suse.de>
2833 PR tree-optimization/81900
2834 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
2835 for blocks with abnormal predecessors.
2836 (compute_antic): Do not set visited flag prematurely.
2838 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
2841 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
2843 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2845 * stor-layout.h (vector_type_mode): Move to...
2846 * tree.h (vector_type_mode): ...here.
2847 * stor-layout.c (vector_type_mode): Move to...
2848 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
2850 2017-08-21 Richard Biener <rguenther@suse.de>
2852 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
2853 register_external_die hooks.
2854 (debug_false_tree_charstarstar_uhwistar): Declare.
2855 (debug_nothing_tree_charstar_uhwi): Likewise.
2856 * debug.c (do_nothing_debug_hooks): Adjust.
2857 (debug_false_tree_charstarstar_uhwistar): New do nothing.
2858 (debug_nothing_tree_charstar_uhwi): Likewise.
2859 * dbxout.c (dbx_debug_hooks): Adjust.
2860 (xcoff_debug_hooks): Likewise.
2861 * sdbout.c (sdb_debug_hooks): Likewise.
2862 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
2863 * dwarf2out.c (macinfo_label_base): New global.
2864 (dwarf2out_register_external_die): New function for the
2865 register_external_die hook.
2866 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
2867 (dwarf2_debug_hooks): Use them.
2868 (dwarf2_lineno_debug_hooks): Adjust.
2869 (struct die_struct): Add with_offset flag.
2870 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
2871 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
2872 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
2873 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
2874 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
2875 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
2876 defining section names for the early LTO debug variants.
2877 (reset_indirect_string): New helper.
2878 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
2879 (print_dw_val): Add support for offsetted symbol references.
2880 (get_ultimate_context): Split out from is_cxx.
2881 (is_cxx): Use get_ultimate_context.
2882 (is_fortran): Add decl overload.
2883 (compute_comp_unit_symbol): Split out worker from
2884 compute_section_prefix.
2885 (compute_section_prefix): Call compute_comp_unit_symbol and
2886 set comdat_type_p here.
2887 (output_die): Skip DIE symbol output for the LTO added one.
2888 Handle DIE symbol references with offset.
2889 (output_comp_unit): Guard section name mangling properly.
2890 For LTO debug sections emit a symbol at the section beginning
2891 which we use to refer to its DIEs.
2892 (add_abstract_origin_attribute): For DIEs registered via
2893 dwarf2out_register_external_die directly refer to the early
2894 DIE rather than indirectly through the shadow one we created.
2895 Remove obsolete call to dwarf2out_abstract_function for
2896 non-function/block origins.
2897 (gen_array_type_die): When generating early LTO debug do
2898 not emit DW_AT_string_length.
2899 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
2900 late when in LTO. As suggested place a gcc_unreachable for
2901 the DECL_ABSTRACT_P case.
2902 (gen_subprogram_die): Avoid another specification DIE
2903 for early built declarations/definitions for the late LTO case.
2904 (gen_variable_die): Add type references for late duplicated VLA dies
2906 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
2907 we have the abstract instance already.
2908 (process_scope_var): Adjust decl DIE contexts in LTO which
2909 first puts them in limbo.
2910 (gen_decl_die): Do not generate type DIEs late apart from
2911 types for VLAs or for decls we do not yet have a DIE. Do not
2912 call dwarf2out_abstract_function late.
2913 (dwarf2out_early_global_decl): Make sure to create DIEs
2914 for abstract instances of a decl first.
2915 (dwarf2out_late_global_decl): Adjust comment.
2916 (output_macinfo_op): With multiple macro sections use
2917 macinfo_label_base to distinguish labels.
2918 (output_macinfo): Likewise. Update macinfo_label_base.
2919 Pass in the line info label.
2920 (note_variable_value_in_expr): When generating LTO resolve
2921 all variable values here by generating DIEs as needed.
2922 (init_sections_and_labels): Add early LTO debug flag parameter
2923 and generate different sections and names if set. Add generation
2924 counter for the labels so we can have multiple of them.
2925 (reset_dies): Helper to allow DIEs to be output multiple times.
2926 (dwarf2out_finish): When outputting DIEs to the fat part of an
2927 LTO object first reset DIEs.
2928 (dwarf2out_early_finish): Output early DIEs when generating LTO.
2929 (modified_type_die): Check for decl_ultimate_origin being self
2931 (gen_type_die_with_usage): Likewise.
2932 (gen_typedef_die): Allow decl_ultimate_origin being self.
2933 (set_decl_abstract_flags): Remove.
2934 (set_block_abstract_flags): Likewise.
2935 (dwarf2out_abstract_function): Treat the early generated DIEs
2936 as the abstract copy and only add DW_AT_inline and
2937 DW_AT_artificial here and call set_decl_origin_self.
2938 If the DIE has an abstract origin don't do anything.
2939 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
2940 if we have none yet (Go fails to build one, PR78628).
2941 (variably_modified_type_p): Prevent endless recursion for Ada
2942 cyclic pointer types.
2943 * lto-streamer-in.c: Include debug.h.
2944 (dref_queue): New global.
2945 (lto_read_tree_1): Stream in DIE references.
2946 (lto_input_tree): Register DIE references.
2947 (input_function): Stream DECL_DEBUG_ARGS.
2948 * lto-streamer-out.c: Include debug.h.
2949 (lto_write_tree_1): Output DIE references.
2950 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
2951 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
2952 (output_function): Stream DECL_DEBUG_ARGS.
2953 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
2954 Stream DECL_ABSTRACT_ORIGIN.
2955 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
2956 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
2957 DECL_CONTEXT for file-scope decls.
2958 * lto-streamer.h (struct dref_entry): Declare.
2959 (dref_queue): Likewise.
2960 * cfgexpand.c (pass_expand::execute): Do not call the
2961 outlining_inline_function hook here.
2962 * lto-wrapper.c (debug_obj): New global.
2963 (tool_cleanup): Unlink it if required.
2964 (debug_objcopy): New function.
2965 (run_gcc): Handle early debug sections in the IL files by
2966 extracting them to separate files, partially linkin them and
2967 feeding the result back as result to the linker.
2968 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
2969 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
2970 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
2971 sections into a separate segment.
2972 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
2974 (darwin_asm_dwarf_section): Likewise.
2975 (darwin_asm_output_dwarf_offset): Likewise.
2976 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
2978 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2979 Alan Hayward <alan.hayward@arm.com>
2980 David Sherwood <david.sherwood@arm.com>
2982 * read-md.h (md_reader::record_potential_iterator_use): Replace
2983 pointer argument with an rtx and an index.
2984 * read-rtl.c (iterator_group::apply_iterator): Likewise.
2985 (apply_mode_iterator): Likewise.
2986 (apply_code_iterator): Likewise.
2987 (apply_int_iterator): Likewise.
2988 (apply_subst_iterator): Likewise.
2989 (record_iterator_use): Likewise.
2990 (record_attribute_use): Likewise.
2991 (md_reader::record_potential_iterator_use): Likewise. Update calls
2992 to record_iterator_use and apply_iterator.
2993 (iterator_use): Replace ptr with x and index.
2994 (attribute_use): Likewise.
2995 (apply_attribute_uses): Update calls to apply_iterator.
2996 (apply_iterators): Likewise. Update initialization of iterator_use.
2997 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
2998 and record_potential_iterator_use.
2999 (rtx_reader::read_rtx_operand): Likewise.
3001 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3002 Alan Hayward <alan.hayward@arm.com>
3003 David Sherwood <david.sherwood@arm.com>
3005 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
3008 2017-08-21 Richard Biener <rguenther@suse.de>
3011 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
3012 at struct end conservatively when comparing common bases.
3014 2017-08-21 Richard Biener <rguenther@suse.de>
3016 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
3017 (mem_ref_in_stmt): Remove.
3018 (determine_max_movement): Use ref index to get at the reference.
3019 (invariantness_dom_walker::before_dom_children): Deal with
3020 lim data already initialized.
3021 (gather_mem_refs_stmt): Initialize lim data and record ref index.
3023 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
3025 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
3026 (TARGET_ISA_ROUND): Ditto.
3027 (TARGET_ROUND): Ditto.
3028 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
3029 * config/i386/i386.md: Ditto.
3030 * config/i386/sse.md: Ditto.
3031 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
3032 with OPTION_MASK_ISA_SSE4_1.
3034 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
3037 * doc/extend.texi (x86 Built-in Functions): Correct the name of
3038 __builtin_ia32_lzcnt_u16.
3040 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
3043 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
3044 (rs6000_set_current_function): Rewrite function to use it.
3046 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
3049 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
3050 and TYPE_WARN_IF_NOT_ALIGN.
3051 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
3052 (handle_warn_if_not_align): New.
3053 (place_union_field): Call handle_warn_if_not_align.
3054 (place_field): Call handle_warn_if_not_align.
3055 Copy TYPE_WARN_IF_NOT_ALIGN.
3056 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
3057 (layout_type): Likewise.
3058 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
3060 (tree_decl_common): Add warn_if_not_align.
3061 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
3062 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
3063 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
3064 (DECL_WARN_IF_NOT_ALIGN): Likewise.
3065 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
3066 * doc/extend.texi: Document warn_if_not_aligned attribute.
3067 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
3069 2017-08-17 Martin Liska <mliska@suse.cz>
3072 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
3073 (get_data_dependence): Use it as pointer type.
3074 (distribute_loop): Likewise.
3076 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3078 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
3079 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
3080 (p8_vmrgow_<mode>_direct): New define_insn.
3081 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
3082 handle endianness for vmrgew and vmrgow permute patterns.
3084 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
3086 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
3087 * config/rs6000/rs6000-cpus.def: Remove comment.
3088 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
3089 (POWERPC_MASKS): Likewise.
3090 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
3091 use of TARGET_VSX_TIMODE.
3092 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
3093 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
3094 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
3095 (rs6000_option_override_internal): Remove dead code.
3096 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
3097 (rs6000_legitimize_reload_address): Likewise.
3098 (rs6000_legitimate_address_p): Likewise.
3099 (rs6000_opt_masks): Delete "vsx-timode".
3100 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
3101 from function comment.
3102 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
3103 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
3104 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
3106 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
3107 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
3108 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
3111 (define_peephole2): Likewise.
3113 2017-08-17 Martin Sebor <msebor@redhat.com>
3116 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
3117 past the end of an array.
3118 (test_pp_format): Add test cases.
3120 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
3122 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
3123 missing ECF_NOTHROW flags.
3125 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
3128 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
3129 operands residing in integer registers.
3130 (*vsx_le_perm_load_<mode>): Likewise.
3131 (*vsx_le_perm_store_<mode>): Likewise.
3132 (define_peephole2): Add peepholes to optimize the above.
3134 2017-08-17 Marek Polacek <polacek@redhat.com>
3137 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
3138 to mimic what shorten_compare did. Change the return type to bool.
3139 (fold_cond_expr_with_comparison): Update call to
3140 operand_equal_for_comparison_p.
3141 (fold_ternary_loc): Likewise.
3143 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
3145 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
3147 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
3149 2017-08-17 Richard Biener <rguenther@suse.de>
3151 * tree-ssa-structalias.c (solve_graph): When propagating
3152 to successors update the graphs succ edges and avoid duplicate work.
3154 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
3157 * config/i386/i386.c (ix86_option_override_internal): Save target
3158 specific options after ix86_stack_protector_guard_reg was changed.
3160 2017-08-17 Richard Biener <rguenther@suse.de>
3162 PR tree-optimization/81827
3163 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
3164 (new_var_info): Initialize it conservatively.
3165 (get_call_vi): Mark register vars.
3166 (new_scalar_tmp_constraint_exp): Likewise.
3167 (handle_rhs_call): Likewise.
3168 (handle_const_call): Likewise.
3169 (create_function_info_for): Likewise.
3170 (solve_constraints): Sort varinfos to separate register from
3171 non-register vars to pack points-to solution bitmaps during
3174 2017-08-17 Marek Polacek <polacek@redhat.com>
3176 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
3178 2017-08-17 Richard Biener <rguenther@suse.de>
3180 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
3181 to true when overflow is undefined and we saturated the result.
3183 2017-08-17 Alan Modra <amodra@gmail.com>
3186 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
3187 Don't use store multiple if only one reg needs saving.
3188 (interesting_frame_related_regno): New function.
3189 (rs6000_frame_related): Don't emit frame info for regs that
3191 (rs6000_emit_epilogue): Likewise.
3193 2017-08-16 Nathan Sidwell <nathan@acm.org>
3195 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
3196 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
3197 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
3198 * tree.c (free_lang_data_in_type): Use else-if chain. Always
3199 clear TYPE_LANG_1. Remove obsolete member-function stripping.
3200 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
3201 (verify_type): Adjust for TYPE_BINFO move.
3202 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
3203 process TYPE_BINFO directly.
3204 (hash_tree): Likewise.
3205 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
3207 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
3210 2017-08-16 David Malcolm <dmalcolm@redhat.com>
3212 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
3214 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
3217 * config/i386/i386.md (*anddi_1_btr): Change predicates of
3218 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
3219 Add ix86_binary_operator_ok to insn constraint.
3220 (*iordi_1_bts): Ditto.
3221 (*xordi_1_btc): Ditto.
3222 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
3223 Update corresponding peephole2 pattern.
3227 2017-08-16 Bin Cheng <bin.cheng@arm.com>
3229 PR tree-optimization/81832
3230 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
3231 copy loop header which has IFN_LOOP_DIST_ALIAS call.
3233 2017-08-16 Marek Polacek <polacek@redhat.com>
3236 * fold-const.c (fold_indirect_ref_1): Restore original behavior
3237 regarding size_zero_node.
3239 2017-08-16 Martin Liska <mliska@suse.cz>
3242 * config.gcc: Respect previously set extra_objs in case
3245 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
3247 PR tree-optimization/81835
3248 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
3249 the phi SSA_NAME. Check that the condition in a COND_EXPR does
3250 not depend on the phi.
3252 2017-08-16 Alan Modra <amodra@gmail.com>
3254 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
3257 2017-08-16 Alan Modra <amodra@gmail.com>
3259 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
3260 (save_reg_p): ..into this. Update all callers.
3261 (first_reg_to_save): Simplify.
3263 2017-08-16 Alan Modra <amodra@gmail.com>
3265 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
3268 2017-08-15 Joseph Myers <joseph@codesourcery.com>
3272 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
3273 constant count if that count is less than 32.
3275 2017-08-15 Nathan Sidwell <nathan@acm.org>
3277 * gcc.c (execute): Emit friendlier message if inferior is killed
3278 by an external cause.
3280 2017-08-15 Richard Biener <rguenther@suse.de>
3282 PR tree-optimization/81790
3283 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
3284 CONSTRUCTORs from simplifying and VN.
3286 2017-08-14 Martin Sebor <msebor@redhat.com>
3288 * builtin-attrs.def: Add comments.
3290 2017-08-14 Martin Sebor <msebor@redhat.com>
3293 * doc/extend.texi (attribute nonstring): Document new attribute.
3295 2017-08-14 Martin Sebor <msebor@redhat.com>
3298 * tree-diagnostic.c (default_tree_printer): Handle %G.
3299 * gimple-pretty-print.h (percent_G_format): Declare new function.
3300 * gimple-pretty-print.c (percent_G_format): Define.
3301 * tree-pretty-print.c (percent_K_format): Add argument.
3303 2017-08-14 Martin Sebor <msebor@redhat.com>
3305 PR translation/79998
3306 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
3307 Remove a stray space.
3309 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
3312 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
3313 (*iordi_1_bts): Ditto.
3314 (*xordi_1_btc): Ditto.
3316 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3319 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
3320 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3322 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
3323 quoted strings, and make more translator-friendly.
3324 (darwin_rs6000_override_options): Likewise.
3325 (rs6000_option_override_internal): Likewise.
3326 (rs6000_return_in_memory): Fix overlong line.
3327 (init_cmulative_args): Use quoted strings, and make more
3328 translator-friendly.
3329 (rs6000_pass_by_reference): Fix overlong line.
3330 (def_builtin): Use quoted strings.
3331 (altivec_expand_predicate_builtin): Use quoted strings, and make
3332 more translator-friendly.
3333 (htm_expand_builtin): Use quoted strings.
3334 (cpu_expand_builtin): Use quoted strings, and make more
3335 translator-friendly.
3336 (altivec_expand_builtin): Likewise.
3337 (paired_expand_predicate_builtin): Likewise.
3338 (rs6000_invalid_builtin): Likewise.
3339 (builtin_function_type): Use quoted strings.
3340 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
3341 more translator-friendly.
3342 (rs6000_trampoline_init): Likewise.
3343 (rs6000_handle_altivec_attribute): Likewise.
3344 (rs6000_inner_target_options): Use quoted strings.
3345 (rs6000_disable_incompatible_switches): Likewise.
3346 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
3347 strings, and make more translator-friendly.
3348 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
3350 2017-08-14 Bin Cheng <bin.cheng@arm.com>
3352 PR tree-optimization/81799
3353 * tree-loop-distribution.c (version_loop_by_alias_check): Force
3354 cond_expr to simple gimple operand.
3356 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
3359 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
3361 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
3365 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
3366 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
3367 TARGET_GASISR_PROLOGUES.
3368 * config/avr/avr.c (avr_option_override): Same.
3369 (avr_pass_pre_proep::execute): Same.
3371 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
3374 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
3375 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
3377 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
3379 * config/i386/i386.md (*load_tp_<mode>): Redefine as
3380 define_insn_and_split. Split to a memory load from 0 in
3381 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
3382 using PTR mode iterator.
3383 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
3384 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
3385 (*add_tp_<mode>): Redefine as define_insn_and_split.
3386 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
3387 address space. Merge with *add_tp_x32 using PTR mode iterator.
3388 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
3389 Split to an add with a memory load from 0 in
3390 DEFAULT_TLS_SEG_REG address space.
3392 2017-08-12 Andrew Pinski <apinski@cavium.com>
3394 * config/aarch64/aarch64-option-extensions.def (rdma):
3395 Fix feature string to what Linux prints out in /proc/cpuinfo.
3397 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
3400 * dwarf2out.c (modified_type_die): For C typedef types that have
3401 an ultimate origin, process the ultimate origin instead of the
3403 (gen_typedef_die): Assert that input DECLs have no ultimate
3405 (gen_type_die_with_usage): For typedef variants that have an
3406 ultimate origin, just call gen_decl_die on the original DECL.
3407 (process_scope_var): Avoid creating DIEs for local typedefs and
3408 concrete static variables.
3410 2017-08-12 Alan Modra <amodra@gmail.com>
3414 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
3415 match gnu-user.h startfile.
3416 (ENDFILE_LINUX_SPEC): Similarly.
3418 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
3421 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
3423 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
3425 2017-08-11 Tamar Christina <tamar.christina@arm.com>
3426 * config/aarch64/aarch64.md (mov<mode>): Change.
3427 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
3428 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
3429 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
3431 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
3433 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
3434 for storage order barriers.
3436 2017-08-11 Martin Liska <mliska@suse.cz>
3439 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
3440 variables of void type.
3442 2017-08-11 Martin Liska <mliska@suse.cz>
3444 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
3445 TARGET_SUPPORTS_ALIASES.
3446 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
3447 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
3448 (optimize_weakref): Likewise.
3449 * symtab.c (symtab_node::noninterposable_alias): Likewise.
3450 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
3451 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
3453 2017-08-11 Martin Liska <mliska@suse.cz>
3456 * config/i386/i386.c (make_resolver_func): Do complete
3457 refactoring of the function.
3459 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
3462 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
3463 * config/i386/i386.c (ix86_stack_protect_guard): Use
3464 ix86_stack_protect_guard_symbol_str to generate varible declaration.
3465 * doc/invoke.texi (x86 Options): Document
3466 -mstack-protector-guard-symbol= option.
3468 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
3470 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
3471 * config/i386/i386.c (ix86_split_stack_guard): New function.
3472 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
3473 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
3474 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
3475 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
3476 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
3477 (split_stack_space_check): Call ix86_split_stack_guard.
3479 2017-08-10 Martin Sebor <msebor@redhat.com>
3481 * print-tree.c (print_node): Print location using the established
3483 Replace spaces with colons.
3484 (debug_raw, debug): Ditto.
3486 2017-08-10 Martin Sebor <msebor@redhat.com>
3489 * pretty-print.c (pp_format): Correct the handling of %s precision.
3491 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
3494 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
3496 (ix86_finalize_stack_frame_flags): This. Also clear
3497 frame_pointer_needed if -fno-omit-frame-pointer is used without
3499 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
3500 with ix86_finalize_stack_frame_flags.
3501 (ix86_expand_epilogue): Likewise.
3502 (ix86_expand_split_stack_prologue): Likewise.
3503 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
3505 2017-08-10 Martin Liska <mliska@suse.cz>
3508 * c-attribs.c (handle_target_attribute):
3509 Report warning for an empty string argument of target attribute.
3511 2017-08-09 Jakub Jelinek <jakub@redhat.com>
3514 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
3516 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
3517 or DECL_NONLOCAL labels.
3518 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
3519 or DECL_NONLOCAL labels here.
3521 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
3523 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
3524 to indicate when early gimple folding has been disabled.
3525 (rs6000_gimple_fold_builtin): Add debug content.
3526 (rs6000_invalid_builtin): Fix whitespace.
3527 (rs6000_expand_builtin): Fix whitespace.
3528 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
3530 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
3533 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
3534 SAVE_MULTIPLE if not all the registers that saves, should be saved.
3536 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
3538 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
3539 (qdf24xx): Likewise.
3540 * config/aarch64/aarch64.md: Include falkor.md.
3541 * config/aarch64/falkor.md: New.
3543 2017-08-09 Marek Polacek <polacek@redhat.com>
3546 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
3547 * diagnostic.c (emit_diagnostic): Add a comment.
3548 (emit_diagnostic_valist): New function.
3550 2017-08-09 Marek Polacek <polacek@redhat.com>
3553 * input.c (make_location): New overload.
3554 * input.h (make_location): Declare.
3556 2017-08-08 Alan Modra <amodra@gmail.com>
3557 H.J. Lu <hongjiu.lu@intel.com>
3560 * gcc.c (NO_PIE_SPEC): Delete.
3561 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
3563 (LINK_PIE_SPEC): ..to here.
3564 (LINK_COMMAND_SPEC): Support -no-pie.
3565 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
3566 chain of crtbegin*.o selection, update for PIE_SPEC changes and
3568 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
3569 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
3570 (ENDFILE_CRTEND_SPEC): Similarly.
3572 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
3575 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
3576 (mstack-protector-guard-offset=): Ditto.
3577 * config/i386/i386.c (ix86_option_override): Handle
3578 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
3580 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
3581 ix86_stack_protect_guard_offset variables.
3582 (TARGET_STACK_PROTECT_GUARD): Always define.
3583 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
3584 and -mstack-protector-guard-offset= options.
3586 2017-08-08 Bin Cheng <bin.cheng@arm.com>
3588 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
3589 boundary case for the last candidate.
3591 2017-08-08 Bin Cheng <bin.cheng@arm.com>
3593 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
3594 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
3596 2017-08-08 Tamar Christina <tamar.christina@arm.com>
3599 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
3600 * config/aarch64/aarch64-builtins.c
3601 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
3602 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
3603 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
3605 2017-08-08 Tamar Christina <tamar.christina@arm.com>
3606 Andrew Pinski <pinskia@gmail.com>
3609 * internal-fn.def (XORSIGN): New.
3610 * optabs.def (xorsign_optab): New.
3611 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
3612 (convert_expand_mult_copysign): New.
3613 (pass_optimize_widening_mul::execute): Call
3614 convert_expand_mult_copysign.
3616 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3618 PR tree-optimization/81354
3619 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
3620 Insert on edges rather than explicitly creating landing pads.
3621 (analyze_candidates_and_replace): Commit edge inserts.
3623 2017-08-08 Richard Biener <rguenther@suse.de>
3626 * tree-ssa-loop-niter.c: Include tree-dfa.h.
3627 (expand_simple_operations): Also look through ADDR_EXPRs with
3628 MEM_REF bases treating them as POINTER_PLUS_EXPR.
3630 2017-08-08 Richard Biener <rguenther@suse.de>
3632 PR tree-optimization/81723
3633 * tree-vect-slp.c (struct bst_traits): New hash traits.
3634 (bst_fail): New global.
3635 (vect_build_slp_tree_2): New worker, split out from ...
3636 (vect_build_slp_tree): ... this now wrapping it with using
3637 bst_fail set to cache SLP tree build fails. Properly handle
3639 (vect_analyze_slp_instance): Allocate and free bst_fail.
3641 2017-08-08 Martin Liska <mliska@suse.cz>
3644 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
3645 LABEL_DECLs that can be from a different function.
3647 2017-08-08 Bin Cheng <bin.cheng@arm.com>
3649 PR tree-optimization/81744
3650 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
3651 loop's number of iterations.
3653 2017-08-08 Martin Liska <mliska@suse.cz>
3655 * asan.c: Include header files.
3656 * attribs.c (build_decl_attribute_variant): New function moved
3658 (build_type_attribute_qual_variant): Likewise.
3659 (cmp_attrib_identifiers): Likewise.
3660 (simple_cst_list_equal): Likewise.
3661 (omp_declare_simd_clauses_equal): Likewise.
3662 (attribute_value_equal): Likewise.
3663 (comp_type_attributes): Likewise.
3664 (build_type_attribute_variant): Likewise.
3665 (lookup_ident_attribute): Likewise.
3666 (remove_attribute): Likewise.
3667 (merge_attributes): Likewise.
3668 (merge_type_attributes): Likewise.
3669 (merge_decl_attributes): Likewise.
3670 (merge_dllimport_decl_attributes): Likewise.
3671 (handle_dll_attribute): Likewise.
3672 (attribute_list_equal): Likewise.
3673 (attribute_list_contained): Likewise.
3674 * attribs.h (lookup_attribute): New function moved from tree.[ch].
3675 (lookup_attribute_by_prefix): Likewise.
3676 * bb-reorder.c: Include header files.
3677 * builtins.c: Likewise.
3678 * calls.c: Likewise.
3679 * cfgexpand.c: Likewise.
3680 * cgraph.c: Likewise.
3681 * cgraphunit.c: Likewise.
3682 * convert.c: Likewise.
3683 * dwarf2out.c: Likewise.
3684 * final.c: Likewise.
3685 * fold-const.c: Likewise.
3686 * function.c: Likewise.
3687 * gimple-expr.c: Likewise.
3688 * gimple-fold.c: Likewise.
3689 * gimple-pretty-print.c: Likewise.
3690 * gimple.c: Likewise.
3691 * gimplify.c: Likewise.
3692 * hsa-common.c: Likewise.
3693 * hsa-gen.c: Likewise.
3694 * internal-fn.c: Likewise.
3695 * ipa-chkp.c: Likewise.
3696 * ipa-cp.c: Likewise.
3697 * ipa-devirt.c: Likewise.
3698 * ipa-fnsummary.c: Likewise.
3699 * ipa-inline.c: Likewise.
3700 * ipa-visibility.c: Likewise.
3702 * lto-cgraph.c: Likewise.
3703 * omp-expand.c: Likewise.
3704 * omp-general.c: Likewise.
3705 * omp-low.c: Likewise.
3706 * omp-offload.c: Likewise.
3707 * omp-simd-clone.c: Likewise.
3708 * opts-global.c: Likewise.
3709 * passes.c: Likewise.
3710 * predict.c: Likewise.
3711 * sancov.c: Likewise.
3712 * sanopt.c: Likewise.
3713 * symtab.c: Likewise.
3714 * toplev.c: Likewise.
3715 * trans-mem.c: Likewise.
3716 * tree-chkp.c: Likewise.
3717 * tree-eh.c: Likewise.
3718 * tree-into-ssa.c: Likewise.
3719 * tree-object-size.c: Likewise.
3720 * tree-parloops.c: Likewise.
3721 * tree-profile.c: Likewise.
3722 * tree-ssa-ccp.c: Likewise.
3723 * tree-ssa-live.c: Likewise.
3724 * tree-ssa-loop.c: Likewise.
3725 * tree-ssa-sccvn.c: Likewise.
3726 * tree-ssa-structalias.c: Likewise.
3727 * tree-ssa.c: Likewise.
3728 * tree-streamer-in.c: Likewise.
3729 * tree-vectorizer.c: Likewise.
3730 * tree-vrp.c: Likewise.
3732 * ubsan.c: Likewise.
3733 * varasm.c: Likewise.
3734 * varpool.c: Likewise.
3735 * tree.c: Remove functions moved to attribs.[ch].
3737 * config/aarch64/aarch64.c: Add attrs.h header file.
3738 * config/alpha/alpha.c: Likewise.
3739 * config/arc/arc.c: Likewise.
3740 * config/arm/arm.c: Likewise.
3741 * config/avr/avr.c: Likewise.
3742 * config/bfin/bfin.c: Likewise.
3743 * config/c6x/c6x.c: Likewise.
3744 * config/cr16/cr16.c: Likewise.
3745 * config/cris/cris.c: Likewise.
3746 * config/darwin.c: Likewise.
3747 * config/epiphany/epiphany.c: Likewise.
3748 * config/fr30/fr30.c: Likewise.
3749 * config/frv/frv.c: Likewise.
3750 * config/ft32/ft32.c: Likewise.
3751 * config/h8300/h8300.c: Likewise.
3752 * config/i386/winnt.c: Likewise.
3753 * config/ia64/ia64.c: Likewise.
3754 * config/iq2000/iq2000.c: Likewise.
3755 * config/lm32/lm32.c: Likewise.
3756 * config/m32c/m32c.c: Likewise.
3757 * config/m32r/m32r.c: Likewise.
3758 * config/m68k/m68k.c: Likewise.
3759 * config/mcore/mcore.c: Likewise.
3760 * config/microblaze/microblaze.c: Likewise.
3761 * config/mips/mips.c: Likewise.
3762 * config/mmix/mmix.c: Likewise.
3763 * config/mn10300/mn10300.c: Likewise.
3764 * config/moxie/moxie.c: Likewise.
3765 * config/msp430/msp430.c: Likewise.
3766 * config/nds32/nds32-isr.c: Likewise.
3767 * config/nds32/nds32.c: Likewise.
3768 * config/nios2/nios2.c: Likewise.
3769 * config/nvptx/nvptx.c: Likewise.
3770 * config/pa/pa.c: Likewise.
3771 * config/pdp11/pdp11.c: Likewise.
3772 * config/powerpcspe/powerpcspe.c: Likewise.
3773 * config/riscv/riscv.c: Likewise.
3774 * config/rl78/rl78.c: Likewise.
3775 * config/rx/rx.c: Likewise.
3776 * config/s390/s390.c: Likewise.
3777 * config/sh/sh.c: Likewise.
3778 * config/sol2.c: Likewise.
3779 * config/sparc/sparc.c: Likewise.
3780 * config/spu/spu.c: Likewise.
3781 * config/stormy16/stormy16.c: Likewise.
3782 * config/tilegx/tilegx.c: Likewise.
3783 * config/tilepro/tilepro.c: Likewise.
3784 * config/v850/v850.c: Likewise.
3785 * config/vax/vax.c: Likewise.
3786 * config/visium/visium.c: Likewise.
3787 * config/xtensa/xtensa.c: Likewise.
3789 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
3792 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
3793 constraints since the -mupper-regs-* switches have been
3795 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
3796 into a vector from a double word element that was extracted from
3797 another vector, and eliminate extra XXPERMDI instructions.
3798 (vsx_concat_<mode>_2): Likewise.
3799 (vsx_concat_<mode>_3): Likewise.
3800 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
3801 concat to allow optimizing inserts from previous extracts.
3803 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
3805 * config/i386/i386.c (ix86_stack_protect_guard): Generate
3806 memory reference to a SSP offset in TLS address space.
3807 (ix86_print_operand) <case '@'>: Remove.
3808 (ix86_print_operand_punct_valid_p): Remove '@' code.
3809 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
3811 (stack_tls_protect_set_<mode>): Remove.
3812 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
3813 (stack_tls_protect_test_<mode>): Remove.
3814 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
3816 2017-08-07 Olivier Hainque <hainque@adacore.com>
3819 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
3821 2017-08-07 Douglas Rupp <rupp@adacore.com>
3823 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
3824 variable was referenced as multidir in command.
3826 2017-08-07 Jakub Jelinek <jakub@redhat.com>
3829 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
3832 2017-08-07 Martin Liska <mliska@suse.cz>
3834 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
3835 * config/rl78/rl78.c: Add include of attribs.h.
3836 * config/sh/sh.c: Likewise.
3837 * config/v850/v850.c: Likewise.
3839 2017-08-07 Tom de Vries <tom@codesourcery.com>
3842 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
3844 2017-08-07 Martin Liska <mliska@suse.cz>
3846 * config/mips/mips.c: Include attribs.h.
3848 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
3851 * doc/invoke.texi: Document change in behvaior for -Ofast for
3854 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
3856 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
3858 (aarch64_pop_regs): Likewise.
3859 (aarch64_gen_load_pair): Likewise.
3860 (aarch64_save_callee_saves): Likewise.
3861 (aarch64_restore_callee_saves): Likewise.
3863 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
3865 * config/i386/i386.c: Revert the last change.
3867 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
3870 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
3872 (ix86_finalize_stack_frame_flags): This. Also clear
3873 frame_pointer_needed if -fno-omit-frame-pointer is used without
3875 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
3876 with ix86_finalize_stack_frame_flags.
3877 (ix86_expand_epilogue): Likewise.
3878 (ix86_expand_split_stack_prologue): Likewise.
3880 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
3883 * config/i386/i386.c (get_builtin_code_for_version): Set priority
3884 to P_AES for Westmere.
3886 2017-08-07 Jonathan Yong <10walls@gmail.com>
3888 * config/i386/mingw.opt (fset-stack-executable): Removed.
3889 * config/i386/cygming.opt (fset-stack-executable): Moved
3891 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
3893 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
3895 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
3897 2017-08-07 Marek Polacek <polacek@redhat.com>
3900 * fold-const.c (fold_indirect_ref_1): Check type_domain.
3902 2017-08-07 Martin Liska <mliska@suse.cz>
3904 * attribs.h (canonicalize_attr_name): New function.
3905 (cmp_attribs): Move from c-format.c and adjusted.
3906 (is_attribute_p): Moved from tree.h.
3907 * tree-inline.c: Add new includes.
3908 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
3909 (private_is_attribute_p): Remove.
3910 (private_lookup_attribute): Likewise.
3911 (private_lookup_attribute_by_prefix): Simplify.
3912 (remove_attribute): Use is_attribute_p.
3913 * tree.h: Remove removed declarations.
3915 2017-08-07 Jakub Jelinek <jakub@redhat.com>
3918 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
3919 instead of computing it in the function. Formatting fix.
3920 (expand_case): Don't rely on default_edge being the first edge,
3921 clear it if removing it, pass default_edge to
3922 emit_case_dispatch_table.
3923 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
3926 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
3928 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
3929 insn in the function, emit NOP after the insn.
3931 2017-08-06 Tom de Vries <tom@codesourcery.com>
3933 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
3936 2017-08-06 Tom de Vries <tom@codesourcery.com>
3938 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
3941 2017-08-04 Yury Gribov <tetra2005@gmail.com>
3943 PR tree-optimization/57371
3944 * match.pd: New pattern.
3946 2017-08-04 Marek Polacek <polacek@redhat.com>
3949 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
3950 perform the computation in offset_int.
3952 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3954 PR tree-optimization/81136
3955 * tree-vectorizer.h: Include tree-hash-traits.h.
3956 (vec_base_alignments): New typedef.
3957 (vec_info): Add a base_alignments field.
3958 (vect_record_base_alignments): Declare.
3959 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
3961 (DR_IS_CONDITIONAL_IN_STMT): New macro.
3962 (create_data_ref): Add an is_conditional_in_stmt argument.
3963 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
3964 the is_conditional_in_stmt field.
3965 (data_ref_loc): Add an is_conditional_in_stmt field.
3966 (get_references_in_stmt): Set the is_conditional_in_stmt field.
3967 (find_data_references_in_stmt): Update call to create_data_ref.
3968 (graphite_find_data_references_in_stmt): Likewise.
3969 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
3970 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
3971 (vect_record_base_alignment): New function.
3972 (vect_record_base_alignments): Likewise.
3973 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
3974 for nested statements even if we fail to compute a misalignment.
3975 Use pooled base alignments for unconditional references.
3976 (vect_find_same_alignment_drs): Compare base addresses instead
3978 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
3979 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
3981 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3983 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
3984 Add an explicit name for the enum. Use auto_vec for slp_instances
3986 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
3988 (_bb_vec_info): Add a constructor and destructor.
3989 (vinfo_for_stmt): Return NULL for uids of -1 as well.
3990 (destroy_loop_vec_info): Delete.
3991 (vect_destroy_datarefs): Likewise.
3992 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
3993 (vec_info::vec_info): New function.
3994 (vec_info::~vec_info): Likewise.
3995 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
3996 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
3997 destroy_loop_vec_info.
3998 * tree-vect-loop.c (new_loop_vec_info): Replace with...
3999 (_loop_vec_info::_loop_vec_info): ...this.
4000 (destroy_loop_vec_info): Replace with...
4001 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
4002 the stmt_vec_infos. Leave handling of vec_info information to its
4003 destructor. Remove explicit vector releases.
4004 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
4005 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
4006 * tree-vect-slp.c (new_bb_vec_info): Replace with...
4007 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
4008 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
4009 (destroy_bb_vec_info): Replace with...
4010 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
4011 information to its destructor.
4012 (vect_slp_analyze_bb_1): Use new and delete instead of
4013 new_bb_vec_info and destroy_bb_vec_info.
4014 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
4017 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
4019 * tree-data-ref.h (subscript): Add access_fn field.
4020 (data_dependence_relation): Add could_be_independent_p.
4021 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
4022 (same_access_functions): Move to tree-data-ref.c.
4023 * tree-data-ref.c (ref_contains_union_access_p): New function.
4024 (access_fn_component_p): Likewise.
4025 (access_fn_components_comparable_p): Likewise.
4026 (dr_analyze_indices): Add a reference to access_fn_component_p.
4027 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
4029 (constant_access_functions): Likewise.
4030 (add_other_self_distances): Likewise.
4031 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
4032 (initialize_data_dependence_relation): Use XCNEW and remove
4033 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
4034 of access functions that have the same type. Allow the
4035 subsequence to end with different bases in some circumstances.
4036 Record the chosen access functions in SUB_ACCESS_FN.
4037 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
4038 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
4039 (subscript_dependence_tester_1): Likewise dra and drb.
4040 (build_classic_dist_vector): Update calls accordingly.
4041 (subscript_dependence_tester): Likewise.
4042 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
4043 DDR_COULD_BE_INDEPENDENT_P.
4044 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
4045 comp_alias_ddrs instead of may_alias_ddrs.
4046 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
4048 (vect_analyze_data_ref_dependence): Use it if
4049 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
4050 distance vectors if that fails.
4051 (dependence_distance_ge_vf): New function.
4052 (vect_prune_runtime_alias_test_list): Use it. Don't clear
4053 LOOP_VINFO_MAY_ALIAS_DDRS.
4055 2017-08-04 Richard Biener <rguenther@suse.de>
4058 * fold-const.c (fold_binary_loc): Properly restrict
4059 minus_var0 && minus_var1 case when associating undefined overflow
4062 2017-08-04 Tom de Vries <tom@codesourcery.com>
4064 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
4066 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4068 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4069 Don't start diagnostic messages with a capital letter.
4070 * config/rs6000/rs6000.c (rs6000_option_override_internal):
4072 (rs6000_invalid_builtin): Likewise.
4073 (rs6000_trampoline_init): Likewise.
4075 2017-08-03 Jakub Jelinek <jakub@redhat.com>
4078 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
4079 after setting changeable df flags.
4081 2017-08-03 Richard Biener <rguenther@suse.de>
4083 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
4084 up if the use is in USE - X.
4086 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
4088 * toplev.c (dumpfile.h): New include.
4089 (internal_error_reentered): New static function. Use it...
4090 (internal_error_function): ...here to handle reentered internal_error.
4092 2017-08-03 Richard Biener <rguenther@suse.de>
4095 * fold-const.c (split_tree): Add minus_var and minus_con
4096 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
4097 here but always use minus_*.
4098 (associate_trees): Assert we never associate with MINUS_EXPR
4099 and NULL first operand. Do not recurse for PLUS_EXPR operands
4100 when associating as MINUS_EXPR either.
4101 (fold_binary_loc): Track minus_var and minus_con.
4103 2017-08-03 Tom de Vries <tom@codesourcery.com>
4106 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
4107 ACCEL_COMPILER, apply finish_options on
4108 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
4110 2017-08-03 Tom de Vries <tom@codesourcery.com>
4113 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
4114 function_entry_patch_area_size > 0.
4116 2017-08-03 Jakub Jelinek <jakub@redhat.com>
4119 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
4120 instead of 10??LU, perform unit multiplication in wide_int,
4121 don't change alloc_object_size_limit if the limit is larger
4124 PR tree-optimization/81655
4125 PR tree-optimization/81588
4126 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
4127 the case when ranges[i].low and high are 1 for unsigned type with
4131 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
4132 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
4134 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4136 * tree-vrp.h: Add include guard.
4138 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
4141 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
4142 (ud2): New insn pattern.
4143 * config/i386/i386.c (ix86_expand_epilogue):
4144 For naked functions, generate ud2 instead of trap insn.
4146 2017-08-02 Marek Polacek <polacek@redhat.com>
4149 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
4151 2017-08-02 Tom de Vries <tom@codesourcery.com>
4152 Cesar Philippidis <cesar@codesourcery.com>
4154 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
4155 Add missing edge probabilities.
4157 2017-08-02 Tamar Christina <tamar.christina@arm.com>
4159 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
4162 2017-08-02 Jakub Jelinek <jakub@redhat.com>
4165 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
4166 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
4167 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
4168 prologue_seq sequences - if any.
4170 2017-08-02 Richard Biener <rguenther@suse.de>
4172 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
4173 via vectors if supported, integer extracts via punning if supported
4174 or otherwise vector extracts.
4176 2017-08-02 Richard Biener <rguenther@suse.de>
4178 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
4180 (bitmap_insert_into_set): ... this.
4182 2017-08-02 Richard Biener <rguenther@suse.de>
4184 PR tree-optimization/81633
4186 2015-08-17 Alan Hayward <alan.hayward@arm.com>
4188 PR tree-optimization/71752
4189 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
4191 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
4193 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
4194 (machine_function::call_ms2sysv_pad_out): Remove field.
4195 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
4196 (ix86_compute_frame_layout): Likewise.
4198 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
4201 * config/i386/i386.c (ix86_set_func_type): Disallow naked
4202 attribute with interrupt attribute.
4204 2017-08-01 Andrew Pinski <apinski@cavium.com>
4206 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
4207 BIT_INSERT_EXPR's operand 1
4208 to see if the types precision matches.
4210 2017-08-01 Martin Liska <mliska@suse.cz>
4213 * builtins.c (expand_builtin_memcpy_args): Remove.
4214 (expand_builtin_memcpy): Call newly added function
4215 expand_builtin_memory_copy_args.
4216 (expand_builtin_memcpy_with_bounds): Likewise.
4217 (expand_builtin_mempcpy): Remove last argument.
4218 (expand_builtin_mempcpy_with_bounds): Likewise.
4219 (expand_builtin_memory_copy_args): New function created from
4220 expand_builtin_mempcpy_args with small modifications.
4221 (expand_builtin_mempcpy_args): Remove.
4222 (expand_builtin_stpcpy): Remove unused argument.
4223 (expand_builtin): Likewise.
4224 (expand_builtin_with_bounds): Likewise.
4226 2017-08-01 Martin Liska <mliska@suse.cz>
4229 Make mempcpy more optimal (PR middle-end/70140).
4231 2017-08-01 Jakub Jelinek <jakub@redhat.com>
4234 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
4235 __builtin_vec_cmpne verify both arguments are compatible vectors
4236 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
4237 verify arg1_type is a pointer or array type. For __builtin_vec_st,
4238 move computation of aligned to after checking the argument types.
4242 * config/rs6000/vsx.md (vextract_fp_from_shorth,
4243 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
4246 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
4247 Doug Rupp <rupp@adacore.com>
4248 Olivier Hainque <hainque@adacore.com>
4250 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
4251 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
4253 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
4254 for TARGET_OS_CPP_BUILTIN.
4255 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
4256 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
4258 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
4259 passing required abi options to the assembler for EABI configurations.
4260 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
4261 of .text.hot and .text.unlikely sections for kernel modules when
4262 using ARM style exceptions.
4263 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
4264 options. Add EXTRA_CC1_SPEC.
4265 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
4267 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
4269 (ARM_TARGET2_DWARF_FORMAT): Define.
4270 * config/arm/t-vxworks: Adjust multilib control to removal of the
4271 Diab command line options.
4273 2017-08-01 Martin Liska <mliska@suse.cz>
4275 PR gcov-profile/81561
4276 * gcov.c (unblock): Make unblocking safe as we need to preserve
4277 index correspondence of blocks and block_lists.
4279 2017-08-01 Richard Biener <rguenther@suse.de>
4281 PR tree-optimization/81181
4282 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4283 (compute_antic): ... end of iteration here.
4285 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
4287 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
4288 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
4289 (ftree-slp-vectorize): Likewise.
4290 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
4291 can no longer be set independent of flag_tree_loop_vectorize.
4292 * omp-general.c (emp_max_vf): Likewise.
4293 * opts.c (enable_fdo_optimizations): Remove references to
4294 flag_tree_vectorize, these are now implicit.
4295 (common_handle_option): Remove handling for OPT_ftree_vectorize,
4296 and leave it for the options machinery.
4298 2017-08-01 Martin Liska <mliska@suse.cz>
4301 * builtins.c (expand_builtin_memcpy_args): Remove.
4302 (expand_builtin_memcpy): Call newly added function
4303 expand_builtin_memory_copy_args.
4304 (expand_builtin_memcpy_with_bounds): Likewise.
4305 (expand_builtin_mempcpy): Remove last argument.
4306 (expand_builtin_mempcpy_with_bounds): Likewise.
4307 (expand_builtin_memory_copy_args): New function created from
4308 expand_builtin_mempcpy_args with small modifications.
4309 (expand_builtin_mempcpy_args): Remove.
4310 (expand_builtin_stpcpy): Remove unused argument.
4311 (expand_builtin): Likewise.
4312 (expand_builtin_with_bounds): Likewise.
4314 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
4317 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
4318 print "ds:" only for immediates in generic address space.
4320 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
4323 * config/i386/i386.c (ix86_funciton_naked): New prototype.
4324 (ix86_function_ok_for_sibcall): Return false for naked functions.
4326 2017-08-01 Richard Biener <rguenther@suse.de>
4328 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
4329 (compute_antic): Seed worklist with exit block predecessors.
4330 * cfganal.c (dfs_find_deadend): For a cycle return the source
4331 of the edge closing it.
4333 2017-08-01 Tamar Christina <tamar.christina@arm.com>
4335 * config/aarch64/aarch64.c
4336 (aarch64_can_const_movi_rtx_p): Move 0 check.
4338 2017-08-01 Bin Cheng <bin.cheng@arm.com>
4340 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
4341 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
4343 * match.pd: Ditto in address comparison pattern.
4345 2017-08-01 Bin Cheng <bin.cheng@arm.com>
4347 PR tree-optimization/81627
4348 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
4349 closed ssa form for store-store chain.
4351 2017-08-01 Bin Cheng <bin.cheng@arm.com>
4353 PR tree-optimization/81620
4354 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
4355 for store-store chain.
4357 2017-08-01 Jakub Jelinek <jakub@redhat.com>
4359 PR tree-optimization/81588
4360 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
4361 ranges[i].in_p, invert comparison code ccode. For >/>=,
4362 swap rhs1 and rhs2 and comparison code unconditionally,
4363 for </<= don't do that. Don't swap rhs1/rhs2 again if
4364 ranges[i].in_p, instead invert comparison code ccode if
4365 opcode or oe->rank is BIT_IOR_EXPR.
4368 * optabs.def (vec_extract_optab, vec_init_optab): Change from
4369 a direct optab to conversion optab.
4370 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
4371 with GET_MODE_INNER as last argument instead of optab_handler.
4372 * expmed.c (extract_bit_field_1): Likewise. Use vector from
4373 vector extraction if possible and optab is available.
4374 * expr.c (store_constructor): Use convert_optab_handler instead
4375 of optab_handler. Use vector initialization from smaller
4376 vectors if possible and optab is available.
4377 * tree-vect-stmts.c (vectorizable_load): Likewise.
4378 * doc/md.texi (vec_extract, vec_init): Document that the optabs
4380 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
4381 of vec_init from half-sized vectors with the same element mode.
4382 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
4383 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
4384 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
4385 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
4386 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
4387 after mode in gen_vec_extract* calls.
4388 (vec_extract<mode>): Renamed to ...
4389 (vec_extract<mode><ssescalarmodelower>): ... this.
4390 (vec_extract<mode><ssehalfvecmodelower>): New expander.
4391 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
4392 element mode after mode in gen_vec_init* calls.
4393 (VEC_INIT_HALF_MODE): New mode iterator.
4394 (vec_init<mode>): Renamed to ...
4395 (vec_init<mode><ssescalarmodelower>): ... this.
4396 (vec_init<mode><ssehalfvecmodelower>): New expander.
4397 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
4398 (vec_extractv2sfsf): ... this.
4399 (vec_initv2sf): Renamed to ...
4400 (vec_initv2sfsf): ... this.
4401 (vec_extractv2si): Renamed to ...
4402 (vec_extractv2sisi): ... this.
4403 (vec_initv2si): Renamed to ...
4404 (vec_initv2sisi): ... this.
4405 (vec_extractv4hi): Renamed to ...
4406 (vec_extractv4hihi): ... this.
4407 (vec_initv4hi): Renamed to ...
4408 (vec_initv4hihi): ... this.
4409 (vec_extractv8qi): Renamed to ...
4410 (vec_extractv8qiqi): ... this.
4411 (vec_initv8qi): Renamed to ...
4412 (vec_initv8qiqi): ... this.
4413 * config/rs6000/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/rs6000/paired.md (vec_initv2sf): Renamed to ...
4419 (vec_initv2sfsf): ... this.
4420 * config/rs6000/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, altivec_vreve<mode>2): Add
4423 element mode after mode in gen_vec_init* calls.
4424 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
4425 (vec_init<mode><Vel>): ... this.
4426 (vec_extract<mode>): Renamed to ...
4427 (vec_extract<mode><Vel>): ... this.
4428 * config/aarch64/iterators.md (Vel): New mode attribute.
4429 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
4430 Add element mode after mode in gen_vec_extract* calls.
4431 * config/s390/vector.md (non_vec_l): New mode attribute.
4432 (vec_extract<mode>): Renamed to ...
4433 (vec_extract<mode><non_vec_l>): ... this.
4434 (vec_init<mode>): Renamed to ...
4435 (vec_init<mode><non_vec_l>): ... this.
4436 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
4437 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
4439 * config/arm/iterators.md (V_elem_l): New mode attribute.
4440 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
4441 (vec_extract<mode><V_elem_l>): ... this.
4442 (vec_extractv2di): Renamed to ...
4443 (vec_extractv2didi): ... this.
4444 (vec_init<mode>): Renamed to ...
4445 (vec_init<mode><V_elem_l>): ... this.
4446 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
4447 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
4448 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
4449 Add element mode after gen_vec_extract* calls.
4450 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
4451 (vec_init<mode><unitmode>): ... this.
4452 (vec_extract<mode>): Renamed to ...
4453 (vec_extract<mode><unitmode>): ... this.
4454 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
4455 (vec_init<mode><unitmode>): ... this.
4456 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
4457 (vec_initv2sfsf): ... this.
4458 (vec_extractv2sf): Renamed to ...
4459 (vec_extractv2sfsf): ... this.
4460 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
4461 Add element mode after gen_vec_extract* calls.
4462 * config/mips/mips.md (unitmode): New mode iterator.
4463 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
4464 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
4465 * config/spu/spu.md (inner_l): New mode attribute.
4466 (vec_init<mode>): Renamed to ...
4467 (vec_init<mode><inner_l>): ... this.
4468 (vec_extract<mode>): Renamed to ...
4469 (vec_extract<mode><inner_l>): ... this.
4470 * config/sparc/sparc.md (veltmode): New mode iterator.
4471 (vec_init<VMALL:mode>): Renamed to ...
4472 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
4473 * config/ia64/vect.md (vec_initv2si): Renamed to ...
4474 (vec_initv2sisi): ... this.
4475 (vec_initv2sf): Renamed to ...
4476 (vec_initv2sfsf): ... this.
4477 (vec_extractv2sf): Renamed to ...
4478 (vec_extractv2sfsf): ... this.
4479 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
4480 (vec_init<mode>): Renamed to ...
4481 (vec_init<mode><VEC_base_l>): ... this.
4482 (vec_extract<mode>): Renamed to ...
4483 (vec_extract<mode><VEC_base_l>): ... this.
4484 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
4485 (vec_initv2sfsf): ... this.
4486 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
4487 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
4488 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
4489 gen_vec_init* calls.
4491 2017-08-01 Richard Biener <rguenther@suse.de>
4493 PR tree-optimization/81297
4494 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
4495 TREE_OVERFLOW from INTEGER_CSTs.
4497 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
4499 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
4501 2017-07-31 Carl Love <cel@us.ibm.com>
4503 * config/rs6000/rs6000-c: Add support for built-in functions
4504 vector signed char vec_xl_be (signed long long, signed char *);
4505 vector unsigned char vec_xl_be (signed long long, unsigned char *);
4506 vector signed int vec_xl_be (signed long long, signed int *);
4507 vector unsigned int vec_xl_be (signed long long, unsigned int *);
4508 vector signed long long vec_xl_be (signed long long, signed long long *);
4509 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
4510 vector signed short vec_xl_be (signed long long, signed short *);
4511 vector unsigned short vec_xl_be (signed long long, unsigned short *);
4512 vector double vec_xl_be (signed long long, double *);
4513 vector float vec_xl_be (signed long long, float *);
4514 * config/rs6000/altivec.h (vec_xl_be): Add #define.
4515 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
4516 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
4518 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
4519 (altivec_expand_builtin): Add switch statement to call
4520 altivec_expand_xl_be for each builtin.
4521 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
4522 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
4523 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
4524 __builtin_vsx_le_be_v16qi.
4525 * doc/extend.texi: Update the built-in documentation file for the
4526 new built-in functions.
4528 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
4531 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
4533 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
4535 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4537 * config.gcc: Add z14.
4538 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
4539 CPU model numbers for z13s and z14.
4540 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
4542 * config/s390/s390-opts.h (enum processor_type): Rename
4543 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
4544 * config/s390/s390.c (processor_table): Add field for CPU name to
4545 be passed to Binutils.
4546 (s390_asm_output_machine_for_arch): Use the new field in
4547 processor_table for Binutils.
4548 (s390_expand_builtin): Replace arch12 with z14.
4549 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
4550 (s390_get_sched_attrmask): Likewise.
4551 (s390_get_unit_mask): Likewise.
4552 * config/s390/s390.opt: Add z14 to processor_type enum.
4554 2017-07-31 Martin Jambor <mjambor@suse.cz>
4557 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
4558 regardless of optimization level.
4560 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
4561 Martin Liska <mliska@suse.cz>
4563 * predict.def: Remove old comment and adjust probability.
4564 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
4567 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
4570 * config/i386/i386.c (ix86_function_naked): New function.
4571 (ix86_can_use_return_insn_p): Return false for naked functions.
4572 (ix86_expand_prologue): Skip prologue for naked functions.
4573 (ix86_expand_epilogue): Skip epilogue for naked functions
4574 and emit trap instruction.
4575 (ix86_warn_func_return): New function.
4576 (ix86_attribute_table): Add "naked" attribute specification.
4577 (TARGET_WARN_FUNC_RETURN): Define.
4578 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
4580 2017-07-31 Martin Liska <mliska@suse.cz>
4582 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
4583 (dump_gimple_bb_header): Always dump BB info.
4584 (pp_cfg_jump): Do not append info about BB when dumping a jump.
4586 2017-07-31 Martin Liska <mliska@suse.cz>
4589 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
4590 also with current_function_decl non-null equality.
4592 2017-07-31 Jakub Jelinek <jakub@redhat.com>
4595 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
4596 change type to the element type, instead add eltype variable and
4597 use it where we are interested in the element type.
4599 PR tree-optimization/81603
4600 * ipa-polymorphic-call.c
4601 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
4602 offset arithmetic in offset_int, bail out if the resulting bit offset
4603 doesn't fit into shwi.
4605 2017-07-31 Martin Liska <mliska@suse.cz>
4607 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
4608 (gimplify_save_expr): Fix comment.
4610 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
4613 * config/i386/i386.c (ix86_function_arg): Update arguments for
4615 (ix86_compute_frame_layout): Set the initial stack offset to
4616 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
4617 INCOMING_FRAME_SP_OFFSET.
4618 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
4619 stack before exception handler returns.
4620 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
4621 the 'ERROR_CODE' for exception handler.
4623 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
4625 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
4626 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
4627 (ASM_OUTPUT_REG_POP): Ditto.
4628 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
4629 instead of asm_fprintf to output pure string.
4631 2017-07-29 Jakub Jelinek <jakub@redhat.com>
4633 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
4634 to imported_module_or_decl hook.
4635 (debug_nothing_tree_tree_tree_bool): Remove.
4636 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
4637 * debug.c (do_nothing_debug_hooks): Use
4638 debug_nothing_tree_tree_tree_bool_bool instead of
4639 debug_nothing_tree_tree_tree_bool.
4640 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
4641 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
4642 * sdbout.c (sdb_debug_hooks): Likewise.
4643 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
4644 (gen_namespace_die): Add DW_AT_export_symbols attribute if
4646 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
4647 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
4648 attribute, don't add anything.
4650 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4652 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
4653 (fold_build2_stat_loc): Likewise.
4654 (fold_build3_stat_loc): Likewise.
4655 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
4656 (fold_build1_loc): Remove macro.
4657 (fold_build2_loc): Likewise.
4658 (fold_build3_loc): Likewise.
4660 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4662 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
4663 (gimple_build_debug_bind_source_stat): Likewise.
4664 * gimple.h (gimple_build_debug_bind): Remove macro.
4665 (gimple_build_debug_bind_source): Likewise.
4667 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4669 * bitmap.c (bitmap_alloc): Adjust.
4670 (bitmap_gc_alloc): Likewise.
4671 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
4673 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4675 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
4676 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
4677 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
4678 (bitmap_gc_alloc_stat): Likewise.
4679 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
4681 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4683 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
4684 * rtl.h (shallow_copy_rtx): Remove macro.
4686 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4688 * emit-rtl.c (gen_raw_REG): Adjust.
4689 * gengenrtl.c (gendef): Likewise.
4690 * rtl.c (rtx_alloc_stat): Remove _stat from name.
4691 * rtl.h (rtx_alloc): Remove macro.
4693 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4695 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
4696 (build_tree_list_stat): Likewise.
4697 * tree.h (build_tree_list): Remove macro.
4698 (build_tree_list_vec): Likewise.
4700 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4702 * tree.c (make_vector_stat): Remove _stat from name.
4703 (build_vector_stat): Likewise.
4704 * tree.h (make_vector_stat): Remove macro.
4705 (build_vector_stat): Likewise.
4707 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4709 * tree.h (build_var_debug_value): Remove prototype.
4711 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4713 * tree.c (tree_cons_stat): Remove _stat from name.
4714 * tree.h (tree_cons): Remove macro.
4716 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4718 * tree.c (build_vl_exp_stat): Remove _stat from name.
4719 * tree.h (build_vl_exp): Remove macro.
4721 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4723 * tree.c (build_decl_stat): Remove _stat from name.
4724 * tree.h (build_decl): Remove macro.
4726 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4728 * gimple.c (gimple_build_with_ops_stat): Adjust.
4729 (gimple_alloc_stat): Remove _stat from name.
4730 * gimple.h (gimple_alloc): Remove macro.
4732 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4734 * tree.c (make_tree_vec_stat): Remove _stat from name.
4735 (grow_tree_vec_stat): Likewise.
4736 * tree.h (make_tree_vec_stat): Adjust prototype.
4737 (grow_tree_vec_stat): Likewise.
4738 (make_tree_vec): Remove macro.
4739 (grow_tree_vec): Likewise.
4741 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4743 * fold-const.c (fold_build1_stat_loc): Adjust.
4744 (fold_build2_stat_loc): Likewise.
4745 (fold_build3_stat_loc): Likewise.
4746 * tree.c (build0_stat): Remove _stat from name.
4747 (build1_stat): Likewise.
4748 (build2_stat): Likewise.
4749 (build3_stat): Likewise.
4750 (build4_stat): Likewise.
4751 (build5_stat): Likewise.
4752 * tree.h (build1_loc): Remove macro, and rename _stat function
4754 (build2_loc): Likewise.
4755 (build3_loc): Likewise.
4756 (build4_loc): Likewise.
4757 (build5_loc): Likewise.
4759 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4761 * tree.c (make_int_cst_stat): Remove _stat from name.
4762 * tree.h (make_int_cst_stat): Adjust prototype.
4763 (make_int_cst): Remove macro.
4765 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4767 * tree.c (make_tre_binfo_stat): Remove _stat from name.
4768 * tree.h (make_tree_binfo_stat): Adjust prototype.
4769 (make_tree_binfo): Remove.
4771 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4773 * tree.c (copy_node_stat): Rename to copy_node.
4774 (build_distinct_type_copy): Adjust.
4775 * tree.h (copy_node_stat): Adjust prototype.
4776 (copy_node): Remove macro.
4778 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4780 * tree.c (make_node_stat): rename to make_node.
4781 (build_tree_list_stat): Adjust.
4782 (build0_stat): Likewise.
4783 (build2_stat): Likewise.
4784 (build3_stat): Likewise.
4785 (build4_stat): Likewise.
4786 (build5_stat): Likewise.
4787 (build_decl_stat): Likewise.
4788 * tree.h (make_node_stat): Adjust prototype.
4789 (make_node): remove macro.
4791 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
4793 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
4794 (PPC_FEATURE2_SCV): Likewise.
4795 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
4797 2017-07-28 Tamar Christina <tamar.christina@arm.com>
4799 * config/aarch64/aarch64.c
4800 (aarch64_internal_mov_immediate): Add new special pattern.
4801 * config/aarch64/aarch64.md (*movdi_aarch64):
4802 Add reg/32bit const mov case.
4804 2017-07-28 Tamar Christina <tamar.christina@arm.com>
4805 Richard Sandiford <richard.sandiford@linaro.org>
4807 * config/aarch64/aarch64.md (mov<mode>): Generalize.
4808 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
4809 Add integer and movi cases.
4810 (movi-split-hf-df-sf split, fp16): New.
4811 (enabled): Added TARGET_FP_F16INST.
4812 * config/aarch64/iterators.md (GPF_HF): New.
4813 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
4815 2017-07-28 Tamar Christina <tamar.christina@arm.com>
4817 * config/aarch64/aarch64.c
4818 (aarch64_simd_container_mode): Add prototype.
4819 (aarch64_expand_mov_immediate): Add HI support.
4820 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
4821 (aarch64_can_const_movi_rtx_p): New.
4822 (aarch64_preferred_reload_class):
4823 Remove restrictions of using FP registers for certain SIMD operations.
4824 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
4825 (aarch64_valid_floating_const): Add integer move validation.
4826 (aarch64_simd_imm_scalar_p): Remove.
4827 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
4828 (aarch64_legitimate_constant_p): Expand list of supported cases.
4829 * config/aarch64/aarch64-protos.h
4830 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
4831 (aarch64_reinterpret_float_as_int): New.
4832 (aarch64_simd_imm_scalar_p): Remove.
4833 * config/aarch64/constraints.md (Uvi): New.
4834 (Dd): Split into Ds and new Dd.
4835 * config/aarch64/aarch64.md (*movsi_aarch64):
4837 (*movdi_aarch64): Add SIMD mov case.
4839 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4841 * tree-predcom.c: (struct chain): Handle store-store chain in which
4842 stores for elimination only store loop invariant values.
4843 (execute_pred_commoning_chain): Ditto.
4844 (prepare_initializers_chain_store_elim): Ditto.
4845 (prepare_finalizers): Ditto.
4846 (is_inv_store_elimination_chain): New function.
4847 (initialize_root_vars_store_elim_1): New function.
4849 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4851 * tree-predcom.c: Revise general description of the pass.
4852 (enum chain_type): New enum type for store elimination.
4853 (struct chain): New field supporting store elimination.
4854 (struct component): Ditto.
4855 (dump_chain): Dump store-stores chain.
4856 (release_chain): Release resources.
4857 (split_data_refs_to_components): Compute and create component
4858 contains only stores for elimination.
4859 (get_chain_last_ref_at): New function.
4860 (make_invariant_chain): Initialization.
4861 (make_rooted_chain): Specify chain type in parameter and record it.
4862 (add_looparound_copies): Skip for store-stores chain.
4863 (determine_roots_comp): Compute type of chain and pass it to
4865 (initialize_root_vars_store_elim_2): New function.
4866 (finalize_eliminated_stores): New function.
4867 (remove_stmt): Handle store for elimination.
4868 (execute_pred_commoning_chain): Execute predictive commoning on
4870 (determine_unroll_factor): Skip unroll for store-stores chain.
4871 (prepare_initializers_chain_store_elim): New function.
4872 (prepare_initializers_chain): Hanlde store-store chain.
4873 (prepare_finalizers_chain, prepare_finalizers): New function.
4874 (tree_predictive_commoning_loop): Return integer value indicating
4875 if loop is unrolled or lcssa form is corrupted.
4876 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
4878 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4880 * tree-predcom.c (initialize_root): Delete.
4881 (execute_pred_commoning_chain): Initialize root vars and replace
4882 reference of non-combined chain directly, rather than call above
4885 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4887 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
4888 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
4890 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4892 * tree-predcom.c (struct chain): New field init_seq.
4893 (release_chain): Release init_seq.
4894 (prepare_initializers_chain): Record intialization stmts in above
4896 (insert_init_seqs): New function.
4897 (tree_predictive_commoning_loop): Call insert_init_seqs.
4899 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4901 * tree-predcom.c (determine_roots_comp): Skip trivial components.
4903 2017-07-28 Richard Biener <rguenther@suse.de>
4905 * match.pd: Remove superfluous :c.
4906 * genmatch.c (simplify::id): Add member.
4907 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
4909 (current_id): New global.
4910 (dt_node::parent): Move from ...
4911 (dt_operand::parent): ... here. Add for_id member.
4912 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
4913 (decision_tree::find_node): Relax order requirement when
4914 merging DT_TRUE nodes to ones inbetween the current simplify
4915 and the one we try to merge with. Add diagnostic whenever
4916 we need to enforce pattern order by not merging.
4917 (decision_tree::insert): Set current_id.
4918 (decision_tree::print_node): Dump parent node and for_id.
4919 (parser::last_id): Add member.
4920 (parser::push_simplify): Assign unique id.
4921 (parser::parser): Initialize last_id.
4923 2017-07-28 Martin Liska <mliska@suse.cz>
4926 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
4927 gimple_build_debug_bind.
4929 2017-07-28 Richard Biener <rguenther@suse.de>
4931 PR tree-optimization/81502
4932 * match.pd: Add pattern combining BIT_INSERT_EXPR with
4934 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
4936 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
4937 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
4938 for BIT_FIELD_REF args.
4939 * fold-const.c (make_bit_field_ref): Likewise.
4940 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
4942 2017-07-28 Jakub Jelinek <jakub@redhat.com>
4945 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
4946 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
4947 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
4948 Or it into SANITIZER_UNDEFINED.
4949 * ubsan.c: Include gimple-fold.h and varasm.h.
4950 (ubsan_expand_ptr_ifn): New function.
4951 (instrument_pointer_overflow): New function.
4952 (maybe_instrument_pointer_overflow): New function.
4953 (instrument_object_size): Formatting fix.
4954 (pass_ubsan::execute): Call instrument_pointer_overflow
4955 and maybe_instrument_pointer_overflow.
4956 * internal-fn.c (expand_UBSAN_PTR): New function.
4957 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
4958 * sanitizer.def (__ubsan_handle_pointer_overflow,
4959 __ubsan_handle_pointer_overflow_abort): New builtins.
4960 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
4961 * internal-fn.def (UBSAN_PTR): New internal function.
4962 * opts.c (sanitizer_opts): Add pointer-overflow.
4963 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
4964 * fold-const.c (build_range_check): Compute pointer range check in
4965 integral type if pointer arithmetics would be needed. Formatting
4968 2017-07-28 Martin Liska <mliska@suse.cz>
4971 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
4972 parameters that are of a variable-length.
4974 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
4976 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
4978 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
4979 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
4980 (CRT_CALL_STATIC_FUNCTION): Likewise.
4981 (ASM_DEFAULT_SPEC): New define.
4982 (ASM_SPEC32): Likewise.
4983 (ASM_SPEC64): Likewise.
4984 (ASM_SPEC_COMMON): Likewise.
4985 (ASM_SPEC): Likewise.
4986 (INVALID_64BIT): Likewise.
4987 (LINK_OS_DEFAULT_SPEC): Likewise.
4988 (LINK_OS_SPEC32): Likewise.
4989 (LINK_OS_SPEC64): Likewise.
4990 (POWERPC_LINUX): Likewise.
4991 (PTRDIFF_TYPE): Likewise.
4992 (RESTORE_FP_PREFIX): Likewise.
4993 (RESTORE_FP_SUFFIX): Likewise.
4994 (SAVE_FP_PREFIX): Likewise.
4995 (SAVE_FP_SUFFIX): Likewise.
4996 (SIZE_TYPE): Likewise.
4997 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
4998 (TARGET_64BIT): Likewise.
4999 (TARGET_64BIT): Likewise.
5000 (TARGET_AIX): Likewise.
5001 (WCHAR_TYPE_SIZE): Likewise.
5002 (WCHAR_TYPE): Undefine.
5003 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
5004 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
5005 (CPP_OS_RTEMS_SPEC): Delete.
5006 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
5007 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
5009 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
5011 2017-07-28 Jakub Jelinek <jakub@redhat.com>
5013 PR tree-optimization/81578
5014 * tree-parloops.c (build_new_reduction): Bail out if
5015 reduction_code isn't one of the standard OpenMP reductions.
5016 Move the details printing after that decision.
5018 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
5020 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
5021 related to reload_in_progress.
5022 (splat_input_operand): Likewise.
5023 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
5025 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
5027 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
5028 (TARGET_INSTANTIATE_DECLS): Likewise.
5029 (legitimate_indexed_address_p): Delete reload_in_progress code.
5030 (rs6000_debug_legitimate_address_p): Likewise.
5031 (rs6000_eliminate_indexed_memrefs): Likewise.
5032 (rs6000_emit_le_vsx_store): Likewise.
5033 (rs6000_emit_move_si_sf_subreg): Likewise.
5034 (rs6000_emit_move): Likewise.
5035 (register_to_reg_type): Likewise.
5036 (rs6000_pre_atomic_barrier): Likewise.
5037 (rs6000_machopic_legitimize_pic_address): Likewise.
5038 (rs6000_allocate_stack_temp): Likewise.
5039 (rs6000_address_for_fpconvert): Likewise.
5040 (rs6000_address_for_altivec): Likewise.
5041 (rs6000_secondary_memory_needed_rtx): Delete function.
5042 (rs6000_check_sdmode): Likewise.
5043 (rs6000_alloc_sdmode_stack_slot): Likewise.
5044 (rs6000_instantiate_decls): Likewise.
5045 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
5046 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
5047 Delete reload_in_progress.
5048 (*vec_reload_and_plus_<mptrsize>): Likewise.
5049 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
5050 (vsx_div_v2di): Likewise.
5051 (vsx_udiv_v2di): Likewise.
5053 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
5055 * config/rs6000/rs6000.opt (mlra): Replace with stub.
5056 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
5057 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
5058 (rs6000_debug_reg_global): Delete print of LRA status.
5059 (rs6000_option_override_internal): Delete dead LRA related code.
5060 (rs6000_lra_p): Delete function.
5061 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
5063 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
5065 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
5066 * config/riscv/rtems.h: New file.
5068 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5069 Sudakshina Das <sudi.das@arm.com>
5071 * config/aarch64/aarch64.md
5072 (define_split for and<mode>3nr_compare): Move
5073 non aarch64_logical_operand to a register.
5074 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
5075 register immediate operand to a register.
5076 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
5078 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
5081 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
5083 2017-07-27 Richard Biener <rguenther@suse.de>
5085 PR tree-optimization/81573
5086 PR tree-optimization/81494
5087 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
5088 multi defuse cycle case.
5090 2017-07-27 Richard Biener <rguenther@suse.de>
5092 PR tree-optimization/81571
5093 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
5096 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
5098 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
5099 earlier and only if MASK_FPU is set. Adjust formatting.
5101 2017-07-27 Martin Liska <mliska@suse.cz>
5103 * opt-functions.awk: Add validation of value of Init.
5104 * optc-gen.awk: Pass new argument.
5106 2017-07-27 Martin Liska <mliska@suse.cz>
5108 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
5109 Fix wrong condition.
5111 2017-07-27 Martin Liska <mliska@suse.cz>
5113 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
5114 BBs and edges seen by autoFDO.
5116 2017-07-27 Richard Biener <rguenther@suse.de>
5118 PR tree-optimization/81502
5119 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
5120 with incompatible but same sized type.
5121 (execute_update_addresses_taken): Likewise.
5123 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
5125 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
5126 flag_tree_loop_vectorize rather than flag_tree_vectorize.
5128 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5131 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
5132 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
5133 Change s_operand to memory_operand.
5135 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
5137 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
5138 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
5139 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
5140 Emit instructions rather than returning an expression. Handle TFmode
5141 and KFmode by casting to TImode.
5142 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
5143 (rs6000_emit_le_vsx_store): Likewise.
5144 * config/rs6000/vsx.md (VSX_TI): New iterator.
5145 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
5146 (*vsx_le_undo_permute_<mode>): Likewise.
5147 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
5148 emit the split sequence.
5149 (*vsx_le_perm_store_<mode>): Likewise.
5151 2017-07-27 Jakub Jelinek <jakub@redhat.com>
5153 PR tree-optimization/81555
5154 PR tree-optimization/81556
5155 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
5156 if true, force CHANGED for the recursive invocation.
5157 (reassociate_bb): Remember original length of ops array, pass
5158 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
5160 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
5161 attributes for noipa attribute. For naked attribute use
5162 lookup_attribute first before lookup_attribute_spec.
5163 * final.c (rest_of_handle_final): Disable IPA RA for functions with
5165 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
5166 for functions with noipa attribute.
5167 (cgraph_externally_visible_p): Return true for functions with noipa
5169 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
5170 for functions with noipa attribute.
5171 * doc/extend.texi: Document noipa function attribute.
5172 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
5173 also for functions with noipa attribute.
5174 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
5176 2017-07-26 Andrew Pinski <apinski@cavium.com>
5178 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
5179 vec_unalign_load_cost and vec_unalign_store_cost.
5181 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5183 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
5184 -mvsx-small-integer option.
5185 (ISA_3_0_MASKS_IEEE): Likewise.
5186 (OTHER_VSX_VECTOR_MASKS): Likewise.
5187 (POWERPC_MASKS): Likewise.
5188 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
5189 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
5190 code, only testing for DImode being allowed in non-VSX floating
5192 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
5193 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
5195 (rs6000_option_override_internal): Delete -mvsx-small-integer.
5196 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
5197 TARGET_P8_VECTOR test.
5198 (rs6000_secondary_reload_simple_move): Likewise.
5199 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
5200 since TARGET_P9_VECTOR was already tested.
5201 (rs6000_opt_masks): Remove -mvsx-small-integer.
5202 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
5203 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
5205 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
5206 test for TARGET_VEXTRACTUB was used, and that uses
5208 (p9 extract splitter): Likewise.
5209 (vsx_extract_<mode>_di_p9): Likewise.
5210 (vsx_extract_<mode>_store_p9): Likewise.
5211 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
5212 for TARGET_P9_VECTOR was used. Delete code that is now dead with
5213 the elimination of TARGET_VSX_SMALL_INTEGER.
5214 (vsx_extract_<mode>_p8): Likewise.
5215 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
5216 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
5217 (vsx_set_<mode>_p9): Likewise.
5218 (vsx_set_v4sf_p9): Likewise.
5219 (vsx_set_v4sf_p9_zero): Likewise.
5220 (vsx_insert_extract_v4sf_p9): Likewise.
5221 (vsx_insert_extract_v4sf_p9_2): Likewise.
5222 * config/rs6000/rs6000.md (sign extend splitter): Change
5223 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
5224 (floatsi<mode>2_lfiwax_mem): Likewise.
5225 (floatunssi<mode>2_lfiwzx_mem): Likewise.
5226 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
5227 since a test for TARGET_P9_VECTOR was used.
5228 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
5229 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
5230 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
5231 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
5232 TARGET_P8_VECTOR test.
5233 (fix_trunc<mode>si2_stfiwx): Likewise.
5234 (fix_trunc<mode>si2_internal): Likewise.
5235 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
5236 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
5238 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
5239 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
5240 TARGET_P8_VECTOR test.
5241 (fixuns_trunc<mode>si2_stfiwx): Likewise.
5242 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
5243 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
5245 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
5246 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
5247 since a test for TARGET_P9_VECTOR was used.
5248 (splitter for loading small constants): Likewise.
5250 2017-07-26 Andrew Pinski <apinski@cavium.com>
5252 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
5255 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
5258 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
5259 (fp_valid_at): Likewise.
5261 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
5263 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
5264 (qdf24xx_addrcost_table): Likewise.
5265 (cortexa57_tunings): Update to use generic_branch_cost.
5266 (cortexa72_tunings): Likewise.
5267 (cortexa73_tunings): Likewise.
5268 (qdf24xx_tunings): Likewise.
5270 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
5272 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
5273 (thunderx2t99_branch_cost): Likewise.
5274 (cortexa35_tunings): Update to use generic_branch_cost.
5275 (cortexa53_tunings): Likewise.
5276 (cortexa57_tunings): Likewise.
5277 (cortexa72_tunings): Likewise.
5278 (cortexa73_tunings): Likewise.
5279 (thunderx2t99_tunings): Likewise.
5281 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
5283 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
5284 (sparc_option_override): Honour MASK_FSMULD.
5285 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
5286 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
5287 * config/sparc/sparc.opt (mfsmuld): New option.
5288 * doc/invoke.texi (mfsmuld): Document option.
5290 2017-07-26 Marek Polacek <polacek@redhat.com>
5293 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
5295 2017-07-26 Richard Biener <rguenther@suse.de>
5297 * gimple-match-head.c (do_valueize): Return OP if valueize
5299 (get_def): New helper to get at the def stmt of a SSA name
5301 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
5302 do_valueize to get at the def stmt.
5303 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
5305 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
5308 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
5310 2017-07-26 Martin Liska <mliska@suse.cz>
5313 * function.c (expand_function_start): Make expansion of
5314 nonlocal_goto_save_area after parm_birth_insn.
5316 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
5318 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
5319 from all CPU target flags enable members.
5321 2017-07-26 Richard Biener <rguenther@suse.de>
5323 * genmatch.c (dt_simplify::gen): Make iterator vars const.
5324 (decision_tree::gen): Make 'type' const.
5325 (write_predicate): Likewise.
5327 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
5329 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
5330 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
5331 (rs6000_option_override_internal): Likewise.
5332 (rs6000_expand_vector_set): Likewise.
5333 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
5334 (TARGET_UPPER_REGS_SF): Likewise.
5335 (TARGET_UPPER_REGS_DI): Likewise.
5336 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
5337 (TARGET_DIRECT_MOVE_64BIT): Likewise.
5338 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
5339 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
5340 (Splitters for DI constants in Altivec registers): Likewise.
5341 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
5342 (vsx_set_v4sf_p9): Likewise.
5343 (vsx_set_v4sf_p9_zero): Likewise.
5344 (vsx_insert_extract_v4sf_p9): Likewise.
5345 (vsx_insert_extract_v4sf_p9_2): Likewise.
5347 2017-07-25 Carl Love <cel@us.ibm.com>
5349 * doc/extend.texi: Update the built-in documentation file for the
5350 existing built-in functions
5351 vector signed char vec_cnttz (vector signed char);
5352 vector unsigned char vec_cnttz (vector unsigned char);
5353 vector signed short vec_cnttz (vector signed short);
5354 vector unsigned short vec_cnttz (vector unsigned short);
5355 vector signed int vec_cnttz (vector signed int);
5356 vector unsigned int vec_cnttz (vector unsigned int);
5357 vector signed long long vec_cnttz (vector signed long long);
5358 vector unsigned long long vec_cnttz (vector unsigned long long);
5360 2017-07-25 Andrew Pinski <apinski@cavium.com>
5362 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
5363 accesses where the use is for the first operand of a BIT_INSERT.
5365 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
5368 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
5369 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
5371 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
5373 * config/i386/gstabs.h: Delete.
5374 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
5376 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
5378 * config/i386/i386.c (ix86_decompose_address): Do not check for
5379 register RTX when looking at index_reg or base_reg.
5380 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
5382 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
5384 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
5385 to update EH info here.
5387 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
5389 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
5391 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
5393 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
5395 2017-07-25 Torsten Duwe <duwe@suse.de>
5397 * common.opt: Introduce -fpatchable-function-entry
5398 command line option, and its variables function_entry_patch_area_size
5399 and function_entry_patch_area_start.
5400 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
5401 including a two-value parser.
5402 * target.def (print_patchable_function_entry): New target hook.
5403 * targhooks.h (default_print_patchable_function_entry): New function.
5404 * targhooks.c (default_print_patchable_function_entry): Likewise.
5405 * toplev.c (process_options): Switch off IPA-RA if
5406 patchable function entries are being generated.
5407 * varasm.c (assemble_start_function): Look at the
5408 patchable-function-entry command line switch and current
5409 function attributes and maybe generate NOP instructions by
5410 calling the print_patchable_function_entry hook.
5411 * doc/extend.texi: Document patchable_function_entry attribute.
5412 * doc/invoke.texi: Document -fpatchable_function_entry
5413 command line option.
5414 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
5416 * doc/tm.texi: Re-generate.
5418 2017-07-25 Jakub Jelinek <jakub@redhat.com>
5421 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
5422 TARGET_AVX512DQ rather than TARGET_AVX512BW.
5424 2017-07-25 Tamar Christina <tamar.christina@arm.com>
5426 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
5428 2017-07-25 Richard Biener <rguenther@suse.de>
5430 PR tree-optimization/81455
5431 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
5432 not walk in cycles when looking for guards.
5434 2017-07-25 Richard Biener <rguenther@suse.de>
5436 PR tree-optimization/81529
5437 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
5438 when optimizing backedge uses.
5440 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
5442 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
5444 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
5445 to dl_section_ref. On AIX, append an expression to subtract
5446 the size of the section length to dl_section_ref.
5448 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
5450 * configure.ac: If any of the config.* scripts fail, exit 1.
5451 * configure: Regenerate.
5453 2017-07-25 Richard Biener <rguenther@suse.de>
5456 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
5457 of immediate uses, be more verbose on errors.
5459 2017-07-25 Richard Biener <rguenther@suse.de>
5461 PR tree-optimization/81510
5462 * tree-vect-loop.c (vect_is_simple_reduction): When the
5463 reduction stmt is not inside the loop bail out.
5465 2017-07-25 Richard Biener <rguenther@suse.de>
5467 PR tree-optimization/81303
5468 * tree-vect-loop-manip.c (vect_loop_versioning): Build
5469 profitability check against LOOP_VINFO_NITERSM1.
5471 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
5473 * domwalk.c (cmp_bb_postorder): Simplify.
5474 (sort_bbs_postorder): New function. Use it...
5475 (dom_walker::walk): ...here to optimize common cases.
5477 2017-07-25 Martin Liska <mliska@suse.cz>
5480 * ipa-visibility.c (function_and_variable_visibility): Make the
5481 redirection just on target that supports aliasing.
5482 Fix GNU coding style.
5484 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
5487 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
5489 * config/arm/rtems.h: Likewise.
5490 * config/bfin/rtems.h: Likewise.
5491 * config/i386/rtemself.h: Likewise.
5492 * config/lm32/rtems.h: Likewise.
5493 * config/m32c/rtems.h: Likewise.
5494 * config/m68k/rtemself.h: Likewise.
5495 * config/microblaze/rtems.h: Likewise.
5496 * config/mips/rtems.h: Likewise.
5497 * config/moxie/rtems.h: Likewise.
5498 * config/nios2/rtems.h: Likewise.
5499 * config/powerpcspe/rtems.h: Likewise.
5500 * config/rs6000/rtems.h: Likewise.
5501 * config/rtems.h: Likewise.
5502 * config/sh/rtems.h: Likewise.
5503 * config/sh/rtemself.h: Likewise.
5504 * config/sparc/rtemself.h: Likewise.
5506 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
5509 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
5510 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
5511 * tree-ssa-structalias.c (alias_get_name): Same.
5513 2017-07-25 Bin Cheng <bin.cheng@arm.com>
5516 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
5517 instructions if no du chain is found.
5519 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
5521 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
5523 2017-07-25 Richard Biener <rguenther@suse.de>
5526 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
5529 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
5531 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
5533 (ISA_2_7_MASKS_SERVER): Likewise.
5534 (ISA_3_0_MASKS_IEEE): Likewise.
5535 (OTHER_P8_VECTOR_MASKS): Likewise.
5536 (OTHER_VSX_VECTOR_MASKS): Likewise.
5537 (POWERPC_MASKS): Likewise.
5538 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
5539 duplicate list of options.
5540 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
5541 explicit -mupper-regs options.
5542 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
5543 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
5544 alias for -mupper-regs-df.
5545 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
5546 (rs6000_init_hard_regno_mode_ok): Likewise.
5547 (rs6000_option_override_internal): Likewise.
5548 (rs6000_opt_masks): Likewise.
5549 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
5550 options in terms of whether -mvsx or -mpower8-vector was used.
5551 (TARGET_UPPER_REGS_DI): Likewise.
5552 (TARGET_UPPER_REGS_SF): Likewise.
5553 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
5554 -mupper-regs-* options.
5556 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
5558 * passes.c (emergency_dump_function): Print some empty lines and a
5559 header before the RTL dump.
5561 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
5563 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
5565 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
5568 * config/aarch64/aarch64.c (aarch64_classify_symbol):
5569 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
5571 2017-07-24 Carl Love <cel@us.ibm.com>
5573 * config/rs6000/rs6000-c.c: Add support for built-in functions
5574 vector float vec_extract_fp32_from_shorth (vector unsigned short);
5575 vector float vec_extract_fp32_from_shortl (vector unsigned short);
5576 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
5577 vec_extract_fp_from_shortl): Add defines for the two builtins.
5578 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
5579 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
5581 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
5582 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
5583 * doc/extend.texi: Update the built-in documentation file for the
5584 new built-in function.
5586 2017-07-24 Jakub Jelinek <jakub@redhat.com>
5589 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
5591 * doc/generic.texi: Likewise.
5592 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
5593 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
5595 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
5597 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
5598 (aarch64_mls_elt_merge<mode>): Likewise.
5600 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
5602 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
5603 having __cxa_atexit.
5605 2017-07-23 Michael Collison <michael.collison@arm.com>
5607 * config/arm/arm.c (arm_option_override): Deprecate
5608 use of -mstructure-size-boundary.
5609 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
5610 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
5612 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5615 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
5616 Reduce cost estimate for direct moves.
5618 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
5621 * config/i386/i386.c (ix86_option_override_internal): Disable
5622 BMI, BMI2 and TBM instructions for -m16.
5624 2017-07-21 Carl Love <cel@us.ibm.com>
5626 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5627 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5628 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5629 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5630 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5631 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
5632 VMULOSW): New enum "unspec" values.
5633 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
5634 altivec_vmulosw): New patterns.
5635 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5636 VMULOSW): Add definitions.
5638 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
5640 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
5641 (qdf24xx): Likewise.
5642 * config/aarch64/aarch64-options-extensions.def (rdma); New.
5643 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
5644 (AARCH64_FL_V8_1): Renumber.
5645 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
5646 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
5647 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
5648 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
5649 rdma to feature modifiers list.
5651 2017-07-21 Yury Gribov <tetra2005@gmail.com>
5654 * ipa-visibility (function_and_variable_visibility): Convert
5655 recursive PLT call to direct call if appropriate.
5657 2017-07-21 Andrew Pinski <apinski@cavium.com>
5659 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
5660 operand 1 to see if the types precision matches.
5661 * fold-const.c (operand_equal_p): Likewise.
5663 2017-07-21 Richard Biener <rguenther@suse.de>
5665 PR tree-optimization/81303
5666 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
5667 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
5668 (vect_peeling_hash_get_lowest_cost): Adjust.
5669 (vect_enhance_data_refs_alignment): Likewise. Use
5670 vect_get_peeling_costs_all_drs to compute the penalty for no
5671 peeling to match up costs.
5673 2017-07-21 Richard Biener <rguenther@suse.de>
5675 PR tree-optimization/81500
5676 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
5677 we didn't identify a reduction path.
5679 2017-07-21 Tom de Vries <tom@codesourcery.com>
5680 Cesar Philippidis <cesar@codesourcery.com>
5682 PR gcov-profile/81442
5683 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
5686 2017-07-21 Tom de Vries <tom@codesourcery.com>
5689 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
5691 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
5692 nvptx_override_options_after_change.
5694 2017-07-21 Ulrich Drepper <drepper@redhat.com>
5696 * dwarf2out.c (output_file_names): Avoid double testing for
5699 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
5701 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
5703 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
5705 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
5707 (try_crossjump_to_edge): Do not punt on partitioned functions.
5709 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
5711 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
5712 Put all BBs reachable only via paths crossing cold region to cold
5714 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
5716 2016-07-21 Richard Biener <rguenther@suse.de>
5718 PR tree-optimization/81303
5719 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
5720 into account prologue and epilogue iterations when raising
5721 min_profitable_iters to sth at least covering one vector iteration.
5723 2017-07-21 Tamar Christina <tamar.christina@arm.com>
5725 * config/arm/arm.c (arm_test_cpu_arch_dat):
5728 2017-07-20 Nathan Sidwell <nathan@acm.org>
5730 Remove TYPE_METHODS.
5731 * tree.h (TYPE_METHODS): Delete.
5732 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
5733 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
5734 (dbxout_type_methods): Scan TYPE_FIELDS.
5735 (dbxout_type): Don't check TYPE_METHODS here.
5736 * function.c (use_register_for_decl): Always ignore register for
5737 class types when not optimizing.
5738 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
5739 * tree.c (free_lang_data_in_type): Stitch out member functions and
5740 templates from TYPE_FIELDS.
5741 (build_distinct_type_copy, verify_type_variant,
5742 verify_type): Member fns are on TYPE_FIELDS.
5743 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
5744 * tree-pretty-print.c (dump_generic_node): Likewise.
5746 2017-07-20 Jakub Jelinek <jakub@redhat.com>
5749 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
5750 V2TImode and V4TImode.
5751 (ix86_expand_vector_extract): Likewise.
5752 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
5753 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
5754 (ssescalarmode): Handle V4TImode and V2TImode.
5755 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
5756 (*vec_extractv2ti, *vec_extractv4ti): New insns.
5757 (VEXTRACTI128_MODE): New mode iterator.
5758 (splitter for *vec_extractv?ti first element): New.
5759 (VEC_INIT_MODE): New mode iterator.
5760 (vec_init<mode>): Consolidate 3 expanders into one using
5761 VEC_INIT_MODE mode iterator.
5763 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
5765 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
5766 non_spilled_static_chain_regno_p.
5768 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
5770 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
5772 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
5774 * bb-reorder.c (connect_traces): Allow copying of blocks within
5777 2017-07-20 Richard Biener <rguenther@suse.de>
5779 * gimple.h (gimple_phi_result): Add gphi * overload.
5780 (gimple_phi_result_ptr): Likewise.
5781 (gimple_phi_arg): Likewise. Adjust index assert to only
5782 allow actual argument accesses rather than all slots available
5784 (gimple_phi_arg_def): Add gphi * overload.
5785 * tree-phinodes.c (make_phi_node): Initialize only actual
5787 (resize_phi_node): Clear memory not covered by old node,
5788 do not initialize excess argument slots.
5789 (reserve_phi_args_for_new_edge): Initialize new argument slot
5792 2017-07-20 Bin Cheng <bin.cheng@arm.com>
5794 PR tree-optimization/81388
5796 2016-07-21 Bin Cheng <bin.cheng@arm.com>
5798 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
5799 by removing computation of may_be_zero.
5801 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5802 Tom de Vries <tom@codesourcery.com>
5805 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
5806 when gimple level profile disagrees with what RTL expander did.
5808 2017-07-20 Richard Biener <rguenther@suse.de>
5810 PR tree-optimization/61171
5811 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
5812 (vect_analyze_stmt): Add slp instance parameter.
5813 (vectorizable_reduction): Likewise.
5814 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
5815 (vect_is_simple_reduction): Deal with chains not detected
5816 as SLP reduction chain, specifically not properly associated
5817 chains containing a mix of plus/minus.
5818 (get_reduction_op): Remove.
5819 (get_initial_defs_for_reduction): Simplify, pass in whether
5820 this is a reduction chain, pass in the SLP node for the PHIs.
5821 (vect_create_epilog_for_reduction): Get the SLP instance as
5823 (vectorizable_reduction): Get the SLP instance as arg.
5824 During analysis remember the SLP node with the PHIs in the
5825 instance. Simplify getting at the vectorized reduction PHIs.
5826 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
5827 through SLP instance.
5828 (vect_slp_analyze_operations): Likewise.
5829 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
5830 (vect_transform_stmt): Likewise.
5832 2017-07-20 Tom de Vries <tom@codesourcery.com>
5834 PR tree-optimization/81489
5835 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
5836 read of phi arg location to before loop that modifies phi.
5838 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
5840 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
5843 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
5846 * except.c (execute): Fix ordering issue.
5848 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
5850 PR rtl-optimization/81423
5851 * combine.c (make_compound_operation_int): Don't try to optimize
5852 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
5854 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
5856 PR rtl-optimization/81423
5857 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
5858 with a constant that is -1 in the truncated to mode.
5860 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
5862 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
5863 (determine_unlikely_bbs): ... here.
5864 * predict.h (propagate_unlikely_bbs_forward): Declare.
5865 * cfgexpand.c (pass_expand::execute): Use it.
5866 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
5868 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
5869 propagate_unlikely_bbs_forward.
5871 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
5874 * except.c (maybe_add_nop_after_section_switch): New function.
5877 2017-07-19 Tom de Vries <tom@codesourcery.com>
5879 * gimple.h (gimple_phi_set_arg): Make assert more strict.
5881 2017-07-19 Tom de Vries <tom@codesourcery.com>
5883 * gimple.h (gimple_phi_arg): Make assert more strict.
5885 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
5887 * config.gcc (powerpc*-*-*): Add mmintrin.h.
5888 * config/rs6000/mmintrin.h: New file.
5889 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
5891 2017-07-19 Jakub Jelinek <jakub@redhat.com>
5893 PR tree-optimization/81346
5894 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
5896 2017-07-19 Tom de Vries <tom@codesourcery.com>
5898 * config/nvptx/nvptx.md (VECIM): Add V2DI.
5900 2017-07-19 Tom de Vries <tom@codesourcery.com>
5902 * config/nvptx/nvptx-modes.def: Add V2DImode.
5903 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
5904 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
5905 (nvptx_output_mov_insn): Handle lack of mov.b128.
5906 (nvptx_print_operand): Handle 'H' and 'L' codes.
5907 (nvptx_vector_mode_supported): Allow V2DImode.
5908 (nvptx_preferred_simd_mode): New function.
5909 (nvptx_data_alignment): New function.
5910 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
5911 nvptx_preferred_simd_mode.
5912 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
5914 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
5916 2017-07-19 Tom de Vries <tom@codesourcery.com>
5918 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
5919 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
5920 (nvptx_vector_mode_supported): New function. Allow V2SImode.
5921 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
5922 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
5923 (mov<VECIM>_insn): New define_insn.
5924 (define_expand "mov<VECIM>): New define_expand.
5926 2017-07-19 Tom de Vries <tom@codesourcery.com>
5928 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
5930 2017-07-19 Jakub Jelinek <jakub@redhat.com>
5932 PR tree-optimization/81346
5933 * fold-const.h (fold_div_compare, range_check_type): Declare.
5934 * fold-const.c (range_check_type): New function.
5935 (build_range_check): Use range_check_type.
5936 (fold_div_compare): No longer static, rewritten into
5937 a match.pd helper function.
5938 (fold_comparison): Don't call fold_div_compare here.
5939 * match.pd (X / C1 op C2): New optimization using fold_div_compare
5942 2017-07-19 Nathan Sidwell <nathan@acm.org>
5944 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
5945 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
5946 * tree.c (find_decls_types_r, verify_type): Use
5947 TYPE_{MIN,MAX}_VALUE_RAW.
5948 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
5949 (hash_tree): Likewise.
5950 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
5952 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
5955 2017-07-18 Tom de Vries <tom@codesourcery.com>
5958 * omp-expand.c (expand_omp_for_static_chunk): Handle
5959 equal-argument loop exit phi.
5961 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
5964 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
5965 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
5966 operand 2 predicate.
5967 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
5968 operand 2 predicate.
5969 (ror,rol -> rorx splitters): Use const_int_operand as
5970 operand 2 predicate.
5972 2017-06-18 Richard Biener <rguenther@suse.de>
5974 PR tree-optimization/81410
5975 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
5976 the gap in the ! slp_perm SLP case after each group.
5978 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5981 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
5984 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5987 * predict.c (set_even_probabilities): Cleanup; do not affect
5988 probabilities that are already known.
5989 (combine_predictions_for_bb): Call even when count is set.
5991 2017-07-18 Nathan Sidwell <nathan@acm.org>
5993 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
5996 2017-07-18 Bin Cheng <bin.cheng@arm.com>
5999 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
6000 optimization for loop niter analysis.
6002 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
6005 * config/avr/avr.c (avr_optimize_casesi): Don't use
6006 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
6008 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
6010 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
6011 body_cost_vec from _vect_peel_extended_info.
6012 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
6013 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
6016 2017-07-18 Bin Cheng <bin.cheng@arm.com>
6018 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
6020 2017-07-18 Richard Biener <rguenther@suse.de>
6022 PR tree-optimization/80620
6023 PR tree-optimization/81403
6024 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
6025 info when re-using a VN table entry.
6027 2017-07-18 Richard Biener <rguenther@suse.de>
6029 PR tree-optimization/81418
6030 * tree-vect-loop.c (vectorizable_reduction): Properly compute
6031 vectype_in. Verify that with lane-reducing reduction operations
6032 we have a single def-use cycle.
6034 2017-07-17 Carl Love <cel@us.ibm.com>
6036 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
6038 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6039 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
6040 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
6041 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
6042 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
6043 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
6044 VMULOSW): New enum "unspec" values.
6045 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
6046 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
6047 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
6048 altivec_vmulosw): New patterns.
6049 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
6050 VMULOSW): Add definitions.
6051 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
6053 * config/alpha/alpha.c: Include predict.h.
6055 2017-07-17 Yury Gribov <tetra2005@gmail.com>
6057 * tree-vrp.c (compare_assert_loc): Fix comparison function
6058 to return predictable results.
6060 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6062 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
6065 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
6066 * doc/invoke.texi (mexpand-adddi): Update text.
6068 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6070 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
6071 that also clobbers the CC register. The old expand code is moved
6073 (*arc_clzsi2): ... here.
6074 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
6075 the CC register. The old expand code is moved to ...
6076 (arc_ctzsi2): ... here.
6078 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6080 * config/arc/arc.opt (mindexed-loads): Use initial value
6081 TARGET_INDEXED_LOADS_DEFAULT.
6082 (mauto-modify-reg): Use initial value
6083 TARGET_AUTO_MODIFY_REG_DEFAULT.
6084 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
6085 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
6086 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
6087 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
6089 2017-07-17 Martin Liska <mliska@suse.cz>
6092 * opts.c (finish_options): Do not allow -fgnu-tm
6093 w/ -fsanitize={kernel-,}address. Say sorry.
6095 2017-07-17 Bin Cheng <bin.cheng@arm.com>
6098 * tree-loop-distribution.c (classify_partition): Only assert on
6099 numer of iterations.
6100 (merge_dep_scc_partitions): Delete prameter. Update function call.
6101 (distribute_loop): Remove code handling loop with unknown niters.
6102 (pass_loop_distribution::execute): Skip loop with unknown niters.
6104 2017-07-17 Bin Cheng <bin.cheng@arm.com>
6107 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
6108 function sort_partitions_by_post_order.
6110 2017-07-17 Bin Cheng <bin.cheng@arm.com>
6112 PR tree-optimization/81374
6113 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
6114 the max index of basic blocks, rather than number of basic blocks.
6116 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6118 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
6120 (arc_legitimate_pic_operand_p): Likewise.
6121 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
6123 (arc_needs_pcl_p): Likewise.
6124 (arc_legitimate_pc_offset_p): Likewise.
6125 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
6126 function is also used in constrains.md.
6127 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
6128 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
6129 PLUS. Only return true/false in known cases, otherwise assert.
6130 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
6131 is already called in arc_legitimate_constant_p.
6132 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
6134 (LEGITIMATE_PIC_OPERAND_P): Use
6135 arc_raw_symbolic_reference_mentioned_p function.
6136 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
6141 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6142 Andrew Burgess <andrew.burgess@embecosm.com>
6144 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
6145 (arc_return_address_register): New function.
6146 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
6147 (arc_handle_fndecl_attribute): Add naked attribute.
6148 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
6149 (TARGET_WARN_FUNC_RETURN): Likewise.
6150 (arc_allocate_stack_slots_for_args): New function.
6151 (arc_warn_func_return): Likewise.
6152 (machine_function): Change type fn_type.
6153 (arc_compute_function_type): Consider new naked function type,
6154 change function return type.
6155 (arc_must_save_register): Adapt to handle new
6156 arc_compute_function_type's return type.
6157 (arc_expand_prologue): Likewise.
6158 (arc_expand_epilogue): Likewise.
6159 (arc_return_address_regs): Delete.
6160 (arc_return_address_register): New function.
6161 (arc_epilogue_uses): Use above function.
6162 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
6163 (arc_function_type): Change encoding, add naked type.
6164 (ARC_INTERRUPT_P): Change to handle the new encoding.
6165 (ARC_FAST_INTERRUPT_P): Likewise.
6166 (ARC_NORMAL_P): Define.
6167 (ARC_NAKED_P): Likewise.
6168 (arc_compute_function_type): Delete prototype.
6169 * config/arc/arc.md (in_ret_delay_slot): Use
6170 arc_return_address_register function.
6171 (simple_return): Likewise.
6172 (p_return_i): Likewise.
6174 2017-07-17 Jakub Jelinek <jakub@redhat.com>
6176 PR tree-optimization/81428
6177 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
6178 can't be built for those types.
6180 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
6182 Remove stuff dead since r239246.
6184 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
6185 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
6186 (avr_inform_devices): Remove dead stuff.
6188 2017-07-17 Tamar Christina <tamar.christina@arm.com>
6190 * config/arm/arm_neon.h: Fix softp typo.
6192 2017-07-17 Jakub Jelinek <jakub@redhat.com>
6194 PR tree-optimization/81365
6195 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
6196 aggregate moves onto bb predecessor edges, make sure there are no
6197 loads that could alias the lhs in between the start of bb and the
6200 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
6203 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
6204 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
6205 [LSHIFTRT, outer_code = TRUNCATE]: Same.
6207 2017-07-17 Jakub Jelinek <jakub@redhat.com>
6209 PR tree-optimization/81396
6210 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
6211 (init_symbolic_number): Initialize it to 1.
6212 (perform_symbolic_merge): Add n_ops from both operands into the new
6214 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
6215 without base_addr as useless if they need more than one operation.
6216 (bswap_replace): Handle !bswap case for NULL base_addr.
6218 2017-07-17 Tom de Vries <tom@codesourcery.com>
6221 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
6224 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
6226 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
6227 conditional builtin define __FIX_LEON3FT_B2BST.
6229 2017-07-17 Daniel Cederman <cederman@gaisler.com>
6231 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
6232 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
6235 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
6237 PR rtl-optimization/81424
6238 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
6239 to remove potential trapping from operands if -fnon-call-exceptions.
6241 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6243 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
6244 profile_proability for scalling.
6245 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
6247 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6249 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
6251 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6253 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
6254 fixpoint arithmetics.
6256 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6258 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
6259 fixpoint arithmetics.
6261 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6263 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
6264 fixpoint arithmetics.
6266 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6268 * profile-count.h (profile_probability::from_reg_br_prob_note,
6269 profile_probability::to_reg_br_prob_note): New functions.
6270 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
6271 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
6272 * predict.c (probability_reliable_p): Update.
6273 (edge_probability_reliable_p): Update.
6274 (br_prob_note_reliable_p): Update.
6275 (invert_br_probabilities): Update.
6276 (add_reg_br_prob_note): New function.
6277 (combine_predictions_for_insn): Update.
6278 * asan.c (asan_clear_shadow): Update.
6279 * cfgbuild.c (compute_outgoing_frequencies): Update.
6280 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
6281 (update_br_prob_note): Update.
6282 (rtl_verify_edges): Update.
6283 (purge_dead_edges): Update.
6284 (fixup_reorder_chain): Update.
6285 * emit-rtl.c (try_split): Update.
6286 * ifcvt.c (cond_exec_process_insns): Update.
6287 (cond_exec_process_if_block): Update.
6288 (dead_or_predicable): Update.
6289 * internal-fn.c (expand_addsub_overflow): Update.
6290 (expand_neg_overflow): Update.
6291 (expand_mul_overflow): Update.
6292 * loop-doloop.c (doloop_modify): Update.
6293 * loop-unroll.c (compare_and_jump_seq): Update.
6294 * optabs.c (emit_cmp_and_jump_insn_1): Update.
6295 * predict.h: Update.
6296 * reorg.c (mostly_true_jump): Update.
6298 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
6299 * config/alpha/alpha.c (emit_unlikely_jump): Update.
6300 * config/arc/arc.c: (emit_unlikely_jump): Update.
6301 * config/arm/arm.c: (emit_unlikely_jump): Update.
6302 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
6303 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
6304 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
6305 (ix86_print_operand): Update.
6306 (ix86_split_fp_branch): Update.
6307 (predict_jump): Update.
6308 * config/ia64/ia64.c (ia64_print_operand): Update.
6309 * config/mmix/mmix.c (mmix_print_operand): Update.
6310 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
6311 (rs6000_expand_split_stack_prologue): Update.
6312 * config/rs6000/rs6000.c: Update.
6313 * config/s390/s390.c (s390_expand_vec_strlen): Update.
6314 (s390_expand_vec_movstr): Update.
6315 (s390_expand_cs_tdsi): Update.
6316 (s390_expand_split_stack_prologue): Update.
6317 * config/sh/sh.c (sh_print_operand): Update.
6318 (expand_cbranchsi4): Update.
6319 (expand_cbranchdi4): Update.
6320 * config/sparc/sparc.c (output_v9branch): Update.
6321 * config/spu/spu.c (get_branch_target): Update.
6322 (ea_load_store_inline): Update.
6323 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
6324 * config/tilepro/tilepro.c: Update.
6326 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
6328 * gimplify.c (mostly_copy_tree_r): Revert latest change.
6329 (gimplify_save_expr): Likewise.
6331 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6333 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
6335 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6337 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
6339 * timevar.def (TV_IPA_FNSUMMARY): Define.
6341 2017-07-16 Daniel Cederman <cederman@gaisler.com>
6343 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
6344 to back store errata sensitive sequence from being generated.
6345 (sqrtdf2_fix): Likewise.
6347 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6349 * tree-ssa-threadupdate.c (compute_path_counts,
6350 update_joiner_offpath_counts): Use profile_probability.
6352 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
6355 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
6357 * config/arm/arm-c.c (arm_cpu_builtins): Define
6358 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
6360 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
6362 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6363 array entries to represent __ieee128 versions of the
6364 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
6365 scalar_extract_sig, and scalar_insert_exp built-in functions.
6366 (altivec_resolve_overloaded_builtin): Add special case handling
6367 for the __builtin_scalar_insert_exp function, as represented by
6368 the P9V_BUILTIN_VEC_VSIEDP constant.
6369 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
6370 exponent support for __ieee128 argument.
6371 (VSESQP): Add scalar extract signature support for __ieee128
6373 (VSTDCNQP): Add scalar test negative support for __ieee128
6375 (VSIEQP): Add scalar insert exponent support for __int128 argument
6376 with __ieee128 result.
6377 (VSIEQPF): Add scalar insert exponent support for __ieee128
6378 argument with __ieee128 result.
6379 (VSTDCQP): Add scalar test data class support for __ieee128
6381 (VSTDCNQP): Add overload support for scalar test negative with
6383 (VSTDCQP): Add overload support for scalar test data class
6385 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
6387 (UNSPEC_VSX_SIEXPQP): New constant.
6388 (xsxexpqp): New insn for VSX scalar extract exponent quad
6390 (xsxsigqp): New insn for VSX scalar extract significand quad
6392 (xsiexpqpf): New insn for VSX scalar insert exponent quad
6393 precision with floating point argument.
6394 (xststdcqp): New expand for VSX scalar test data class quad
6396 (xststdcnegqp): New expand for VSX scalar test negative quad
6398 (xststdcqp): New insn to match expansions for VSX scalar test data
6399 class quad precision and VSX scalar test negative quad precision.
6400 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6401 special case operand checking to enforce that second operand of
6402 VSX scalar test data class with quad precision argument is a 7-bit
6404 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
6405 prototypes and descriptions of __ieee128 versions of
6406 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
6407 scalar_test_data_class, and scalar_test_neg built-in functions.
6409 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6411 PR tree-optimization/81162
6412 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
6413 replace a negate with an add.
6415 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
6417 * doc/invoke.texi (arm/-mcpu): Document +crypto.
6419 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
6421 * config/arm/arm-c.c (arm_cpu_builtins): Define
6422 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
6424 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
6426 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
6427 (armv8-r): Set ARM Cortex-R52 as default CPU.
6428 * config/arm/arm-tables.opt: Regenerate.
6429 * config/arm/arm-tune.md: Regenerate.
6430 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
6432 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
6433 extension for -mcpu=cortex-r52.
6435 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
6437 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
6438 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
6439 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
6440 (fp-armv8): Define it as FP_ARMv8 only.
6441 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
6442 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
6444 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
6445 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
6446 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
6447 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
6448 than TARGET_FPU_ARMV8.
6449 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
6450 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
6451 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
6453 * config/arm/neon.md (neon_vrint): Likewise.
6454 (neon_vcvt): Likewise.
6455 (neon_<fmaxmin_op><mode>): Likewise.
6456 (<fmaxmin><mode>3): Likewise.
6457 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
6458 * config/arm/predicates.md (arm_cond_move_operator): Check against
6459 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
6461 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
6463 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
6466 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6468 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
6469 loop in comment with memset.
6471 2017-07-14 Martin Liska <mliska@suse.cz>
6473 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
6474 * dwarf2out.c (is_java): Remove the function.
6475 (output_pubname): Remove usage of the function.
6476 (lower_bound_default): Remove usage of DW_LANG_Java.
6477 (gen_compile_unit_die): Likewise.
6478 * gcc.c: Remove compiler defaults for .java and .zip files.
6479 * gimple-expr.c (remove_suffix): Change as there's no longer
6480 extension than 4-letter one.
6481 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
6482 (gimplify_save_expr): Likewise.
6483 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
6484 as it's possible even for other languages than Java.
6485 * langhooks.h (struct lang_hooks): Remove Java from a comment.
6486 * lto-opts.c (lto_write_options): Remove reference to Java.
6487 * opts.c (strip_off_ending): Update file extension handling.
6488 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
6489 * tree-eh.c (lower_resx): Likewise.
6490 * tree.c (free_lang_data_in_type): Remove dead code.
6491 (find_decls_types_r): Likewise.
6492 (build_common_builtin_nodes): Remove Java from a comment.
6493 (verify_type): Remove dead code.
6494 * varasm.c (assemble_external): Remove Java from a comment.
6496 2017-07-14 Martin Liska <mliska@suse.cz>
6498 * opts.c (finish_options): Add quotes.
6499 (common_handle_option): Likewise.
6501 2017-07-14 Martin Liska <mliska@suse.cz>
6503 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
6504 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
6506 * dwarf2out.c (lower_bound_default): Do not handle
6508 (gen_compile_unit_die): Likewise.
6509 * gcc.c: Remove default extension binding for GNU Pascal.
6510 * stmt.c: Remove Pascal language from a comment.
6511 * xcoffout.c: Likewise.
6513 2017-07-13 David Malcolm <dmalcolm@redhat.com>
6516 * diagnostic-show-locus.c (fixit_cmp): New function.
6517 (layout::layout): Sort m_fixit_hints.
6518 (column_range::column_range): Assert that the values are valid.
6519 (struct char_span): New struct.
6520 (correction::overwrite): New method.
6521 (struct source_line): New struct.
6522 (line_corrections::add_hint): Add assertions. Reimplement memcpy
6523 calls in terms of classes source_line and char_span, and
6524 correction::overwrite.
6525 (selftest::test_overlapped_fixit_printing_2): New function.
6526 (selftest::diagnostic_show_locus_c_tests): Call it.
6528 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
6530 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
6531 early if there is no lhs.
6533 2017-07-13 Martin Liska <mliska@suse.cz>
6535 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
6536 (gen_reference_type_die): Likewise.
6537 * stor-layout.c: Remove Pascal-related comment.
6539 2017-07-13 Martin Liska <mliska@suse.cz>
6541 * opts.c (finish_options): Add quotes to error messages.
6542 (parse_sanitizer_options): Likewise.
6544 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
6546 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
6548 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
6550 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
6552 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
6554 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
6556 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
6558 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
6561 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
6562 provides the hardware capability bits, define the macro
6563 __BUILTIN_CPU_SUPPORTS__.
6564 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
6565 if GLIBC does not provide the hardware capability bits. Add a
6566 gcc_unreachable call if the built-in cpu function is neither
6567 __builtin_cpu_is nor __builtin_cpu_supports.
6568 (rs6000_get_function_versions_dispatcher): Change the warning
6569 that an old GLIBC is used which does not export the capability
6570 bits to be an error.
6571 * doc/extend.texi (target_clones attribute): Document the
6572 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
6573 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
6574 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
6575 the macros defined by GCC if the newer GLIBC is available.
6577 2017-07-12 Jeff Law <law@redhat.com>
6579 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
6580 remaining includes slightly.
6581 * config/riscv/riscv-builtins.c: Include profile-count.h.
6583 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
6586 * config/avr/avr.c (avr_set_current_function): In diagnostic
6587 messages: Quote keywords and (parts of) identifiers.
6588 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
6591 2017-07-12 Carl Love <cel@us.ibm.com>
6593 * config/rs6000/rs6000-c.c: Add support for built-in functions
6594 vector bool char vec_revb (vector bool char);
6595 vector bool short vec_revb (vector short char);
6596 vector bool int vec_revb (vector bool int);
6597 vector bool long long vec_revb (vector bool long long);
6598 * doc/extend.texi: Update the built-in documentation file for the
6599 new built-in functions.
6601 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6603 * config/s390/s390.md: Remove movcc splitter.
6605 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6607 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
6608 load/store on condition.
6610 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
6613 * config/avr/avr.c (avr_encode_section_info)
6614 [progmem && !TREE_READONLY]: Error if progmem object needs
6617 2017-07-11 Michael Collison <michael.collison@arm.com>
6619 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
6622 2017-07-11 Carl Love <cel@us.ibm.com>
6624 * config/rs6000/rs6000-c.c: Add support for builtins
6625 vector unsigned int vec_parity_lsbb (vector signed int);
6626 vector unsigned int vec_parity_lsbb (vector unsigned int);
6627 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
6628 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
6629 vector unsigned long long vec_parity_lsbb (vector signed long long);
6630 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
6631 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
6632 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
6633 * doc/extend.texi: Update the built-in documentation file for the
6634 new built-in functions.
6636 2017-07-11 David Malcolm <dmalcolm@redhat.com>
6638 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
6639 (layout::m_primary_loc): New field.
6640 (layout::layout): Initialize new field. Move location filtering
6641 logic from here to...
6642 (layout::maybe_add_location_range): ...this new method. Add
6643 support for filtering to just the lines already specified by other
6645 (layout::will_show_line_p): New method.
6646 (gcc_rich_location::add_location_if_nearby): New method.
6647 (selftest::test_add_location_if_nearby): New test function.
6648 (selftest::diagnostic_show_locus_c_tests): Call it.
6649 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
6652 2017-07-11 Tom de Vries <tom@codesourcery.com>
6654 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
6655 (bb_first_real_insn): New function.
6656 (nvptx_single): Add extra initialization of broadcasted condition
6659 2017-07-11 Nathan Sidwell <nathan@acm.org>
6661 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
6663 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
6665 * doc/extend.texi (AVR Function Attributes): Remove weblink to
6666 Binutils doc as TEXI will mess them up.
6667 * doc/invoke.texi (AVR Options): Same here.
6669 2017-07-11 Daniel Cederman <cederman@gaisler.com>
6671 * config/sparc/sparc.opt (mfix-ut700): New option.
6672 (mfix-gr712rc): Likewise.
6673 (sparc_fix_b2bst): New variable.
6674 * doc/invoke.texi (SPARC options): Document them.
6675 (ARM options): Fix warnings.
6676 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
6677 instructions to prevent sequences that can trigger the store-store
6678 errata for certain LEON3FT processors.
6679 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
6680 (sparc_option_override): Set sparc_fix_b2bst appropriately.
6681 * config/sparc/sparc.md (fix_b2bst): New attribute.
6682 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
6684 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
6687 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
6689 (*rsqrtsf2_sse): Ditto.
6691 (div<mode>3): Macroize insn from divdf3 and divsf3
6692 using MODEF mode iterator.
6694 2017-07-10 Martin Sebor <msebor@redhat.com>
6696 PR tree-optimization/80397
6697 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
6698 instead of testing for equality to INTEGER_TYPE.
6700 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
6702 * config.gcc: Remove uclibc from arc target spec.
6704 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
6706 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
6708 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6711 * lto-wrapper.c (remove_option): New function.
6712 (merge_and_complain): Merge PIC/PIE options more realistically.
6714 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
6716 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
6720 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
6721 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
6722 * config.in: Regenerate.
6723 * configure: Regenerate.
6724 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
6725 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
6726 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
6727 (TARGET_GASISR_PROLOGUES): ...target mask.
6728 * common/config/avr/avr-common.c
6729 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
6730 Set -mgas-isr-prologues.
6731 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
6732 INSERT_PASS_BEFORE for it.
6733 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
6734 * config/avr/avr.c (avr_option_override)
6735 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
6736 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
6737 (avr_attribute_table) <no_gccisr>: Add new function attribute.
6738 (avr_set_current_function) <is_no_gccisr>: Init machine field.
6739 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
6741 (make_avr_pass_pre_proep): New function.
6742 (emit_push_sfr) <treg>: Add argument to function and use it
6744 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
6745 and set machine->gasisr.yes.
6746 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
6747 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
6748 __gcc_isr.n_pushed to .L__stack_usage.
6749 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
6750 (avr_asm_final_postscan_insn): ...this new static function.
6751 * config/avr/avr.h (machine_function)
6752 <is_no_gccisr, use_L__stack_usage>: New fields.
6753 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
6754 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
6755 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
6756 (gasisr, *gasisr): New expander and insn.
6757 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
6758 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
6759 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
6761 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
6763 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
6766 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
6768 Move jump-tables out of .text again.
6771 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
6772 (ASM_OUTPUT_ADDR_VEC): New function.
6773 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
6774 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
6775 INSN_ADDRESSes as asm comment.
6776 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
6777 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
6778 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
6779 * config/avr/avr.md (*tablejump): Adjust comment.
6780 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
6781 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
6783 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
6784 (avr_output_addr_vec): New proto.
6785 (avr_log_t) <insn_addresses>: New field.
6787 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
6790 * config/i386/i386.c (ix86_function_arg_advance): Set
6791 outgoing_args_on_stack to true if there are outgoing arguments
6793 (ix86_function_arg): Likewise.
6794 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
6795 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
6796 * config/i386/i386.h (machine_function): Add
6797 outgoing_args_on_stack.
6799 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
6801 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
6803 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
6805 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
6807 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
6808 (REAL_H): Remove $(MACHMODE_H).
6809 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
6811 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
6812 $(MACHMODE_H) and double-int.h.
6813 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
6815 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
6818 2017-07-07 Andrew Pinski <apinski@cavium.com>
6820 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
6821 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
6823 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6825 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
6826 Add warning if GCC was not configured to link against a GLIBC that
6827 exports the hardware capability bits.
6828 (make_resolver_func): Make resolver function private and not a
6829 COMDAT function. Create the name with clone_function_name instead
6830 of make_unique_name.
6833 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
6834 correct operand in doing the split.
6836 2017-07-07 Carl Love <cel@us.ibm.com>
6838 * config/rs6000/rs6000-c: Add support for built-in function
6839 vector unsigned short vec_pack_to_short_fp32 (vector float,
6841 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
6842 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
6843 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
6844 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
6845 (convert_4f32_8i16): Add define_expand.
6846 * doc/extend.texi: Update the built-in documentation file for the
6847 new built-in function.
6849 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6851 * config/sparc/m8.md: New file.
6852 * config/sparc/sparc.md: Include m8.md.
6854 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6856 * config/sparc/sparc.opt: New option -mvis4b.
6857 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
6858 (sparc_option_override): Handle VIS4B.
6859 (enum sparc_builtins): Define
6860 SPARC_BUILTIN_DICTUNPACK{8,16,32},
6861 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
6862 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
6863 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
6864 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
6865 (check_constant_argument): New function.
6866 (sparc_vis_init_builtins): Define builtins
6867 __builtin_vis_dictunpack{8,16,32},
6868 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
6869 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
6870 __builtin_vis_fpcmpde{8,16,32}shl and
6871 __builtin_vis_fpcmpur{8,16,32}shl.
6872 (sparc_expand_builtin): Check that the constant operands to
6873 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
6874 constant and in range.
6875 * config/sparc/sparc-c.c (sparc_target_macros): Handle
6877 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
6878 (SPARC_IMM5_P): Likewise.
6879 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
6880 (enabled): Handle vis4b.
6881 (UNSPEC_DICTUNPACK): New unspec.
6882 (UNSPEC_FPCMPSHL): Likewise.
6883 (UNSPEC_FPUCMPSHL): Likewise.
6884 (UNSPEC_FPCMPDESHL): Likewise.
6885 (UNSPEC_FPCMPURSHL): Likewise.
6886 (cpu_feature): New CPU feature `vis4b'.
6887 (dictunpack{8,16,32}): New insns.
6888 (FPCSMODE): New mode iterator.
6889 (fpcscond): New code iterator.
6890 (fpcsucond): Likewise.
6891 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
6892 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
6893 (fpcmpde{8,16,32}{si,di}shl): Likewise.
6894 (fpcmpur{8,16,32}{si,di}shl): Likewise.
6895 * config/sparc/constraints.md: Define constraints `q' for unsigned
6896 2-bit integer constants and `t' for unsigned 5-bit integer
6898 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
6900 (imm5_operand_dictunpack16): Likewise.
6901 (imm5_operand_dictunpack32): Likewise.
6902 (imm2_operand): Likewise.
6903 * doc/invoke.texi (SPARC Options): Document -mvis4b.
6904 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
6905 ditunpack* and fpcmp*shl builtins.
6907 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6909 * config.gcc: Handle m8 in --with-{cpu,tune} options.
6910 * config.in: Add HAVE_AS_SPARC6 define.
6911 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
6913 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
6915 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
6916 (CPP_CPU_SPEC): Handle m8.
6917 (ASM_CPU_SPEC): Likewise.
6918 * config/sparc/sparc-opts.h (enum processor_type): Add
6920 * config/sparc/sparc.c (m8_costs): New struct.
6921 (sparc_option_override): Handle TARGET_CPU_m8.
6922 (sparc32_initialize_trampoline): Likewise.
6923 (sparc64_initialize_trampoline): Likewise.
6924 (sparc_issue_rate): Likewise.
6925 (sparc_register_move_cost): Likewise.
6926 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
6927 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
6928 (ASM_CPU64_DEFAULT_SPEC): Likewise.
6929 (CPP_CPU_SPEC): Handle M8.
6930 (ASM_CPU_SPEC): Likewise.
6931 (AS_M8_FLAG): Define.
6932 * config/sparc/sparc.md: Add m8 to the cpu attribute.
6933 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
6934 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
6936 * configure: Regenerate.
6937 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
6940 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6942 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
6944 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
6945 ("*movdi_insn_sp32"): Do not set v3pipe.
6946 ("*movsi_insn"): Likewise.
6947 ("*movdi_insn_sp64"): Likewise.
6948 ("*movsf_insn"): Likewise.
6949 ("*movdf_insn_sp32"): Likewise.
6950 ("*movdf_insn_sp64"): Likewise.
6951 ("*zero_extendsidi2_insn_sp64"): Likewise.
6952 ("*sign_extendsidi2_insn"): Likewise.
6953 ("*mov<VM32:mode>_insn"): Likewise.
6954 ("*mov<VM64:mode>_insn_sp64"): Likewise.
6955 ("*mov<VM64:mode>_insn_sp32"): Likewise.
6956 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
6957 ("<vlop:code><VL:mode>3"): Likewise.
6958 ("*not_<vlop:code><VL:mode>3"): Likewise.
6959 ("*nand<VL:mode>_vis"): Likewise.
6960 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
6961 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
6962 ("one_cmpl<VL:mode>2"): Likewise.
6963 ("faligndata<VM64:mode>_vis"): Likewise.
6964 ("alignaddrsi_vis"): Likewise.
6965 ("alignaddrdi_vis"): Likweise.
6966 ("alignaddrlsi_vis"): Likewise.
6967 ("alignaddrldi_vis"): Likewise.
6968 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
6969 ("bmaskdi_vis"): Likewise.
6970 ("bmasksi_vis"): Likewise.
6971 ("bshuffle<VM64:mode>_vis"): Likewise.
6972 ("cmask8<P:mode>_vis"): Likewise.
6973 ("cmask16<P:mode>_vis"): Likewise.
6974 ("cmask32<P:mode>_vis"): Likewise.
6975 ("pdistn<P:mode>_vis"): Likewise.
6976 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
6978 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6980 * config/sparc/sparc.md ("subtype"): New insn attribute.
6981 ("*wrgsr_sp64"): Set insn subtype.
6982 ("*rdgsr_sp64"): Likewise.
6983 ("alignaddrsi_vis"): Likewise.
6984 ("alignaddrdi_vis"): Likewise.
6985 ("alignaddrlsi_vis"): Likewise.
6986 ("alignaddrldi_vis"): Likewise.
6987 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
6988 ("fexpand_vis"): Likewise.
6989 ("fpmerge_vis"): Likewise.
6990 ("faligndata<VM64:mode>_vis"): Likewise.
6991 ("bshuffle<VM64:mode>_vis"): Likewise.
6992 ("cmask8<P:mode>_vis"): Likewise.
6993 ("cmask16<P:mode>_vis"): Likewise.
6994 ("cmask32<P:mode>_vis"): Likewise.
6995 ("fchksm16_vis"): Likewise.
6996 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
6997 ("fmean16_vis"): Likewise.
6998 ("fp<plusminus_insn>64_vis"): Likewise.
6999 ("<plusminus_insn>v8qi3"): Likewise.
7000 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
7001 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
7002 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
7003 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
7004 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
7005 ("*movqi_insn"): Likewise.
7006 ("*movhi_insn"): Likewise.
7007 ("*movsi_insn"): Likewise.
7008 ("movsi_pic_gotdata_op"): Likewise.
7009 ("*movdi_insn_sp32"): Likewise.
7010 ("*movdi_insn_sp64"): Likewise.
7011 ("movdi_pic_gotdata_op"): Likewise.
7012 ("*movsf_insn"): Likewise.
7013 ("*movdf_insn_sp32"): Likewise.
7014 ("*movdf_insn_sp64"): Likewise.
7015 ("*zero_extendhisi2_insn"): Likewise.
7016 ("*zero_extendqihi2_insn"): Likewise.
7017 ("*zero_extendqisi2_insn"): Likewise.
7018 ("*zero_extendqidi2_insn"): Likewise.
7019 ("*zero_extendhidi2_insn"): Likewise.
7020 ("*zero_extendsidi2_insn_sp64"): Likewise.
7021 ("ldfsr"): Likewise.
7022 ("prefetch_64"): Likewise.
7023 ("prefetch_32"): Likewise.
7024 ("tie_ld32"): Likewise.
7025 ("tie_ld64"): Likewise.
7026 ("*tldo_ldub_sp32"): Likewise.
7027 ("*tldo_ldub1_sp32"): Likewise.
7028 ("*tldo_ldub2_sp32"): Likewise.
7029 ("*tldo_ldub_sp64"): Likewise.
7030 ("*tldo_ldub1_sp64"): Likewise.
7031 ("*tldo_ldub2_sp64"): Likewise.
7032 ("*tldo_ldub3_sp64"): Likewise.
7033 ("*tldo_lduh_sp32"): Likewise.
7034 ("*tldo_lduh1_sp32"): Likewise.
7035 ("*tldo_lduh_sp64"): Likewise.
7036 ("*tldo_lduh1_sp64"): Likewise.
7037 ("*tldo_lduh2_sp64"): Likewise.
7038 ("*tldo_lduw_sp32"): Likewise.
7039 ("*tldo_lduw_sp64"): Likewise.
7040 ("*tldo_lduw1_sp64"): Likewise.
7041 ("*tldo_ldx_sp64"): Likewise.
7042 ("*mov<VM32:mode>_insn"): Likewise.
7043 ("*mov<VM64:mode>_insn_sp64"): Likewise.
7044 ("*mov<VM64:mode>_insn_sp32"): Likewise.
7046 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7048 * config/sparc/sparc.md ("type"): New insn type viscmp.
7049 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
7051 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
7052 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
7053 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
7054 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
7056 ("n7_vis_logical_11cycle"): Likewise.
7057 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
7058 * config/sparc/niagara2.md ("niag3_vis": Likewise.
7059 * config/sparc/niagara.md ("niag_vis"): Likewise.
7060 * config/sparc/ultra3.md ("us3_fga"): Likewise.
7061 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
7063 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7065 * config/sparc/sparc.md: New instruction type `bmask'.
7066 (bmaskdi_vis): Use the `bmask' type.
7067 (bmasksi_vis): Likewise.
7068 * config/sparc/ultra3.md (us3_array): Likewise.
7069 * config/sparc/niagara7.md (n7_array): Likewise.
7070 * config/sparc/niagara4.md (n4_array): Likewise.
7071 * config/sparc/niagara2.md (niag2_vis): Likewise.
7072 (niag3_vis): Likewise.
7073 * config/sparc/niagara.md (niag_vis): Likewise.
7075 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
7077 * ipa-comdats.c: Remove optimize check from gate.
7078 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
7079 for functions not optimized.
7080 (ipa_fn_summary_read): Skip optimize check.
7081 (ipa_fn_summary_write): Likewise.
7082 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
7084 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
7086 (can_inline_edge_p): Check flag_pcc_struct_return for match.
7087 (check_callers): Give up on caller which is not optimized.
7088 (inline_small_functions): Likewise.
7089 (ipa_inline): Do not give up when not optimizing.
7090 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
7091 away unoptimizes cdtors.
7092 (whole_program_function_and_variable_visibility): Do
7093 ipa_discover_readonly_nonaddressable_vars in LTO mode.
7094 * ipa.c (process_references): Do not check optimize.
7095 (symbol_table::remove_unreachable_nodes): Update optimize check.
7096 (set_writeonly_bit): Update optimize check.
7097 (pass_ipa_cdtor_merge::gate): Do not check optimize.
7098 (pass_ipa_single_use::gate): Remove.
7100 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
7102 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
7103 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
7104 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
7105 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
7106 permute_load, permute_store, adjust_extract, adjust_splat,
7107 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
7108 replace_swap_with_copy, dump_swap_insn_table,
7109 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
7110 recombine_lvx_pattern, recombine_stvx_pattern,
7111 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
7112 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
7113 to file rs6000-p8swap.c.
7114 * config/rs6000/rs6000-p8swap.c: New file.
7115 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
7116 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
7117 and rs6000*-*-* targets.
7119 2017-07-06 David Malcolm <dmalcolm@redhat.com>
7121 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
7123 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
7125 * lto-wrapper.c (merge_and_complain): Do not merge
7126 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
7127 fsigned_zeros, ftrapping_math, fwrapv.
7128 (append_compiler_options): Do not track these options.
7129 (append_linker_options): Likewie
7131 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
7133 * cgraphunit.c (cgraph_node::finalize_function): When
7134 !flag_toplevel_reorde set no_reorder flag.
7135 (varpool_node::finalize_decl): Likewise.
7136 (symbol_table::compile): Drop no toplevel reorder path.
7138 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
7140 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
7141 edges; zero probability is not better than uninitialized.
7143 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
7145 * asan.h (asan_sanitize_allocas_p): Declare.
7146 * asan.c (asan_sanitize_allocas_p): New function.
7147 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
7148 (handle_builtin_alloca): Likewise.
7149 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
7150 if !asan_sanitize_allocas_p.
7151 * params.def (asan-instrument-allocas): Add new option.
7152 * params.h (ASAN_PROTECT_ALLOCAS): Define.
7153 * opts.c (common_handle_option): Disable allocas sanitization for
7156 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
7158 * asan.c: Include gimple-fold.h.
7159 (get_last_alloca_addr): New function.
7160 (handle_builtin_stackrestore): Likewise.
7161 (handle_builtin_alloca): Likewise.
7162 (asan_emit_allocas_unpoison): Likewise.
7163 (get_mem_refs_of_builtin_call): Add new parameter, remove const
7164 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
7165 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
7166 (instrument_builtin_call): Pass gimple iterator to
7167 get_mem_refs_of_builtin_call.
7168 (last_alloca_addr): New global.
7169 * asan.h (asan_emit_allocas_unpoison): Declare.
7170 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
7171 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
7172 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
7173 if function calls alloca.
7174 * gimple-fold.c (replace_call_with_value): Remove static keyword.
7175 * gimple-fold.h (replace_call_with_value): Declare.
7176 * internal-fn.c: Include asan.h.
7177 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
7178 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
7180 2017-07-06 David Malcolm <dmalcolm@redhat.com>
7182 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
7183 (C_SELFTEST_FLAGS): New.
7184 (CPP_SELFTEST_FLAGS): New.
7185 (SELFTEST_DEPS): New, from deps of s-selftest.
7186 (C_SELFTEST_DEPS): New, from deps of s-selftest.
7187 (CPP_SELFTEST_DEPS): New.
7188 (selftest): Add dependency on s-selftest-c++.
7189 (s-selftest): Rename to...
7190 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
7191 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
7192 than SELFTEST_FLAGS.
7193 (selftest-gdb): Rename to...
7194 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
7196 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
7197 (selftest-valgrind): Rename to...
7198 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
7200 (selftest-valgrind): Reintroduce as an alias for
7201 selftest-c-valgrind.
7202 (s-selftest-c++): New.
7203 (selftest-c++-gdb): New.
7204 (selftest-c++-valgrind): New.
7206 2017-07-06 Olivier Hainque <hainque@adacore.com>
7208 * gcc.c (process_command): When deciding if undefined variables
7209 should be ignored when processing specs, accept "gcc -v" as well.
7211 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
7213 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
7214 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
7216 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
7218 * config/arm/arm-cpus.in (armv8-r): Add new entry.
7219 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
7220 * config/arm/arm-tables.opt: Regenerate.
7221 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
7223 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
7225 2017-07-06 Carl Love <cel@us.ibm.com>
7227 * ChangeLog: Clean up from mid air collision
7229 2017-07-06 Carl Love <cel@us.ibm.com>
7231 * config/rs6000/rs6000-c.c: Add support for built-in functions
7232 vector signed int vec_subc (vector signed int, vector signed int);
7233 vector signed __int128 vec_subc (vector signed __int128,
7234 vector signed __int128);
7235 vector unsigned __int128 vec_subc (vector unsigned __int128,
7236 vector unsigned __int128);
7237 vector signed int vec_sube (vector signed int, vector signed int,
7239 vector unsigned int vec_sube (vector unsigned int,
7240 vector unsigned int,
7241 vector unsigned int);
7242 vector signed __int128 vec_sube (vector signed __int128,
7243 vector signed __int128,
7244 vector signed__int128);
7245 vector unsigned __int128 vec_sube (vector unsigned __int128,
7246 vector unsigned __int128,
7247 vector unsigned __int128);
7248 vector signed int vec_subec (vector signed int, vector signed int,
7250 vector unsigned int vec_subec (vector unsigned int,
7251 vector unsigned int,
7252 vector unsigned int);
7253 vector signed __int128 vec_subec (vector signed __int128,
7254 vector signed __int128,
7255 vector signed__int128);
7256 vector unsigned __int128 vec_subec (vector unsigned __int128,
7257 vector unsigned __int128,
7258 vector unsigned __int128);
7259 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
7260 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
7261 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
7262 BU_ALTIVEC_OVERLOAD_X definitions.
7263 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
7264 * doc/extend.texi: Update the built-in documentation file for the new
7267 2017-07-06 David Malcolm <dmalcolm@redhat.com>
7270 * diagnostic-show-locus.c (layout::layout): Use start and finish
7271 spelling location for the start and finish of each range.
7272 * genmatch.c (linemap_client_expand_location_to_spelling_point):
7273 Add unused aspect param.
7274 * input.c (expand_location_1): Add "aspect" param, and use it
7275 to access the correct part of the location.
7276 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
7278 (expand_location_to_spelling_point): Likewise.
7279 (linemap_client_expand_location_to_spelling_point): Add "aspect"
7280 param, and pass it to expand_location_1.
7282 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
7284 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
7285 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
7286 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
7287 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
7288 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
7289 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
7290 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
7291 _mm_maskz_getmant_ss): New intrinsics.
7292 (__builtin_ia32_getexpss128_mask): Changed to ...
7293 __builtin_ia32_getexpss128_round ... this.
7294 (__builtin_ia32_getexpsd128_mask): Changed to ...
7295 __builtin_ia32_getexpsd128_round ... this.
7296 * config/i386/i386-builtin-types.def
7297 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
7298 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
7299 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
7300 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
7301 __builtin_ia32_getmantss_mask_round): New builtins.
7302 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
7303 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
7304 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
7305 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
7306 * config/i386/sse.md
7307 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
7308 avx512f_sgetexp<mode><mask_scalar_name>
7309 <round_saeonly_scalar_name> ... this.
7310 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
7311 %0, %1, %2<round_saeonly_op3>}): Changed to ...
7312 vgetexp<ssescalarmodesuffix>
7313 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
7314 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
7315 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
7316 avx512f_vgetmant<mode><mask_scalar_name>
7317 <round_saeonly_scalar_name> ... this.
7318 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
7319 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
7320 vgetmant<ssescalarmodesuffix>
7321 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
7322 %0<mask_scalar_operand4>, %1, %2
7323 <round_saeonly_scalar_mask_op4>, %3} ... this.
7324 * config/i386/subst.md (mask_scalar_operand4,
7325 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
7326 round_saeonly_scalar_nimm_predicate): New subst attributes.
7328 2017-07-06 Julia Koval <julia.koval@intel.com>
7330 * config/i386/i386.c (ix86_erase_embedded_rounding):
7331 Remove code for old rounding pattern.
7333 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
7335 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
7337 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
7339 * doc/sourcebuild.texi (Test Directives, Variants of
7340 dg-require-support): Add documentation for dg-require-stack-check.
7342 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
7344 * config/i386/subst.md (mask_scalar, round_scalar,
7345 round_saeonly_scalar): New meta-templates.
7346 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
7347 round_scalar_mask_operand3, round_scalar_mask_op3,
7348 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
7349 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
7350 round_saeonly_scalar_constraint,
7351 round_saeonly_scalar_prefix): New subst attribute.
7352 * config/i386/sse.md
7353 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
7354 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
7355 <round_scalar_name> ... this.
7356 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
7357 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
7358 <round_scalar_name> ... this.
7359 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
7360 <sse>_vm<code><mode>3<mask_scalar_name>
7361 <round_saeonly_scalar_name> ... this.
7362 (v<plusminus_mnemonic><ssescalarmodesuffix>
7363 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
7364 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
7365 v<plusminus_mnemonic><ssescalarmodesuffix>
7366 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
7367 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
7368 (v<multdiv_mnemonic><ssescalarmodesuffix>
7369 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
7370 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
7371 v<multdiv_mnemonic><ssescalarmodesuffix>
7372 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
7373 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
7374 (v<maxmin_float><ssescalarmodesuffix>
7375 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
7376 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
7377 v<maxmin_float><ssescalarmodesuffix>
7378 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
7379 %0<mask_scalar_operand3>, %1, %<iptr>2
7380 <round_saeonly_scalar_mask_op3>} ... this.
7382 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
7384 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
7385 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
7387 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
7388 Alan Hayward <alan.hayward@arm.com>
7389 David Sherwood <david.sherwood@arm.com>
7391 * combine.c (simplify_if_then_else): Remove "enum" before
7393 * compare-elim.c (can_eliminate_compare): Likewise.
7394 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
7396 (aarch64_lookup_simd_builtin_type): Likewise.
7397 (aarch64_simd_builtin_type): Likewise.
7398 (aarch64_init_simd_builtin_types): Likewise.
7399 (aarch64_simd_expand_args): Likewise.
7400 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
7402 (aarch64_reverse_mask): Likewise.
7403 (aarch64_simd_emit_reg_reg_move): Likewise.
7404 (aarch64_gen_adjusted_ldpstp): Likewise.
7405 (aarch64_ccmp_mode_to_code): Likewise.
7406 (aarch64_operands_ok_for_ldpstp): Likewise.
7407 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
7408 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
7410 (aarch64_min_divisions_for_recip_mul): Likewise.
7411 (aarch64_reassociation_width): Likewise.
7412 (aarch64_get_condition_code_1): Likewise.
7413 (aarch64_simd_emit_reg_reg_move): Likewise.
7414 (aarch64_simd_attr_length_rglist): Likewise.
7415 (aarch64_reverse_mask): Likewise.
7416 (aarch64_operands_ok_for_ldpstp): Likewise.
7417 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
7418 (aarch64_gen_adjusted_ldpstp): Likewise.
7419 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
7421 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
7422 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
7423 (arm_lookup_simd_builtin_type): Likewise.
7424 (arm_simd_builtin_type): Likewise.
7425 (arm_init_simd_builtin_types): Likewise.
7426 (arm_expand_builtin_args): Likewise.
7427 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
7428 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
7429 (ft32_setup_incoming_varargs): Likewise.
7430 (ft32_function_arg): Likewise.
7431 (ft32_function_arg_advance): Likewise.
7432 (ft32_pass_by_reference): Likewise.
7433 (ft32_arg_partial_bytes): Likewise.
7434 (ft32_valid_pointer_mode): Likewise.
7435 (ft32_addr_space_pointer_mode): Likewise.
7436 (ft32_addr_space_legitimate_address_p): Likewise.
7437 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
7439 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
7440 (ix86_emit_outlined_ms2sysv_restore): Likewise.
7441 (iamcu_alignment): Likewise.
7442 (canonicalize_vector_int_perm): Likewise.
7443 (ix86_noce_conversion_profitable_p): Likewise.
7444 (ix86_mpx_bound_mode): Likewise.
7445 (ix86_operands_ok_for_move_multiple): Likewise.
7446 * config/microblaze/microblaze-protos.h
7447 (microblaze_expand_conditional_branch_reg): Likewise.
7448 * config/microblaze/microblaze.c
7449 (microblaze_expand_conditional_branch_reg): Likewise.
7450 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
7452 (rs6000_reassociation_width): Likewise.
7453 (rs6000_invalid_binary_op): Likewise.
7454 (fusion_p9_p): Likewise.
7455 (emit_fusion_p9_load): Likewise.
7456 (emit_fusion_p9_store): Likewise.
7457 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
7459 (riscv_hard_regno_mode_ok_p): Likewise.
7460 (riscv_address_insns): Likewise.
7461 (riscv_split_symbol): Likewise.
7462 (riscv_legitimize_move): Likewise.
7463 (riscv_function_value): Likewise.
7464 (riscv_hard_regno_nregs): Likewise.
7465 (riscv_expand_builtin): Likewise.
7466 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
7467 (riscv_build_integer): Likewise.
7468 (riscv_split_integer): Likewise.
7469 (riscv_legitimate_constant_p): Likewise.
7470 (riscv_cannot_force_const_mem): Likewise.
7471 (riscv_regno_mode_ok_for_base_p): Likewise.
7472 (riscv_valid_base_register_p): Likewise.
7473 (riscv_valid_offset_p): Likewise.
7474 (riscv_valid_lo_sum_p): Likewise.
7475 (riscv_classify_address): Likewise.
7476 (riscv_legitimate_address_p): Likewise.
7477 (riscv_address_insns): Likewise.
7478 (riscv_load_store_insns): Likewise.
7479 (riscv_force_binary): Likewise.
7480 (riscv_split_symbol): Likewise.
7481 (riscv_force_address): Likewise.
7482 (riscv_legitimize_address): Likewise.
7483 (riscv_move_integer): Likewise.
7484 (riscv_legitimize_const_move): Likewise.
7485 (riscv_legitimize_move): Likewise.
7486 (riscv_address_cost): Likewise.
7487 (riscv_subword): Likewise.
7488 (riscv_output_move): Likewise.
7489 (riscv_canonicalize_int_order_test): Likewise.
7490 (riscv_emit_int_order_test): Likewise.
7491 (riscv_function_arg_boundary): Likewise.
7492 (riscv_pass_mode_in_fpr_p): Likewise.
7493 (riscv_pass_fpr_single): Likewise.
7494 (riscv_pass_fpr_pair): Likewise.
7495 (riscv_get_arg_info): Likewise.
7496 (riscv_function_arg): Likewise.
7497 (riscv_function_arg_advance): Likewise.
7498 (riscv_arg_partial_bytes): Likewise.
7499 (riscv_function_value): Likewise.
7500 (riscv_pass_by_reference): Likewise.
7501 (riscv_setup_incoming_varargs): Likewise.
7502 (riscv_print_operand): Likewise.
7503 (riscv_elf_select_rtx_section): Likewise.
7504 (riscv_save_restore_reg): Likewise.
7505 (riscv_for_each_saved_reg): Likewise.
7506 (riscv_register_move_cost): Likewise.
7507 (riscv_hard_regno_mode_ok_p): Likewise.
7508 (riscv_hard_regno_nregs): Likewise.
7509 (riscv_class_max_nregs): Likewise.
7510 (riscv_memory_move_cost): Likewise.
7511 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
7512 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
7513 (rl78_addr_space_address_mode): Likewise.
7514 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7516 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
7517 (rs6000_reassociation_width): Likewise.
7518 (rs6000_invalid_binary_op): Likewise.
7519 (fusion_p9_p): Likewise.
7520 (emit_fusion_p9_load): Likewise.
7521 (emit_fusion_p9_store): Likewise.
7522 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
7523 (ok_for_simple_move_operands): Likewise.
7524 (ok_for_simple_move_strict_operands): Likewise.
7525 (ok_for_simple_arith_logic_operands): Likewise.
7526 (visium_legitimize_reload_address): Likewise.
7527 (visium_select_cc_mode): Likewise.
7528 (output_cbranch): Likewise.
7529 (visium_split_double_move): Likewise.
7530 (visium_expand_copysign): Likewise.
7531 (visium_expand_int_cstore): Likewise.
7532 (visium_expand_fp_cstore): Likewise.
7533 * config/visium/visium.c (visium_pass_by_reference): Likewise.
7534 (visium_function_arg): Likewise.
7535 (visium_function_arg_advance): Likewise.
7536 (visium_libcall_value): Likewise.
7537 (visium_setup_incoming_varargs): Likewise.
7538 (visium_legitimate_constant_p): Likewise.
7539 (visium_legitimate_address_p): Likewise.
7540 (visium_legitimize_address): Likewise.
7541 (visium_secondary_reload): Likewise.
7542 (visium_register_move_cost): Likewise.
7543 (visium_memory_move_cost): Likewise.
7544 (prepare_move_operands): Likewise.
7545 (ok_for_simple_move_operands): Likewise.
7546 (ok_for_simple_move_strict_operands): Likewise.
7547 (ok_for_simple_arith_logic_operands): Likewise.
7548 (visium_function_value_1): Likewise.
7549 (rtx_ok_for_offset_p): Likewise.
7550 (visium_legitimize_reload_address): Likewise.
7551 (visium_split_double_move): Likewise.
7552 (visium_expand_copysign): Likewise.
7553 (visium_expand_int_cstore): Likewise.
7554 (visium_expand_fp_cstore): Likewise.
7555 (visium_split_cstore): Likewise.
7556 (visium_select_cc_mode): Likewise.
7557 (visium_split_cbranch): Likewise.
7558 (output_cbranch): Likewise.
7559 (visium_print_operand_address): Likewise.
7560 * expmed.c (flip_storage_order): Likewise.
7561 * expmed.h (emit_cstore): Likewise.
7562 (flip_storage_order): Likewise.
7563 * genrecog.c (validate_pattern): Likewise.
7564 * hsa-gen.c (gen_hsa_addr): Likewise.
7565 * internal-fn.c (expand_arith_overflow): Likewise.
7566 * ira-color.c (allocno_copy_cost_saving): Likewise.
7567 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7568 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
7569 (process_invariant_for_inheritance): Likewise.
7570 * lra-eliminations.c (move_plus_up): Likewise.
7571 * omp-low.c (lower_oacc_reductions): Likewise.
7572 * simplify-rtx.c (simplify_subreg): Likewise.
7573 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
7574 (TARGET_CHKP_BOUND_MODE): Likewise..
7575 * targhooks.c (default_chkp_bound_mode): Likewise.
7576 (default_setup_incoming_vararg_bounds): Likewise.
7577 * targhooks.h (default_chkp_bound_mode): Likewise.
7578 (default_setup_incoming_vararg_bounds): Likewise.
7579 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
7580 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
7581 (have_whole_vector_shift): Likewise.
7582 * tree-vect-stmts.c (vectorizable_load): Likewise.
7583 * doc/tm.texi: Regenerate.
7585 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
7587 Graceful degrade if Binutils PR21472 is not available.
7590 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
7591 .rodata in flash test fails.
7592 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
7593 * confgure: Regenerate.
7594 * config.in: Regenerate.
7595 * config/avr/avr.c (avr_asm_named_section)
7596 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
7597 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
7598 (avr_asm_init_sections): Same.
7600 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
7602 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
7603 (fma<VH:mode>4_intrinsic): Likewise.
7604 (*fmsub<VCVTF:mode>4): Likewise.
7605 (*fmsub<VH:mode>4_intrinsic): Likewise.
7607 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
7610 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
7611 Don't depend on "optimize > 0".
7612 (out_movhi_r_mr, out_movqi_mr_r): Same.
7613 (out_movhi_mr_r, out_movqi_r_mr): Same.
7614 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
7615 io_address_operand on "optimize > 0".
7617 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7619 * tree-loop-distribution.c: Add general explanantion on the pass.
7620 (generate_loops_for_partition): Mark distributed loop.
7621 (pg_add_dependence_edges): New parameter. Handle alias data
7622 dependence specially and record it in the parameter if asked.
7623 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
7624 (init_partition_graph_vertices, add_partition_graph_edge): New.
7625 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
7626 (free_partition_graph_vdata, build_partition_graph): New.
7627 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
7628 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
7629 (data_ref_segment_size, latch_dominated_by_data_ref): New.
7630 (compute_alias_check_pairs, version_loop_by_alias_check): New.
7631 (version_for_distribution_p, finalize_partitions): New.
7632 (distribute_loop): Handle alias data dependence specially. Factor
7633 out loop fusion code as functions and call these functions.
7635 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7637 * tree-loop-distribution.c (classify_partition): New parameter and
7638 better handle reduction statement.
7639 (rdg_build_partitions): Revise comment.
7640 (distribute_loop): Compute statements in all partitions and pass it
7641 to classify_partition.
7643 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7645 * tree-loop-distribution.c (enum partition_type): New.
7646 (struct partition): New field type.
7647 (partition_merge_into): Add parameter. Update partition type.
7648 (data_dep_in_cycle_p, update_type_for_merge): New functions.
7649 (build_rdg_partition_for_vertex): Compute partition type.
7650 (rdg_build_partitions): Dump partition type.
7651 (distribute_loop): Update calls to partition_merge_into.
7653 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7655 * tree-loop-distribution.c (struct ddr_hasher): New.
7656 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
7658 (classify_partition): Call get_data_dependence.
7659 (pg_add_dependence_edges): Ditto.
7660 (distribute_loop): Release data dependence hash table.
7662 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7664 * tree-loop-distribution.c (ref_base_address): Delete.
7665 (similar_memory_accesses): Rename ...
7666 (share_memory_accesses): ... to this. Check if partitions access
7667 the same memory reference.
7668 (distribute_loop): Call share_memory_accesses.
7670 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7672 * tree-loop-distribution.c (struct partition): New field recording
7674 (partition_alloc, partition_free): Init and release data refs.
7675 (partition_merge_into): Merge data refs.
7676 (build_rdg_partition_for_vertex): Collect data refs for partition.
7677 (pg_add_dependence_edges): Change parameters from vector to bitmap.
7679 (distribute_loop): Remve data refs from vertice data of partition
7682 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7684 * tree-loop-distribution.c (params.h): Include header file.
7685 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
7686 (datarefs_vec): New global var.
7687 (create_rdg_vertices): Use datarefs_vec directly.
7688 (free_rdg): Don't free data references.
7689 (build_rdg): Update use. Don't free data references.
7690 (distribute_loop): Compute global variable for data references.
7691 Bail out if there are too many data references.
7693 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7695 * tree-loop-distribution.c (loop_nest): New global var.
7696 (build_rdg): Use loop directly, rather than loop nest.
7697 (pg_add_dependence_edges): Remove loop nest parameter. Use global
7699 (distribute_loop): Compute global variable loop nest. Update use.
7701 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7703 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
7704 (partition_merge_into): New parameter. Dump reason for fusion.
7705 (distribute_loop): Update use of partition_merge_into.
7707 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7709 * tree-loop-distribution.c (bb_top_order_index): New.
7710 (bb_top_order_index_size, bb_top_order_cmp): New.
7711 (stmts_from_loop): Use topological order.
7712 (pass_loop_distribution::execute): Compute and release topological
7713 order for basic blocks.
7715 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7717 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
7720 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7722 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
7723 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
7724 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
7725 * internal-fn.def (LOOP_DIST_ALIAS): New.
7726 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
7727 (fold_loop_internal_call): ... this.
7728 (vect_loop_dist_alias_call): New function.
7729 (set_uid_loop_bbs): Call fold_loop_internal_call.
7730 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
7733 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
7736 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
7737 Require dead FLAGS_REG at the beginning of a peephole.
7739 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
7742 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
7743 arguments in the call to __builtin_ia32_sbb_u32.
7744 (_subborrow_u64): Swap _X and _Y arguments in the call to
7745 __builtin_ia32_sbb_u64.
7747 2017-07-04 Jakub Jelinek <jakub@redhat.com>
7750 * tree-vrp.c (compare_assert_loc): Turn into a function template
7751 with stable template parameter. Only test if a->e is NULL,
7752 !a->e == !b->e has been verified already. Use e == NULL or
7753 e != NULL instead of e or ! e tests. If stable is true, don't use
7754 iterative_hash_expr, on the other side allow a or b or both NULL
7755 and sort the NULLs last.
7756 (process_assert_insertions): Sort using compare_assert_loc<false>
7757 instead of compare_assert_loc, later sort using
7758 compare_assert_loc<true> before calling process_assert_insertions_for
7759 in a loop. Use break instead of continue once seen NULL pointer.
7761 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
7763 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
7764 Cortex-R7 and Cortex-R8 processors.
7766 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
7768 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
7769 uninitialized while src is not.
7771 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
7773 * common/config/arm/arm-common.c: Adjust include path for
7775 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
7776 (arm-cpu.h): Create in build directory. Adjust dependency rules.
7777 (arm-cpu-data.h): Likewise.
7778 (arm-cpu-cdata.h): Likewise.
7779 * config/arm/arm-cpu.h: Delete.
7780 * config/arm/arm-cpu-cdata.h: Delete.
7781 * config/arm/arm-cpu-data.h: Delete.
7783 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
7785 * config/arm/arm-cpus.in (cortex-a55): New.
7786 (cortex-a75): Likewise.
7787 (cortex-a75.cortex-a55): Likewise.
7788 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
7790 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
7791 * config/arm/arm-cpu-cdata.h: Regenerate.
7792 * config/arm/arm-cpu-data.h: Regenerate.
7793 * config/arm/arm-cpu.h: Regenerate.
7794 * config/arm/arm-tables.opt: Regenerate.
7795 * config/arm/arm-tune.md: Regenerate.
7797 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
7799 * haifa-sched.c (sched_create_recovery_edges): Update profile.
7801 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
7803 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
7806 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
7808 PR tree-optimization/81292
7809 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
7810 full_string_p, also call adjust_related_strinfos if the adjustment
7811 is simple, otherwise invalidate related strinfos.
7813 2017-07-04 Martin Liska <mliska@suse.cz>
7816 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
7817 newly created variable as DECL_IGNORED_P.
7819 2017-07-04 Martin Liska <mliska@suse.cz>
7822 * ipa-inline.c (inline_small_functions):
7823 Use xstrdup_for_dump.
7825 2017-07-04 Tom de Vries <tom@codesourcery.com>
7827 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
7829 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
7832 * config/darwin.c (darwin_function_switched_text_sections):
7835 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
7837 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
7839 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
7841 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
7843 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7845 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
7846 min_profitable_iters, and th as inclusive lower bounds.
7847 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
7848 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
7849 for min_profitable_iters and min_profitable_estimate.
7850 (vect_transform_loop): Treat th as an inclusive lower bound.
7851 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
7853 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
7856 * config/darwin.c (darwin_function_switched_text_sections):
7857 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
7858 in two pieces, and suppress the use of buf.
7860 2017-07-03 Nathan Sidwell <nathan@acm.org>
7862 * hash-table.h (hash_table_mod1): Fix indentation.
7864 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7867 * predict.c (force_edge_cold): Be more careful about propagation
7869 * profile-count.h (profile_probability::guessed,
7870 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
7872 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
7874 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
7876 * doc/invoke.texi (rcpc architecture extension): Document it.
7878 2017-07-03 Richard Biener <rguenther@suse.de>
7880 PR tree-optimization/60510
7881 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
7882 the scalar reduction PHI and use it.
7883 (vectorizable_reduction): Properly guard the single_defuse_cycle
7884 path for non-SLP reduction chains where we cannot use it.
7885 Rework reduc_def/index and vector type deduction. Rework
7886 vector operand gathering during reduction op code-gen.
7887 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
7888 chains dissolve the chain and leave it to non-SLP reduction
7891 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7893 * tree-data-ref.h (dr_alignment): Declare.
7894 * tree-data-ref.c (dr_alignment): New function.
7895 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
7896 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
7898 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
7900 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7902 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
7903 and base_misalignment fields.
7904 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
7905 * tree-data-ref.c: Include builtins.h.
7906 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
7907 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
7908 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
7909 * tree-vect-data-refs.c: Include tree-cfg.h.
7910 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
7911 fields instead of calculating an alignment here.
7912 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
7913 innermost_loop_behavior fields.
7915 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7917 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
7919 (DR_STEP_ALIGNMENT): New macro.
7920 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
7921 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
7922 (create_data_ref): Print it.
7923 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
7924 to tell whether the step preserves vector (mis)alignment.
7925 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
7926 Move the check for an integer step and generalise to all INTEGER_CST.
7927 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
7928 Print the outer step alignment.
7930 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7932 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
7933 with offset_alignment.
7934 (DR_ALIGNED_TO): Delete.
7935 (DR_OFFSET_ALIGNMENT): New macro.
7936 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
7937 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
7938 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
7939 (create_data_ref): Likewise.
7940 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
7941 (vect_analyze_data_refs): Likewise.
7942 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
7943 creating dummy innermost behavior.
7945 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7947 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
7948 with a "innermost_loop_behavior *" and refeence tree.
7949 * tree-data-ref.c (dr_analyze_innermost): Likewise.
7950 (create_data_ref): Update call accordingly.
7951 * tree-predcom.c (find_looparound_phi): Likewise.
7953 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7955 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
7956 fields with dr_wrt_vec_loop.
7957 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
7958 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
7959 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
7960 (vect_dr_behavior): New function.
7961 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
7962 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
7963 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
7964 track whether the step preserves the misalignment.
7965 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
7966 Use vect_dr_behavior.
7967 (vect_setup_realignment): Update call accordingly.
7968 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
7969 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
7970 call to vect_create_addr_base_for_vector_ref.
7971 (vect_create_cond_for_align_checks): Likewise.
7972 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
7973 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
7974 (vect_recog_mask_conversion_pattern): Likewise.
7975 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
7976 (new_stmt_vec_info): Remove redundant zeroing.
7978 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
7980 * common/config/arm/arm-common.c (arm_be8_option): New function.
7981 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
7982 (ISA_ARMv6): Add isa_bit_be8.
7983 * config/arm/arm.h (arm_be8_option): Add prototype.
7984 (BE8_SPEC_FUNCTION): New define.
7985 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
7986 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
7987 (mlittle-endian): Similarly.
7988 (mbe8, mbe32): New options.
7989 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
7990 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
7992 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7994 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
7996 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7998 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
7999 (cleanup_tree_cfg_bb): Use it.
8000 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
8002 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
8004 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8007 * loop-doloop.c (add_test): Update profile.
8009 2017-07-03 Martin Liska <mliska@suse.cz>
8012 * sanopt.c (rewrite_usage_of_param): New function.
8013 (sanitize_rewrite_addressable_params): Likewise.
8014 (pass_sanopt::execute): Call rewrite_usage_of_param.
8016 2017-07-03 Richard Biener <rguenther@suse.de>
8018 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
8019 back to using VIEW_CONVERT_EXPR.
8021 2017-07-03 Martin Liska <mliska@suse.cz>
8024 * doc/extend.texi: Document when a resolver function is
8025 generated for target_clones.
8027 2017-07-03 Martin Liska <mliska@suse.cz>
8029 * asan.c (asan_emit_stack_protection): Unpoison just red zones
8030 and shadow memory of auto variables which are subject of
8031 use-after-scope sanitization.
8032 (asan_expand_mark_ifn): Add do set only when is_poison.
8034 2016-07-03 Richard Biener <rguenther@suse.de>
8036 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
8038 (vect_force_simple_reduction): Record reduction def -> phi mapping.
8039 (vectorizable_reduction): Perform reduction PHI creation when
8040 visiting a reduction PHI and adjust and simplify code generation
8041 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
8042 (vect_transform_loop): Visit reduction PHIs.
8043 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
8044 defs into the SLP tree.
8045 (vect_build_slp_tree): Reduction defs terminate the recursion.
8046 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
8048 (vect_get_vec_defs_for_stmt_copy): Export.
8049 (vect_get_vec_defs): Likewise.
8050 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
8052 (vect_get_vec_defs_for_stmt_copy): Declare.
8053 (vect_get_vec_defs): Likewise.
8055 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8057 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
8058 parameter with a "loop" parameter and use it instead of the
8059 loop containing DR_STMT. Don't check simple_iv when doing
8060 BB analysis. Describe the two analysis modes in the comment.
8062 2017-07-03 Tom de Vries <tom@codesourcery.com>
8064 PR tree-optimization/69468
8065 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
8066 (find_same_succ_bb): Handle ignore_edge_flags.
8068 2017-07-03 Tom de Vries <tom@codesourcery.com>
8070 PR tree-optimization/81192
8071 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
8073 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
8075 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
8077 2017-07-03 Tom de Vries <tom@codesourcery.com>
8079 PR tree-optimization/81192
8080 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
8081 BB_SAME_SUCC (bb) == NULL.
8083 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8085 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
8088 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8090 * dumpfile.c: Include profile-count.h
8091 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
8093 (insert_cond_bb): Update profile.
8094 * tree-cfg.h (insert_cond_bb): Update prototype.
8095 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
8096 * tree-dump.c: Do not include tree-cfg.
8098 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8100 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
8102 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8104 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
8107 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8109 * tree-complex.c (expand_complex_div_wide): update profile.
8111 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
8112 Alan Hayward <alan.hayward@arm.com>
8113 David Sherwood <david.sherwood@arm.com>
8115 * Makefile.in (MACHMODE_H): Remove insn-modes.h
8116 (CORETYPES_H): New define.
8117 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
8118 (insn-modes-inline.h, s-modes-inline-h): New rules.
8119 (generated_files): Add insn-modes-inline.h.
8120 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
8121 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
8122 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
8123 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
8124 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
8125 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
8126 (build/gencodes.o, build/genconditions.o): Likewise.
8127 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
8128 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
8129 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
8130 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
8131 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
8132 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
8133 * coretypes.h: Include everything up to real.h for generators.
8134 Include insn-modes.h first. Include wide-int-print.h after
8135 wide-int.h. Include insn-modes-inline.h and then machmode.h.
8136 * machmode.h: Don't include insn-modes.h here.
8137 * function-tests.c: Remove includes of signop.h, machmode.h,
8138 double-int.h and wide-int.h.
8140 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
8142 * optc-save-gen.awk: Likewise.
8143 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
8144 * godump.c: Remove include of wide-int-print.h.
8145 * pretty-print.h: Likewise.
8146 * wide-int-print.cc: Likewise.
8147 * wide-int.cc: Likewise.
8148 * hash-map-tests.c: Remove include of signop.h.
8149 * hash-set-tests.c: Likewise.
8150 * rtl-tests.c: Likewise.
8151 * mkconfig.sh: Remove include of machmode.h.
8152 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
8154 (emit_insn_modes_inline_h): ...this new function. Emit the code
8155 into an insn-modes-inline.h header file, adding appropriate
8156 include guards and end comments.
8157 (emit_insn_modes_c_header): Remove include of machmode.h.
8158 (emit_min_insn_modes_c_header): Include coretypes.h rather than
8160 (main): Handle -i flag and call emit_insn_modes_inline_h when
8163 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
8165 * tree-ssa-strlen.c (strinfo): Rename the length field to
8166 nonzero_chars. Add a full_string_p field.
8167 (compare_nonzero_chars, zero_length_string_p): New functions.
8168 (get_addr_stridx): Add an offset_out parameter.
8169 Use compare_nonzero_chars.
8170 (get_stridx): Update accordingly. Use compare_nonzero_chars.
8171 (new_strinfo): Update after above changes to strinfo.
8172 (set_endptr_and_length): Set full_string_p.
8173 (get_string_length): Update after above changes to strinfo.
8174 (unshare_strinfo): Update call to new_strinfo.
8175 (maybe_invalidate): Likewise.
8176 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
8177 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
8178 as a uhwi instead of an shwi. Update after above changes to
8179 strinfo and new_strinfo.
8180 (zero_length_string): Assert that chainsi contains full strings.
8181 Use zero_length_string_p. Update call to new_strinfo.
8182 (adjust_related_strinfos): Update after above changes to strinfo.
8183 Copy full_string_p from origsi.
8184 (adjust_last_stmt): Use zero_length_string_p.
8185 (handle_builtin_strlen): Update after above changes to strinfo and
8186 new_strinfo. Install the lhs as the string length if the previous
8187 entry didn't describe a full string.
8188 (handle_builtin_strchr): Update after above changes to strinfo
8190 (handle_builtin_strcpy): Likewise.
8191 (handle_builtin_strcat): Likewise.
8192 (handle_builtin_malloc): Likewise.
8193 (handle_pointer_plus): Likewise.
8194 (handle_builtin_memcpy): Likewise. Track nonzero characters
8195 that aren't necessarily followed by a nul terminator.
8196 (handle_char_store): Likewise.
8198 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
8200 PR tree-optimization/80769
8201 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
8202 for malloc and calloc. Document the new invariant that all related
8203 strinfos have delayed lengths or none do.
8204 (verify_related_strinfos): Move earlier in file.
8205 (set_endptr_and_length): New function, split out from...
8206 (get_string_length): ...here. Also set the lengths of related
8208 (zero_length_string): Assert that chainsi has known (rather than
8210 (adjust_related_strinfos): Likewise.
8212 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
8214 PR tree-optimization/81136
8215 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
8216 assert that two references with the same misalignment have the same
8217 compile-time misalignment if those compile-time misalignments
8220 2017-07-01 Andi Kleen <ak@linux.intel.com>
8222 * print-tree.c (print_node): Print all attributes.
8224 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8226 * cfg.c (scale_bbs_frequencies): New function.
8227 * cfg.h (scale_bbs_frequencies): Declare it.
8228 * cfgloopanal.c (single_likely_exit): Cleanup.
8229 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
8231 (scale_loop_profile): Likewise.
8232 (loop_version): Likewise.
8233 (create_empty_loop_on_edge): Update.
8234 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
8235 scale_loop_frequencies, scale_loop_profile, loopify,
8236 loop_version): Update prototypes.
8237 * modulo-sched.c (sms_schedule): Update.
8238 * predict.c (unlikely_executed_edge_p): Also check probability.
8239 (probably_never_executed_edge_p): Fix typo.
8240 * tree-if-conv.c (version_loop_for_if_conversion): Update.
8241 * tree-parloops.c (gen_parallel_loop): Update.
8242 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
8243 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8244 * tree-ssa-loop-split.c (split_loop): Update.
8245 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
8246 * tree-vect-loop-manip.c (vect_do_peeling): Update.
8247 (vect_loop_versioning): Update.
8248 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8250 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8252 * trans-mem.c (split_bb_make_tm_edge): Update profile.
8254 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8256 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
8257 to keep profile consistent.
8259 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8261 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
8262 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
8263 * profile-count.h (max_safe_multiplier): Make unsigned.
8264 (profile_count::guessed_zero): New.
8266 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8268 * bb-reorder.c (fix_up_crossing_landing_pad,
8269 fix_crossing_conditional_branches): Use make_single_succ_edge
8270 to keep profile consistent.
8272 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8274 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
8277 2017-07-01 Jakub Jelinek <jakub@redhat.com>
8280 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
8281 the right scopes, make sure cond_jump isn't preserved between multiple
8282 iterations. Search for fallthru edge whenever there are 3+ edges and
8283 use find_fallthru_edge for it.
8285 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8287 Patch by Alexander Monakov <amonakov@ispras.ru>
8288 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
8289 probabilities consistently.
8291 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8293 * pa.c (pa_expand_compare_and_swap_loop): Update call of
8294 emit_cmp_and_jump_insns.
8296 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8299 * tree-inline.c (expand_call_inline): Combine profile statuses.
8301 2017-06-30 Andrew Pinski <apinski@cavium.com>
8303 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
8304 fold_stmt returned true.
8306 2017-06-30 Nathan Sidwell <nathan@acm.org>
8308 * ggc.h (empty_string): Delete.
8309 * cfgexpand.c (expand_asm_stmt): Use plain "".
8310 * optabs.c (expand_asm_memory_barrier): Likewise.
8311 * stringpool.c (empty_string): Delete.
8312 (digit_vector, digit_string): Delete.
8313 (ggc_alloc_string): Use plain "", don't optimize single digit
8314 strings. Use ggc_alloc_atomic.
8316 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
8318 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
8319 comparison set and one other set, use the cost of the non-comparison
8322 2017-06-30 Nathan Sidwell <nathan@acm.org>
8324 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
8327 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
8329 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
8330 loops. Remove now unneeded calls to gimple_switch_set_label() that
8331 just set removed labels to NULL_TREE.
8333 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
8335 * tree-ssanames.c (set_range_info_raw): Abstract from ...
8336 (set_range_info): ...here. Only call set_range_info_raw if domain
8338 (set_nonzero_bits): Call set_range_info_raw.
8339 * tree-ssanames.h (set_range_info_raw): New.
8341 2017-06-30 Jakub Jelinek <jakub@redhat.com>
8344 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
8345 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
8346 of nonimmediate_operand and <store_mask_constraint> instead of m
8347 for the input operand. For V8FI iterator, always split if input
8348 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
8349 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
8350 <store_mask_predicate> instead of register_operand and
8351 <store_mask_constraint> instead of v for the input operand. Make
8352 sure both operands aren't MEMs for if not <mask_applied>.
8354 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
8356 * lto-wrapper.c (copy_file) Close both file descriptors before
8359 2017-06-30 Martin Liska <mliska@suse.cz>
8362 * multiple_target.c (create_dispatcher_calls): Make ifunc
8363 also for function that don't have calls or are not referenced.
8365 2017-06-30 Richard Biener <rguenther@suse.de>
8367 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
8368 analyze the first scalar stmt. Move vector type computation
8369 for the BB case here from ...
8370 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
8371 live operation processing in the SLP case properly.
8373 2017-06-30 Richard Biener <rguenther@suse.de>
8375 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
8377 2017-06-30 Martin Liska <mliska@suse.cz>
8380 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
8381 before BUILT_IN_UNWIND_RESUME when ASAN is used.
8383 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
8385 * doc/invoke.texi (AArch64): Add missing options and remove redundant
8388 2017-06-30 Richard Biener <rguenther@suse.de>
8390 PR tree-optimization/81249
8391 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
8392 condition reduction result to original scalar type.
8394 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8396 * profile-count.h (enum profile_quality): Fix typos and whitespace
8399 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8401 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
8402 type for branch probabilities.
8404 2017-06-29 Julian Brown <julian@codesourcery.com>
8405 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8407 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
8408 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
8409 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
8410 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
8412 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8414 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
8415 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
8416 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
8417 CC usage from generic code to here.
8418 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
8419 CC usage into the target macros.
8421 2017-06-29 Maya Rashish <coypu@sdf.org>
8423 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
8426 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8428 * arm/arm-builtins.c: Include profile-count.h
8429 * except.c (sjlj_emit_function_enter): Use
8430 profile_probability::unlikely.
8432 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8434 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
8435 and tocrel_offset be pointer args rather than implicitly using
8437 (legitimate_constant_pool_address_p, rs6000_emit_move,
8438 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
8439 tocrel_offset and use in toc_relative_expr_p call.
8440 (print_operand, print_operand_address): Use static tocrel_base_oac
8441 and tocrel_offset_oac.
8442 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
8445 2017-06-29 Maya Rashish <coypu@sdf.org>
8447 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
8449 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
8451 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
8452 objects, take into account only the alignment of 'op0' and 'mode1' if
8455 2017-06-29 Steve Ellcey <sellcey@cavium.com>
8457 * ccmp.c (ccmp_tree_comparison_p): New function.
8458 (ccmp_candidate_p): Update to use above function.
8459 (get_compare_parts): New function.
8460 (expand_ccmp_next): Update to use new functions.
8461 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
8463 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
8464 take mode as argument.
8465 * ccmp.h (expand_ccmp_expr): Add mode as argument.
8466 * expr.c (expand_expr_real_1): Pass mode as argument.
8468 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
8470 * combine.c (combine_instructions): Print insns to dump_file, together
8473 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8475 * asan.c (asan_emit_stack_protection): Update.
8476 (create_cond_insert_point): Update.
8477 * auto-profile.c (afdo_propagate_circuit): Update.
8478 * basic-block.h (struct edge_def): Turn probability to
8479 profile_probability.
8480 (EDGE_FREQUENCY): Update.
8481 * bb-reorder.c (find_traces_1_round): Update.
8482 (better_edge_p): Update.
8483 (sanitize_hot_paths): Update.
8484 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
8485 (make_single_succ_edge): Update.
8486 (check_bb_profile): Update.
8487 (dump_edge_info): Update.
8488 (update_bb_profile_for_threading): Update.
8489 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
8490 probabilitycount to 0.
8491 * cfgbuild.c (compute_outgoing_frequencies): Update.
8492 * cfgcleanup.c (try_forward_edges): Update.
8493 (outgoing_edges_match): Update.
8494 (try_crossjump_to_edge): Update.
8495 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
8496 (expand_gimple_tailcall): Update.
8497 (construct_init_block): Use make_single_succ_edge.
8498 (construct_exit_block): Use make_single_succ_edge.
8499 * cfghooks.c (verify_flow_info): Update.
8500 (redirect_edge_succ_nodup): Update.
8501 (split_edge): Update.
8502 (account_profile_record): Update.
8503 * cfgloopanal.c (single_likely_exit): Update.
8504 * cfgloopmanip.c (scale_loop_profile): Update.
8505 (set_zero_probability): Remove.
8506 (duplicate_loop_to_header_edge): Update.
8507 * cfgloopmanip.h (loop_version): Update prototype.
8508 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
8509 (force_nonfallthru_and_redirect): Update.
8510 (update_br_prob_note): Update.
8511 (rtl_verify_edges): Update.
8512 (purge_dead_edges): Update.
8513 (rtl_lv_add_condition_to_bb): Update.
8514 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
8515 * cgraphunit.c (init_lowered_empty_function): Update.
8516 (cgraph_node::expand_thunk): Update.
8517 * cilk-common.c: Include profile-count.h
8518 * dojump.c (inv): Remove.
8519 (jumpifnot): Update.
8520 (jumpifnot_1): Update.
8521 (do_jump_1): Update.
8523 (do_jump_by_parts_greater_rtx): Update.
8524 (do_compare_rtx_and_jump): Update.
8525 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
8526 do_jump_1. do_compare_rtx_and_jump): Update prototype.
8527 * dwarf2cfi.c: Include profile-count.h
8528 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
8529 (sjlj_emit_dispatch_table): Likewise.
8530 * explow.c: Include profile-count.h
8531 * expmed.c (emit_store_flag_force): Update.
8532 (do_cmp_and_jump): Update.
8533 * expr.c (compare_by_pieces_d::generate): Update.
8534 (compare_by_pieces_d::finish_mode): Update.
8535 (emit_block_move_via_loop): Update.
8536 (store_expr_with_bounds): Update.
8537 (store_constructor): Update.
8538 (expand_expr_real_2): Update.
8539 (expand_expr_real_1): Update.
8540 * expr.h (try_casesi, try_tablejump): Update prototypes.
8541 * gimple-pretty-print.c (dump_probability): Update.
8542 (dump_profile): New.
8543 (dump_gimple_label): Update.
8544 (dump_gimple_bb_header): Update.
8545 * graph.c (draw_cfg_node_succ_edges): Update.
8546 * hsa-gen.c (convert_switch_statements): Update.
8547 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
8548 (find_if_case_1): Update.
8549 (find_if_case_2): Update.
8550 * internal-fn.c (expand_arith_overflow_result_store): Update.
8551 (expand_addsub_overflow): Update.
8552 (expand_neg_overflow): Update.
8553 (expand_mul_overflow): Update.
8554 (expand_vector_ubsan_overflow): Update.
8555 * ipa-cp.c (good_cloning_opportunity_p): Update.
8556 * ipa-split.c (split_function): Use make_single_succ_edge.
8557 * ipa-utils.c (ipa_merge_profiles): Update.
8558 * loop-doloop.c (add_test): Update.
8559 (doloop_modify): Update.
8560 * loop-unroll.c (compare_and_jump_seq): Update.
8561 (unroll_loop_runtime_iterations): Update.
8562 * lra-constraints.c (lra_inheritance): Update.
8563 * lto-streamer-in.c (input_cfg): Update.
8564 * lto-streamer-out.c (output_cfg): Update.
8565 * mcf.c (adjust_cfg_counts): Update.
8566 * modulo-sched.c (sms_schedule): Update.
8567 * omp-expand.c (expand_omp_for_init_counts): Update.
8568 (extract_omp_for_update_vars): Update.
8569 (expand_omp_ordered_sink): Update.
8570 (expand_omp_for_ordered_loops): Update.
8571 (expand_omp_for_generic): Update.
8572 (expand_omp_for_static_nochunk): Update.
8573 (expand_omp_for_static_chunk): Update.
8574 (expand_cilk_for): Update.
8575 (expand_omp_simd): Update.
8576 (expand_omp_taskloop_for_outer): Update.
8577 (expand_omp_taskloop_for_inner): Update.
8578 * omp-simd-clone.c (simd_clone_adjust): Update.
8579 * optabs.c (expand_doubleword_shift): Update.
8580 (expand_abs): Update.
8581 (emit_cmp_and_jump_insn_1): Update.
8582 (expand_compare_and_swap_loop): Update.
8583 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
8584 * predict.c (predictable_edge_p): Update.
8585 (edge_probability_reliable_p): Update.
8586 (set_even_probabilities): Update.
8587 (combine_predictions_for_insn): Update.
8588 (combine_predictions_for_bb): Update.
8589 (propagate_freq): Update.
8590 (estimate_bb_frequencies): Update.
8591 (force_edge_cold): Update.
8592 * profile-count.c (profile_count::dump): Add missing space into dump.
8593 (profile_count::debug): Add newline.
8594 (profile_count::differs_from_p): Explicitly convert to unsigned.
8595 (profile_count::stream_in): Update.
8596 (profile_probability::dump): New member function.
8597 (profile_probability::debug): New member function.
8598 (profile_probability::differs_from_p): New member function.
8599 (profile_probability::differs_lot_from_p): New member function.
8600 (profile_probability::stream_in): New member function.
8601 (profile_probability::stream_out): New member function.
8602 * profile-count.h (profile_count_quality): Rename to ...
8603 (profile_quality): ... this one.
8604 (profile_probability): New.
8605 (profile_count): Update.
8606 * profile.c (compute_branch_probabilities): Update.
8607 * recog.c (peep2_attempt): Update.
8608 * sched-ebb.c (schedule_ebbs): Update.
8609 * sched-rgn.c (find_single_block_region): Update.
8610 (compute_dom_prob_ps): Update.
8611 (schedule_region): Update.
8612 * sel-sched-ir.c (compute_succs_info): Update.
8613 * stmt.c (struct case_node): Update.
8614 (do_jump_if_equal): Update.
8615 (get_outgoing_edge_probs): Update.
8616 (conditional_probability): Update.
8617 (emit_case_dispatch_table): Update.
8618 (expand_case): Update.
8619 (expand_sjlj_dispatch_table): Update.
8620 (emit_case_nodes): Update.
8621 * targhooks.c: Update.
8622 * tracer.c (better_p): Update.
8623 (find_best_successor): Update.
8624 * trans-mem.c (expand_transaction): Update.
8625 * tree-call-cdce.c: Update.
8626 * tree-cfg.c (gimple_split_edge): Upate.
8627 (move_sese_region_to_fn): Upate.
8628 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
8629 * tree-eh.c (lower_resx): Upate.
8630 (cleanup_empty_eh_move_lp): Upate.
8631 * tree-if-conv.c (version_loop_for_if_conversion): Update.
8632 * tree-inline.c (copy_edges_for_bb): Update.
8633 (copy_cfg_body): Update.
8634 * tree-parloops.c (gen_parallel_loop): Update.
8635 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
8636 (gimple_gen_time_profiler): Update.
8637 * tree-ssa-dce.c (remove_dead_stmt): Update.
8638 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
8639 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
8640 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
8641 (unloop_loops): Update.
8642 (try_peel_loop): Update.
8643 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8644 * tree-ssa-loop-split.c (connect_loops): Update.
8645 (split_loop): Update.
8646 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
8647 (hoist_guard): Update.
8648 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
8649 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
8650 (value_replacement): Update.
8651 * tree-ssa-reassoc.c (branch_fixup): Update.
8652 * tree-ssa-tail-merge.c (replace_block_by): Update.
8653 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
8654 (create_edge_and_update_destination_phis): Update.
8655 (compute_path_counts): Update.
8656 (recompute_probabilities): Update.
8657 (update_joiner_offpath_counts): Update.
8658 (freqs_to_counts_path): Update.
8659 (duplicate_thread_path): Update.
8660 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
8661 (struct switch_conv_info): Update.
8662 (gen_inbound_check): Update.
8663 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
8664 (vect_do_peeling): Update.
8665 (vect_loop_versioning): Update.
8666 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8667 (optimize_mask_stores): Update.
8668 * ubsan.c (ubsan_expand_null_ifn): Update.
8669 * value-prof.c (gimple_divmod_fixed_value): Update.
8670 (gimple_divmod_fixed_value_transform): Update.
8671 (gimple_mod_pow2): Update.
8672 (gimple_mod_pow2_value_transform): Update.
8673 (gimple_mod_subtract): Update.
8674 (gimple_mod_subtract_transform): Update.
8675 (gimple_ic): Update.
8676 (gimple_stringop_fixed_value): Update.
8677 (gimple_stringops_transform): Update.
8678 * value-prof.h: Update.
8680 2017-06-29 Carl Love <cel@us.ibm.com>
8682 * config/rs6000/rs6000-c.c: Add support for built-in functions
8683 vector signed int vec_signed (vector float);
8684 vector signed long long vec_signed (vector double);
8685 vector signed int vec_signed2 (vector double, vector double);
8686 vector signed int vec_signede (vector double);
8687 vector signed int vec_signedo (vector double);
8688 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
8689 instruction generator.
8690 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
8691 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
8692 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
8694 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
8695 vunsignede_v2df): Add define_expands.
8696 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
8697 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
8698 VEC_UNSIGNEDO): Add definitions.
8699 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
8700 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
8701 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
8702 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
8703 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
8704 * config/rs6000/altivec.h (vec_signed, vec_signed2,
8705 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
8706 vec_unsignede, vec_unsignedo): Add builtin defines.
8707 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
8709 * doc/extend.texi: Update the built-in documentation file for the
8710 new built-in functions.
8712 2017-06-29 Richard Biener <rguenther@suse.de>
8714 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
8715 reduction chains to LOOP_VINFO_REDUCTIONS.
8716 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
8717 SLP reductions after processing reduction chains.
8719 2017-06-29 Nathan Sidwell <nathan@acm.org>
8721 * builtins.c (fold_builtin_FUNCTION): Use
8722 lang_hooks.decl_printable_name.
8724 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
8727 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
8728 with only one label.
8729 * stmt.c (expand_case): Assert NCASES is greater than one.
8731 2017-06-29 Richard Biener <rguenther@suse.de>
8733 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
8735 (group_case_labels): Likewise.
8736 (find_taken_edge): Push sanity checking on val to workers...
8737 (find_taken_edge_cond_expr): ... here
8738 (find_taken_edge_switch_expr): ... and here, handle cases
8739 with just a default label.
8740 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
8741 (group_case_labels): Likewise.
8742 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
8743 group_case_labels does anything cleanup the CFG again.
8745 2017-06-29 Bin Cheng <bin.cheng@arm.com>
8747 PR tree-optimization/81196
8748 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
8749 exit condition comparing two IVs.
8751 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
8753 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
8754 profile to the dummy entry at the end of the list of architectures.
8755 * config/arm/arm-cpu-cdata.h: Regenerated.
8757 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8758 Michael Collison <michael.collison@arm.com>
8761 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
8763 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
8764 (*aarch64_reg_<mode>3_minus_mask): New pattern.
8765 (*aarch64_<optab>_reg_di3_mask2): New pattern.
8766 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
8767 of shift when the shift amount is masked with constant equal to
8768 the size of the mode.
8769 * config/aarch64/predicates.md (subreg_lowpart_operator): New
8772 2017-06-29 Martin Liska <mliska@suse.cz>
8774 * config/i386/i386.opt: Change range from [1,5] to [0,5].
8776 2017-06-29 Yury Gribov <tetra2005@gmail.com>
8779 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
8782 (edge_growth_cache_entry::edge_growth_cache_entry): New
8784 (reset_edge_growth_cache): Update to use constructor.
8786 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8788 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
8789 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
8790 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
8792 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
8794 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
8795 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
8797 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
8799 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
8800 (*-linux-uclibc*): Add t-uclibc tmake_file.
8801 * config/t-musl: New.
8802 * config/t-uclibc: New.
8804 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
8806 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
8808 (gen_comm_data): Emit architectural setting of arch_prof.
8809 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
8811 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
8812 (armv8-m.base, armv8-m.main): Likewise.
8813 * arm-protos.h (arm_build_target): Add profile field.
8814 (arch_option): Likewise.
8815 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
8817 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
8818 arm_active_target.profile.
8820 2017-06-28 Richard Biener <rguenther@suse.de>
8823 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
8824 TYPE_OVERFLOW_WRAPS.
8825 * match.pd (negate_expr_p): Likewise.
8826 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
8827 fold_build2, not fold_binary.
8829 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
8831 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
8832 Convert memory address to Pmode.
8833 (aarch64_print_operand): Assert MEM operands are always Pmode.
8835 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
8838 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
8839 Remove redundant if.
8840 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
8841 * config/arm/aarch-common-protos.h
8842 (aarch_forward_to_shift_is_not_shifted_re): Remove.
8843 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
8845 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
8848 * multiple_target.c (create_dispatcher_calls): Set the default
8849 clone to be static, not public.
8851 2017-06-28 Richard Biener <rguenther@suse.de>
8853 * tree-vect-loop.c (vectorizable_reduction): Move special
8854 cond reduction IV var creation ...
8855 (vect_create_epilog_for_reduction): ... here. Remove induction_index
8856 parameter. Use STMT_VINFO_VECTYPE.
8857 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
8860 2017-06-28 Martin Liska <mliska@suse.cz>
8863 * ipa-visibility.c (non_local_p): Handle visibility.
8865 2017-06-28 Martin Liska <mliska@suse.cz>
8868 * common.opt: Add IntegerRange to various options.
8869 * opt-functions.awk (integer_range_info): New function.
8870 * optc-gen.awk: Add integer_range_info to cl_options struct.
8871 * opts-common.c (decode_cmdline_option): Handle
8872 CL_ERR_INT_RANGE_ARG.
8873 (cmdline_handle_error): Likewise.
8874 * opts.c (print_filtered_help): Show valid interval in
8875 when --help is provided.
8876 * opts.h (struct cl_option): Add range_min and range_max fields.
8877 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
8879 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
8881 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
8882 (x * C EQ/NE y * C): New transformation.
8884 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
8886 * genmultilib (combination_space): Accept '+' in option names.
8888 2017-06-28 Martin Liska <mliska@suse.cz>
8891 * asan.c (instrument_derefs): Bail out inner references
8892 that are hard register variables.
8894 2017-06-28 Jakub Jelinek <jakub@redhat.com>
8897 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
8898 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
8900 2017-06-28 Richard Biener <rguenther@suse.de>
8902 * tree-vectorizer.h (vect_get_vec_defs): Remove.
8903 (vect_get_slp_defs): Adjust.
8904 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
8906 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
8908 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
8909 get_initial_defs_for_reduction instead of vect_get_vec_defs.
8910 (vectorizable_reduction): Adjust.
8911 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
8913 (vect_get_slp_defs): Likewise.
8914 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
8915 (vectorizable_bswap): Adjust.
8916 (vectorizable_call): Likewise.
8917 (vectorizable_conversion): Likewise.
8918 (vectorizable_assignment): Likewise.
8919 (vectorizable_shift): Likewise.
8920 (vectorizable_operation): Likewise.
8921 (vectorizable_store): Likewise.
8922 (vectorizable_condition): Likewise.
8923 (vectorizable_comparison): Likewise.
8925 2017-06-28 Michael Collison <michael.collison@arm.com>
8928 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
8930 (arm_gen_movmemqi): Removed unused variable 'i'.
8931 Convert 'for' loop into 'while' loop.
8932 (arm_expand_prologue): Remove last unnecessary set of insn.
8933 (thumb_pop): Remove unused variable 'pushed_words'.
8934 (thumb_exit): Remove last unnecessary set of regs_to_pop.
8936 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8938 * config/s390/predicates.md: Use s390_rel_address_ok_p.
8939 * config/s390/s390-protos.h: Add prototype of
8940 s390_rel_address_ok_p.
8941 * config/s390/s390.c (s390_got_symbol): New function.
8942 (s390_rel_address_ok_p): New function.
8943 (legitimize_pic_address): Use s390_rel_address_ok_p.
8944 (s390_load_got): Use s390_got_symbol.
8945 (s390_option_override): Issue error if
8946 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
8947 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
8949 * config/s390/s390.opt: New option mpic-data-is-text-relative.
8951 2017-06-27 Andrew Pinski <apinski@cavium.com>
8953 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
8954 (X * copysign (1.0, X)): New pattern.
8955 (X * copysign (1.0, -X)): New pattern.
8956 (copysign (-1.0, CST)): New pattern.
8958 2017-06-27 Joseph Myers <joseph@codesourcery.com>
8960 * genmultilib (combination_space): Remove variable.
8961 Validate reuse rules against regular expression for any sequence
8962 of multilib options in any order.
8964 2017-06-27 Michael Collison <michael.collison@arm.com>
8966 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
8967 call aarch64_split_simd_combine.
8968 * (aarch64_combine_internal<mode>): Delete pattern.
8969 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
8970 Allow register and subreg operands.
8972 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8974 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
8975 specific need, just fallback on defaults.
8976 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
8978 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8979 Olivier Hainque <hainque@adacore.com>
8981 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
8983 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
8984 targets. Pick a default if no particular attempt applied.
8985 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
8988 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8990 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
8991 (x86_64-wrs-vxworks7): Likewise.
8993 2017-06-27 Marek Polacek <polacek@redhat.com>
8996 * ubsan.c (instrument_null): Check get_base_address's result for null.
8998 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
9000 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
9002 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
9004 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
9005 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
9006 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
9007 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
9009 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
9010 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
9011 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
9012 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
9013 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
9014 BUILT_IN_FEUPDATEENV): New builtins.
9015 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
9016 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
9017 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
9018 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
9020 (builtin_structptr_types): Adjust size.
9021 * tree.c (builtin_structptr_types): Add four entries.
9023 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
9024 Olivier Hainque <hainque@adacore.com>
9026 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
9027 (TLS_SYM): New local macro, forcing reference to __tls__ on
9028 link command lines for VxWorks 7 RTPs, triggering initialization
9030 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
9031 OS features TLS support, true for RTPs on VxWorks 7.
9032 * config/vxworks.c (vxworks_override_options): Setup emutls
9035 2017-06-27 Jakub Jelinek <jakub@redhat.com>
9037 * predict.c (test_prediction_value_range): Use -1U instead of -1
9038 to avoid narrowing conversion warning.
9039 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
9040 to avoid narrowing conversion warning.
9041 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
9043 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
9045 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
9047 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
9048 64bit configurations.
9049 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
9050 (SIZE_TYPE): Likewise.
9051 * config/vxworks.c (vxworks_emutls_var_fields): Use
9052 long_unsigned_type_node instead of unsigned_type_node as the offset
9053 field type, which is "pointer" mode in emutls.c.
9055 2017-06-27 Jakub Jelinek <jakub@redhat.com>
9058 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
9061 * gimple-fold.c (replace_call_with_call_and_fold): Handle
9062 gimple_vuse copying separately from gimple_vdef copying.
9064 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9066 * value-prof.c (free_hist): Remove call to memset and the enclosing if
9069 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
9070 Olivier Hainque <hainque@adacore.com>
9072 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
9073 for all vxworks7 targets.
9074 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
9075 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
9076 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
9077 variations for VX6/VX7 and 32/64bits later on in ...
9078 (VXWORKS_LIB_SPEC): Leverage new macros.
9079 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
9080 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
9082 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
9084 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
9085 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
9087 2017-06-26 Carl Love <cel@us.ibm.com>
9089 * config/rs6000/rs6000-c.c: Add support for built-in functions
9090 vector bool char vec_reve (vector bool char);
9091 vector signed char vec_reve (vector signed char);
9092 vector unsigned char vec_reve (vector unsigned char);
9093 vector bool int vec_reve (vector bool int);
9094 vector signed int vec_reve (vector signed int);
9095 vector unsigned int vec_reve (vector unsigned int);
9096 vector bool long long vec_reve (vector bool long long);
9097 vector signed long long vec_reve (vector signed long long);
9098 vector unsigned long long vec_reve (vector unsigned long long);
9099 vector bool short vec_reve (vector bool short);
9100 vector signed short vec_reve (vector signed short);
9101 vector double vec_reve (vector double);
9102 vector float vec_reve (vector float);
9103 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
9104 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
9105 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
9106 (altivec_vreve): New pattern.
9107 * config/rs6000/altivec.h (vec_reve): New define.
9108 * doc/extend.texi (vec_rev): Update the built-in documentation file
9109 for the new built-in functions.
9111 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9113 PR tree-optimization/71815
9114 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
9116 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
9118 (slsr_process_phi): Likewise.
9119 (replace_uncond_cands_and_profitable_phis): Don't replace a
9120 multiply candidate with a stride of 1 (copy or cast).
9121 (phi_incr_cost): Call uses_consumed_by_stmt rather than
9123 (lowest_cost_path): Likewise.
9124 (total_savings): Likewise.
9126 2017-06-26 Richard Biener <rguenther@suse.de>
9129 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
9130 Use def_builtin_pure for all gather builtins.
9132 2017-06-26 Richard Biener <rguenther@suse.de>
9134 PR tree-optimization/81203
9135 * tree-tailcall.c (find_tail_calls): Do not move stmts into
9138 2017-06-26 Marek Polacek <polacek@redhat.com>
9141 * doc/invoke.texi: Document -Wmultistatement-macros.
9143 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
9145 * doc/sourcebuild.texi (ARM-specific attributes): Document new
9146 arm_neon_ok_no_float_abi effective target.
9148 2017-06-26 Richard Biener <rguenther@suse.de>
9150 PR tree-optimization/80928
9151 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
9152 (copy_bbs): Set BB_DUPLICATED flag early.
9153 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
9155 (execute_on_shrinking_pred): Likewise.
9156 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
9157 BB_DUPLICATED blocks.
9158 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
9159 iterate over all PHIs considering removal of *gsi.
9161 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
9163 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
9166 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9168 * config/rs6000/rs6000-string.c: (expand_block_clear,
9169 do_load_for_compare, select_block_compare_mode,
9170 compute_current_alignment, expand_block_compare,
9171 expand_strncmp_align_check, expand_strn_compare,
9172 expand_block_move, rs6000_output_load_multiple)
9173 Move functions related to string/block move/compare
9175 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
9176 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
9177 for this function which is now used in two files.
9178 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
9179 * config.gcc: Add rs6000-string.o to extra_objs for
9180 targets powerpc*-*-* and rs6000*-*-*.
9182 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
9185 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
9186 32-bit, since indexed is not valid for DImode.
9187 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
9188 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
9189 (define_peephole2 for Altivec d-form load): Add 32-bit support.
9190 (define_peephole2 for Altivec d-form store): Likewise.
9193 * multiple_target.c (create_dispatcher_calls): Only create the
9194 dispatcher call if the function is the default clone of a
9197 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
9200 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
9201 a call, force the call to not be a tail call.
9203 2017-06-23 Jeff Law <law@redhat.com>
9205 * doc/contrib.texi: Add entry for Steven Pemberton's work on
9208 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
9210 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
9211 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
9212 handling for early expansion of vector shifts (sl,sr,sra,rl).
9213 (builtin_function_type): Add vector shift right instructions
9214 to the unsigned argument list.
9216 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
9218 rtl-optimizatoin/79286
9219 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
9220 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
9221 trap. PIC register plus a const unspec without offset can never trap.
9223 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
9225 * tree.h (builtin_structptr_type): New type.
9226 (builtin_structptr_types): Declare new array.
9227 * tree.c (builtin_structptr_types): New array.
9228 (free_lang_data, build_common_tree_nodes): Use it.
9230 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
9233 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
9236 2017-06-22 Matt Turner <mattst88@gmail.com>
9238 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
9239 Lake models to skylake case. Assume skylake for unknown
9240 models with clflushopt.
9242 2017-06-22 Jeff Law <law@redhat.com>
9244 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
9245 frame sizes that do not satisfy aarch64_uimm12_shift.
9247 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
9249 * profile-count.h (apply_probability,
9250 apply_scale, probability_in): Fix checks for zero.
9252 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9254 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
9255 * doc/cppdiropts.texi (-I @var{dir}): Document it.
9257 2016-06-22 Richard Biener <rguenther@suse.de>
9259 * tree-vect-loop.c (vect_model_reduction_cost): Handle
9260 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
9261 REDUC_MAX_EXPR support.
9262 (vectorizable_reduction): Likewise.
9263 (vect_create_epilog_for_reduction): Likewise.
9265 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
9267 * match.pd (A / (1 << B) -> A >> B): New.
9268 * generic-match-head.c: Include optabs-tree.h.
9269 * gimple-match-head.c: Likewise.
9270 * optabs-tree.h (target_supports_op_p): New.
9271 * optabs-tree.c (target_supports_op_p): New.
9273 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9275 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
9276 $gcc_cv_ld --help output.
9277 (gcc_cv_ld_demangle): Likewise.
9278 (gcc_cv_ld_eh_frame_hdr): Likewise.
9279 (gcc_cv_ld_pie): Likewise.
9280 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
9281 (gcc_cv_ld_buildid): Likewise.
9282 (gcc_cv_ld_sysroot): Likewise.
9283 (ld_bndplt_support): Likewise.
9284 (ld_pushpopstate_support): Likewise.
9285 * configure: Regenerate.
9286 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
9288 2017-06-21 Jakub Jelinek <jakub@redhat.com>
9291 * config/i386/sse.md (round<mode>2): Renumber match_dup and
9292 operands indexes to avoid gap between operands and match_dups.
9294 2017-06-21 Andrew Pinski <apinski@cavium.com>
9296 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
9297 Increment Arith_shift and Arith_shift_reg by 1.
9298 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
9300 * config/aarch64/aarch64.c (thunderx_tunings): Enable
9301 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
9302 (aarch64_strip_extend): Add new argument and test for it.
9303 (aarch64_cheap_mult_shift_p): New function.
9304 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
9305 add a cost if it is true.
9306 Update calls to aarch64_strip_extend.
9307 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
9309 2017-06-21 Andrew Pinski <apinski@cavium.com>
9311 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
9313 (thunderxt88): Likewise.
9314 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
9315 (thunderx_prefetch_tune): New variable.
9316 (thunderx2t99_prefetch_tune): Update for the correct values.
9317 (thunderxt88_tunings): New variable.
9318 (thunderx_tunings): Use thunderx_prefetch_tune instead of
9319 generic_prefetch_tune.
9320 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
9322 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9324 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
9325 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
9326 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
9327 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
9328 (aarch64_atomic_cas<mode>, GPI): Likewise.
9330 2017-06-21 Martin Liska <mliska@suse.cz>
9332 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
9333 statements on cold and hot labels.
9334 * predict.c (tree_estimate_probability_bb): Remove the
9335 prediction from this place.
9337 2017-06-21 Martin Liska <mliska@suse.cz>
9339 PR tree-optimization/79489
9340 * gimplify.c (maybe_add_early_return_predict_stmt): New
9342 (gimplify_return_expr): Call the function.
9343 * predict.c (tree_estimate_probability_bb): Remove handling
9345 * predict.def: Update comment about early return predictor.
9346 * gimple-predict.h (is_gimple_predict): New function.
9347 * predict.def: Change default value of early return to 66.
9348 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
9350 * passes.def: Put pass_strip_predict_hints to the beginning of
9353 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
9355 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
9356 FUNCTION_DECL declarations.
9357 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
9359 (dwaf2out_decl): Likewise.
9360 * godump.c (go_early_global_decl): Skip call to the real debug hook
9361 for FUNCTION_DECL declarations.
9362 * passes.c (rest_of_decl_compilation): Skip call to the
9363 early_global_decl debug hook for FUNCTION_DECL declarations, unless
9364 -fdump-go-spec is passed.
9366 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
9368 * config/i386/i386.c (struct builtin_isa): New field pure_p.
9369 Reorder for compactness.
9370 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
9371 (def_builtin_pure, def_builtin_pure2): New functions.
9372 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
9374 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
9376 * match.pd (nop_convert): New predicate.
9377 ((A +- CST1) +- CST2): Allow some NOP conversions.
9379 2017-06-21 Jakub Jelinek <jakub@redhat.com>
9382 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
9383 with ctors/dtors if GOVD_SHARED is set.
9385 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
9387 * config/aarch64/aarch64.md (movti_aarch64):
9388 Emit mov rather than orr.
9389 (movtf_aarch64): Likewise.
9390 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
9391 Emit mov rather than orr.
9393 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
9395 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
9396 Swap alternatives, make integer dup more expensive.
9398 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
9400 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
9401 Return true for non-tls symbols.
9403 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
9405 * config/aarch64/aarch64-cores.def (cortex-a55): New.
9406 (cortex-a75): Likewise.
9407 (cortex-a75.cortex-a55): Likewise.
9408 * config/aarch64/aarch64-tune.md: Regenerate.
9409 * doc/invoke.texi (-mtune): Document new values for -mtune.
9411 2017-06-21 Tom de Vries <tom@codesourcery.com>
9413 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
9415 (Effective-Target Keywords, Other attributes): Suggest using
9416 dg-add-options stack_size feature to get stack limit in stack_size
9417 effective target documentation.
9419 2017-06-21 Julian Brown <julian@codesourcery.com>
9420 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9422 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
9423 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
9424 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
9426 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
9427 attribute type list for neon_multiply.
9428 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
9429 attribute type list for neon_multiply.
9430 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
9431 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
9432 attribute type list for neon_multiply.
9433 * config/arm/types.md (crypto_pmull): Add.
9434 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
9435 attribute type list.
9437 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
9439 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
9442 2017-06-20 Jakub Jelinek <jakub@redhat.com>
9444 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
9445 to make sure not to dereference a NULL cost_classes_ptr pointer.
9447 2017-06-20 Carl Love <cel@us.ibm.com>
9449 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9450 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9451 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9452 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9453 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9454 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
9455 VMULOSW): New enum "unspec" values.
9456 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
9457 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
9458 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
9459 altivec_vmulosw): New patterns.
9460 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9461 VMULOSW): Add definitions.
9463 2017-06-20 Julia Koval <julia.koval@intel.com>
9465 * config/i386/i386.c: Fix rounding expand for new pattern.
9466 * config/i386/subst.md: Fix pattern (parallel -> unspec).
9468 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
9470 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
9471 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
9473 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
9475 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
9478 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
9480 * config/aarch64/aarch64-cores.def: Rearrange to sort by
9481 architecture, then by implementer ID.
9482 * config/aarch64/aarch64-tune.md: Regenerate.
9484 2017-06-20 Richard Biener <rguenther@suse.de>
9487 * fold-const.c (split_tree): Fold to type before negating.
9489 2017-06-20 David Malcolm <dmalcolm@redhat.com>
9491 * diagnostic-show-locus.c
9492 (selftest::test_fixit_deletion_affecting_newline): New function.
9493 (selftest::diagnostic_show_locus_c_tests): Call it.
9495 2017-06-20 Andreas Schwab <schwab@suse.de>
9498 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
9501 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9503 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
9504 __ARM_FEATURE_COPROC according to support.
9506 2017-06-20 Jakub Jelinek <jakub@redhat.com>
9508 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
9509 Rewritten to avoid overflow for > 32-bit pointers.
9512 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
9513 by removing enum keyword.
9514 (ubsan_type_descriptor): Likewise. Formatting fix.
9517 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
9518 splitter): Require TARGET_SSE2 in the condition.
9520 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
9523 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
9524 for doing vector set of SFmode on ISA 3.0.
9525 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
9526 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
9528 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
9529 SFmode value into a V4SF variable that was extracted from another
9530 V4SF variable without converting the element to double precision
9531 and back to single precision vector format.
9532 (vsx_insert_extract_v4sf_p9_2): Likewise.
9534 2017-06-19 Jakub Jelinek <jakub@redhat.com>
9536 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
9537 in UWHI to avoid undefined overflow.
9540 * ubsan.h (enum ubsan_encode_value_phase): New.
9541 (ubsan_encode_value): Change second argument to
9542 enum ubsan_encode_value_phase with default value of
9543 UBSAN_ENCODE_VALUE_GENERIC.
9544 * ubsan.c (ubsan_encode_value): Change second argument to
9545 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
9546 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
9547 create_tmp_var_raw instead of create_tmp_var and use a
9549 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
9550 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
9551 ubsan_encode_value callers.
9554 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
9555 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
9556 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
9558 2017-06-19 Richard Biener <rguenther@suse.de>
9561 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
9562 estimates if we changed anything.
9564 2017-06-19 Richard Biener <rguenther@suse.de>
9566 PR tree-optimization/80887
9567 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
9568 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
9569 simplified lookups, then reset mprts_hook.
9570 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
9572 (try_to_simplify): Likewise.
9574 2017-06-19 Martin Liska <mliska@suse.cz>
9577 * gimplify.c (gimplify_switch_expr):
9578 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
9580 2017-06-19 Martin Liska <mliska@suse.cz>
9582 * doc/install.texi: Document that PGO runs in 4 stages.
9584 2017-06-19 Martin Liska <mliska@suse.cz>
9587 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
9588 to dispatcher function name.
9589 * multiple_target.c (replace_function_decl): New function.
9590 (create_dispatcher_calls): Redirect both edges and references.
9592 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
9594 * profile-count.c (profile_count::dump): Dump quality.
9595 (profile_count::differs_from_p): Update for unsigned val.
9596 * profile-count.h (profile_count_quality): New enum.
9597 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
9599 2017-06-19 Richard Biener <rguenther@suse.de>
9601 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
9602 struct function as arg.
9603 (estimate_numbers_of_iterations): Export overload with loop arg.
9604 (free_numbers_of_iterations_estimates_loop): Use an overload of
9605 free_numbers_of_iterations_estimates instead.
9606 * tree-cfg.c (remove_bb): Adjust.
9607 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
9608 * tree-parloops.c (gen_parallel_loop): Likewise.
9609 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
9611 (tree_unroll_loops_completely): Likewise.
9612 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
9613 Use an overload instead and export.
9614 (estimated_loop_iterations): Adjust.
9615 (max_loop_iterations): Likewise.
9616 (likely_max_loop_iterations): Likewise.
9617 (estimate_numbers_of_iterations): Take struct function as arg
9619 (loop_exits_before_overflow): Adjust.
9620 (free_numbers_of_iterations_estimates_loop): Use an overload.
9621 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
9622 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
9624 2017-06-19 Richard Biener <rguenther@suse.de>
9627 * ipa-prop.c (find_constructor_constant_at_offset): Handle
9628 RANGE_EXPR conservatively.
9630 2017-06-16 Carl Love <cel@us.ibm.com>
9632 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9633 definitions for vec_float, vec_float2, vec_floato,
9634 vec_floate built-ins.
9635 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
9636 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
9638 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
9639 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
9640 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
9641 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
9642 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
9643 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
9644 vec_floato): Add builtin defines.
9645 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
9646 Update the built-in documentation file for the new built-in
9649 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9651 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
9652 (mthumb): Mark as the negative of -marm.
9654 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9656 * doc/invoke.texi (ARM Options, -mcpu): Document supported
9658 (ARM Options, -mtune): Document that this accepts the same
9659 extension options as -mcpu.
9660 (ARM Options, -mfpu): Document addition of -mfpu=auto.
9662 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9664 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
9665 permitted extensions.
9667 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9669 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
9670 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
9671 (armv8-m.main): Add option +nodsp.
9672 * config/arm/arm-cpu-cdata.h: Regenerated.
9674 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9676 * config/arm/t-fuchsia: New file.
9677 * config.gcc (arm*-*-fuchsia*): Use it.
9679 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9681 * config/arm/t-symbian: Rewrite for new option infrastructure.
9683 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9685 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
9686 (MULTILIB_REQUIRED): Likewise.
9688 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9690 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
9691 (MULTILIB_RESUE): Likewise.
9692 (MULTILIB_MATCHES): Likewise.
9693 (MULTLIB_REQUIRED): Likewise.
9695 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9697 * config/arm/t-rtems: Rewrite for new option framework.
9699 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9701 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
9702 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
9703 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
9704 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
9705 * config/arm/t-multilib: ... here.
9706 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
9707 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
9708 armv7-a and armv8*-a when A-profile libraries have not been built.
9709 * config/arm/t-rmprofile: Rewrite.
9711 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9713 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
9714 with a backslash. Remove the backslash after substituting unescaped
9716 * doc/fragments.texi (MULTILIB_REUSE): Document it.
9718 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9720 * config.gcc: (arm*-*-*): When building a-profile libraries, force
9721 the driver to pass through the default setting of -mfloat-abi.
9722 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
9724 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
9725 (all_feat_combs): New rule.
9726 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
9728 * config/arm/t-aprofile: Rewrite.
9730 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9732 * config/arm/arm.h (FPUTYPE_AUTO): Define.
9733 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
9734 fpu is not specified by the user/command-line.
9735 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
9736 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
9737 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
9738 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
9739 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
9740 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
9742 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9744 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
9745 * config/arm/t-arm-elf: Rewritten.
9747 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9749 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
9750 have some floating-point instructions.
9751 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
9752 (TARGET_MAYBE_HARD_FLOAT): New macro.
9753 * config/arm/arm-builtins.c (arm_init_builtins): Use
9754 TARGET_MAYBE_HARD_FLOAT.
9755 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
9757 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9759 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
9760 (configargs.h): Include it.
9761 (arm_print_hint_for_fpu_option): New function.
9762 (arm_parse_fpu_option): New function.
9763 (candidate_extension): New class.
9764 (arm_canon_for_multilib): New function.
9765 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
9766 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
9767 (ARCH_CANONICAL_SPECS): New macro.
9768 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
9770 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9772 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
9773 are set after handling multilib fragments. Set target_cpu_default2
9776 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9778 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
9780 (arm*-*-*): Set target_cpu_default2 to a quoted string.
9781 * config/arm/parsecpu.awk (check_cpu): Validate any extension
9783 (check_arch): Likewise.
9784 * config/arm/arm.c (arm_configure_build_target): Handle
9785 TARGET_CPU_DEFAULT being a string constant. Scan any feature
9786 options in the default.
9788 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9790 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
9791 when an option is an alias of another.
9792 * config/arm/parsecpu.awk (optalias): New parser token.
9793 (gen_comm_data): Mark non-alias options as such. Emit entries
9794 for extension aliases.
9795 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
9796 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
9797 (armv6kz, armv6zk, armv6t2): Likewise.
9798 (armv7): Make vfpv3-d16 an alias.
9799 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
9801 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
9802 Sort in canonical order.
9803 (armv8-a): Sort in canonical order.
9804 (armv8.1-a, armv8.2-a): Likewise.
9805 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
9807 (cortex-a9): Sort in canonical order.
9808 * config/arm/arm.c (selftests.h): Include it.
9809 (arm_test_cpu_arch_data): New function.
9810 (arm_run_self_tests): New function.
9811 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
9812 (targetm): Move declaration to the end of the file.
9813 * arm-cpu-cdata.h: Regenerated.
9815 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9817 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
9818 call to target_mode_check describing the type of option passed.
9819 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
9820 (arm_target_thumb_only): Use arm_parse_arch_option_name or
9821 arm_parse_cpu_option_name to match parameters against list of
9823 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
9824 arm_arch_core_flags data structure.
9825 * config/arm/arm-cpu_cdata.h: Regenerated.
9827 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9829 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
9831 (arm_print_hint_for_cpu_option): Likewise.
9832 (arm_print_hint_for_arch_option): Likewise.
9833 (arm_parse_cpu_option_name): Likewise.
9834 (arm_parse_arch_option_name): Likewise.
9835 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
9836 of entries in the all_fpus list.
9837 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
9838 (arm_parse_cpu_option_name): Declare.
9839 (arm_parse_arch_option_name): Declare.
9840 (arm_parse_option_features): Declare.
9841 (arm_intialize_isa): Declare.
9842 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
9844 (gen_comm_data): ... here. Make definitions non-static.
9845 * config/arm/arm-cpu-data.h: Regenerated.
9846 * config/arm/arm-cpu-cdata.h: Regenerated.
9848 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9850 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
9851 (cpu_arch_extension): New structure.
9852 (cpu_arch_option, arch_option, cpu_option): New structures.
9853 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
9855 (gen_data): Generate new format data tables.
9856 * config/arm/arm.c (cpu_tune): New structure.
9857 (cpu_option, processors): Delete.
9858 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
9859 (arm_print_hint_for_cpu_option): ... this and ...
9860 (arm_print_hint_for_arch_option): ... this.
9861 (arm_parse_arch_cpu_name): Delete. Replace with ...
9862 (arm_parse_cpu_option_name): ... this and ...
9863 (arm_parse_arch_option_name): ... this.
9864 (arm_unrecognized_feature): Change type of target parameter to
9866 (arm_parse_arch_cpu_features): Delete. Replace with ...
9867 (arm_parse_option_features): ... this.
9868 (arm_configure_build_target): Rework to use new configuration data
9870 (arm_print_tune_info): Rework for new configuration data tables.
9871 * config/arm/arm-cpu-data.h: Regenerated.
9872 * config/arm/arm-cpu.h: Regenerated.
9874 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9876 * Makefile.in (OBJS): Move sbitmap.o from here ...
9877 (OBJS-libcommon): ... to here.
9879 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9881 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
9882 (ISA_ALL_CRYPTO): New macro.
9883 (ISA_ALL_SIMD): New macro
9884 (ISA_ALL_FP): New macro.
9885 * config/arm/arm.c (fpu_bitlist): Update initializer.
9886 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
9888 (arm9e): Add fpu. Add option for nofp
9889 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
9890 (arm926ej-s, arm1026ej-s): Likewise.
9891 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
9892 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
9893 neon-fp16, neon-vfpv4, nofp and nosimd.
9894 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
9895 (cortex-a8): Add fpu. Add option for nofp.
9896 (cortex-a9): Add fpu. Add options for nosimd and nofp.
9897 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
9898 (cortex-r4f): Add fpu.
9899 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
9900 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
9902 (cortex-r8): Likewise.
9903 (cortex-m4): Add fpu. Add option for nofp.
9904 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
9905 (cortex-a17.cortex-a7): Likewise.
9906 (cortex-a32): Add fpu. Add options for crypto and nofp.
9907 (cortex-a35, cortex-a53): Likewise.
9908 (cortex-a57): Add fpu. Add option for crypto.
9909 (cortex-a72, cortex-a73): Likewise.
9910 (exynos-m1): Likewise.
9911 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
9912 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
9913 (cortex-m33): Add fpu. Add option for nofp.
9914 * config/arm/arm-cpu-cdata.h: Regenerated
9915 * config/arm/arm-cpu-data.h: Regenerated.
9917 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9919 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
9920 (armv5te, armv5tej): Likewise.
9921 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
9922 (armv7): Add options fp and vfpv3-d16.
9923 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
9924 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
9926 (armv7ve): Likewise.
9927 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
9928 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
9929 (armv8-a): Add nocrypto option.
9930 (armv8.1-a, armv8.2-a): Likewise.
9931 (armv8-m.main): add options fp, fp.dp and nofp.
9933 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9935 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
9937 (armv8-a+crc): Delete.
9938 (armv8.1-a): Add options simd, crypto and nofp.
9939 (armv8.2-a): Add options fp16, simd, crypto and nofp.
9940 (armv8.2-a+fp16): Delete.
9941 (armv8-m.main): Add option dsp.
9942 (armv8-m.main+dsp): Delete.
9943 (cortex-a8): Add fpu. Add nofp option.
9944 (cortex-a9): Add fpu. Add nofp and nosimd options.
9945 * config/arm/parsecpu.awk (gen_data): Generate option tables and
9946 link to main cpu and architecture data structures.
9947 (gen_comm_data): Only put isa attributes from the main architecture
9949 (option): New statement for architecture and CPU entries.
9950 * arm.c (struct cpu_option): New structure.
9951 (struct processors): Add entry for options.
9952 (arm_unrecognized_feature): New function.
9953 (arm_parse_arch_cpu_name): Ignore any characters after the first
9955 (arm_parse_arch_cpu_feature): New function.
9956 (arm_configure_build_target): Separate out any CPU and architecture
9957 features and parse separately. Don't error out if -mfpu=auto is
9958 used with only an architecture string.
9959 (arm_print_asm_arch_directives): New function.
9960 (arm_file_start): Call it.
9961 * config/arm/arm-cpu-cdata.h: Regenerated.
9962 * config/arm/arm-cpu-data.h: Likewise.
9963 * config/arm/arm-tables.opt: Likewise.
9965 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9967 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
9968 assembler when it is not -mfpu=auto.
9970 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9972 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
9973 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
9974 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
9975 (ASM_CPU_SPEC): Rewrite.
9976 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
9977 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
9978 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
9979 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
9980 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
9981 copied string is NUL-terminated. Also strip any characters prefixed
9983 (arm_rewrite_selected_arch): New function.
9984 (arm_rewrite_march): New function.
9986 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9988 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
9989 (x_arm_cpu_string, x_arm_tune_string): Likewise.
9990 (march, mcpu, mtune): Convert to string-based options.
9991 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
9992 (arm_parse_arch_cpu_name): New function.
9993 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
9994 identify selected architecture or CPU.
9995 (arm_option_save): New function.
9996 (TARGET_OPTION_SAVE): Redefine.
9997 (arm_option_restore): Restore string options.
9998 (arm_option_print): Print string options.
10000 2017-06-16 Martin Sebor <msebor@redhat.com>
10002 PR tree-optimization/80933
10003 PR tree-optimization/80934
10004 * builtins.c (fold_builtin_3): Do not handle bcmp here.
10005 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
10006 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
10007 (gimple_fold_builtin): Call them.
10009 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10011 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
10012 as unlikely; update profile.
10014 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10016 * predict.c (force_edge_cold): Handle declaring edges impossible
10019 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10021 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
10023 (try_unroll_loop_completely): Fix reporting.
10025 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10027 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
10029 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
10032 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
10033 if given a non-constant argument for an intrinsic which requires a
10036 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10038 * profile.c (compare_freqs): New function.
10039 (branch_prob): Sort edge list.
10040 (find_spanning_tree): Assume that the list is priority sorted.
10042 2017-06-16 Richard Biener <rguenther@suse.de>
10044 PR tree-optimization/81090
10045 * passes.def (pass_record_bounds): Remove.
10046 * tree-pass.h (make_pass_record_bounds): Likewise.
10047 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
10048 make_pass_record_bounds): Likewise.
10049 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
10050 not free niter estimates at the beginning but at the end.
10051 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
10053 2017-06-16 Richard Biener <rguenther@suse.de>
10055 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
10056 initializer to workaround ICE in host GCC 4.8.
10058 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10060 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
10062 (clone_inlined_nodes): Update.
10064 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10066 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
10067 prefetch settings, and enable prefetching by default at -O3.
10069 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10071 * config/aarch64/aarch64.c (aarch64_override_options_internal):
10072 Set flag_prefetch_loop_arrays according to tuning data.
10074 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10076 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
10077 New tune structure.
10078 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
10079 [Unrelated to main purpose of the patch] Place the pointer field last
10080 to enable type checking errors when tune structure are wrongly merged.
10081 * config/aarch64/aarch64.c (generic_prefetch_tune,)
10082 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
10083 (thunderx2t99_prefetch_tune): New tune constants.
10084 (tune_params *_tunings): Update all tunings (no functional change).
10085 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
10086 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
10087 from tunings structures.
10089 2017-06-16 Jakub Jelinek <jakub@redhat.com>
10092 * ubsan.c (instrument_null): Add T argument, use it instead
10093 of computing it based on IS_LHS.
10094 (instrument_object_size): Likewise.
10095 (pass_ubsan::execute): Adjust instrument_null and
10096 instrument_object_size callers to pass gimple_get_lhs or
10097 gimple_assign_rhs1 result to it. Use instrument_null instead of
10098 calling get_base_address and instrument_mem_ref. Handle
10099 aggregate call arguments for object-size sanitization.
10101 2017-06-16 Yury Gribov <tetra2005@gmail.com>
10103 PR tree-optimization/81089
10104 * tree-vrp.c (is_masked_range_test): Validate operands of
10107 2017-06-15 Martin Sebor <msebor@redhat.com>
10110 * dumpfile.c (dump_register): Avoid calling memset to initialize
10111 a class with a default ctor.
10112 * gcc.c (struct compiler): Remove const qualification.
10113 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
10114 * hash-table.h: Ditto.
10115 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
10117 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
10118 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
10120 * params.h (struct param_info): Make struct members non-const.
10121 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
10122 with default initialization.
10123 * vec.h (vec_copy_construct, vec_default_construct): New helper
10125 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
10126 with vec_copy_construct.
10127 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
10128 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
10129 * doc/invoke.texi (-Wclass-memaccess): Document.
10131 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10133 * emit-rtl.h (is_leaf): Update comment about local
10134 register allocator.
10136 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
10139 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
10140 for a variable to have a section before checking if the section has a
10142 Set section to.persistent if persistent attribute is set.
10143 Warn if .persistent attribute is used on an automatic variable.
10145 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
10147 PR rtl-optimization/80474
10148 * reorg.c (update_block): Do not ignore instructions in a delay slot.
10150 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
10152 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
10155 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
10157 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
10158 (casesi): Emit bounds checking as RTL.
10159 (casesi_internal_mips16_<mode>): Remove bounds checking.
10161 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
10163 * config/xtensa/xtensa.c (xtensa_option_override): Append
10164 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
10165 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
10166 xtensa_doloop_hooks): Define unconditionally.
10167 (xtensa_reorg_loops): Only call reorg_loops in the presence of
10169 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
10170 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
10171 for it in xtensa_option_override.
10172 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
10173 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
10175 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
10177 * doc/cppopts.texi: Document '-' special value to -MF.
10179 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
10181 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
10182 (cortex_a53_fconst): Likewise.
10183 (cortex_a53_fpmul): Likewise.
10184 (cortex_a53_f_load_64): Likewise.
10185 (cortex_a53_f_load_many): Likewise.
10186 (cortex_a53_advsimd_alu): Likewise.
10187 (cortex_a53_advsimd_alu_q): Likewise.
10188 (cortex_a53_advsimd_mul): Likewise.
10189 (cortex_a53_advsimd_mul_q): Likewise.
10190 (fpmac bypass): Add new bypass for fpmac-fpmac case.
10191 Add missing fmul, r2f_cvt and fconst cases.
10193 2017-06-14 Richard Biener <rguenther@suse.de>
10195 PR middle-end/81088
10196 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
10198 (fold_binary_loc): When associating do not treat pre-existing
10199 TREE_OVERFLOW on literal constants as a reason to allow
10200 TREE_OVERFLOW on associated literal constants.
10202 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
10204 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
10205 (MASK_FEATURES): New macro.
10206 * config/sparc/sparc.c (sparc_option_override): Remove the special
10207 handling of -mfpu and generalize it to all MASK_FEATURES switches.
10209 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
10211 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
10212 a division of 0 if non-call exceptions are enabled.
10214 2017-06-14 Andrew Pinski <apinski@cavium.com>
10215 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10218 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
10219 Improve vector initialization code gen for only variable case.
10221 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
10223 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
10225 2017-06-14 Richard Biener <rguenther@suse.de>
10227 PR tree-optimization/81083
10228 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
10231 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10233 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
10234 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
10235 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
10236 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
10237 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
10238 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
10240 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10242 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
10243 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
10245 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10247 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
10249 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10251 * config/rs6000/t-rtems: Don't handle SPE.
10253 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10255 * config/rs6000/t-linux: Don't handle SPE.
10257 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10259 * config/rs6000/eabispe.h: Delete file.
10261 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10263 * config/rs6000/t-spe: Delete file.
10265 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10267 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
10268 (rs6000_legitimate_offset_address_p): Return false for anything in
10269 V2SImode or V2SFmode.
10271 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10273 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
10274 except V2SF and V2SI. Rearrange the vector modes, and add comments.
10275 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
10277 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
10278 (rs6000_legitimate_offset_address_p): Ditto.
10279 (rs6000_emit_move): Ditto.
10280 (rs6000_init_builtins): Remove V4HI_type_node.
10282 2017-06-13 Martin Liska <mliska@suse.cz>
10285 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
10286 (gate_asan): Likewise.
10287 * asan.h (asan_no_sanitize_address_p): Remove the function.
10288 (sanitize_flags_p): New function.
10289 * builtins.def: Fix coding style.
10290 * common.opt: Use renamed enum value.
10291 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
10292 * doc/extend.texi: Document no_sanitize attribute.
10293 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
10294 to SANITIZE_UNDEFINED_NONDEFAULT.
10295 * gcc.c (sanitize_spec_function): Use the renamed enum value.
10296 * gimple-fold.c (optimize_atomic_compare_exchange_p):
10297 Use sanitize_flags_p.
10298 * gimplify.c (gimplify_function_tree): Likewise.
10299 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
10300 * opts.c (parse_no_sanitize_attribute): New function.
10301 (common_handle_option): Use renamed enum value.
10302 * opts.h (parse_no_sanitize_attribute): Declare.
10303 * tree.c (sanitize_flags_p): New function.
10304 * tree.h: Declared here.
10305 * tsan.c: Use sanitize_flags_p.
10306 * ubsan.c (ubsan_expand_null_ifn): Likewise.
10307 (instrument_mem_ref): Likewise.
10308 (instrument_bool_enum_load): Likewise.
10309 (do_ubsan_in_current_function): Remove the function.
10310 (pass_ubsan::execute): Use sanitize_flags_p.
10311 * ubsan.h: Remove do_ubsan_in_current_function
10312 * tree-cfg.c (print_no_sanitize_attr_value): New function.
10313 (dump_function_to_file): Use it here.
10315 2017-06-13 Martin Jambor <mjambor@suse.cz>
10317 PR tree-optimization/80803
10318 PR tree-optimization/81063
10319 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
10320 (propagate_subaccesses_across_link): Enqueue subtree whenever
10321 necessary instead of relying on the caller.
10323 2017-06-13 Martin Jambor <mjambor@suse.cz>
10325 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
10326 that have a first_link.
10327 (sort_and_splice_var_accesses): Do not check first_link before
10329 (subtree_mark_written_and_enqueue): Likewise.
10330 (propagate_all_subaccesses): Likewise and do not stop at first
10331 parent with a first_link.
10333 2017-06-13 Martin Jambor <mjambor@suse.cz>
10335 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
10338 2017-06-13 Yury Gribov <tetra2005@gmail.com>
10340 * match.pd: New pattern.
10342 2017-06-13 Yury Gribov <tetra2005@gmail.com>
10344 * tree-vrp.c (is_masked_range_test): New function.
10345 (register_edge_assert_for): Determine ranges for
10348 2017-06-13 Yury Gribov <tetra2005@gmail.com>
10350 PR tree-optimization/67328
10351 * fold-const.c (maskable_range_p): New function.
10352 (build_range_check): Generate bittests if possible.
10354 2017-06-13 Martin Liska <mliska@suse.cz>
10356 * gimple-pretty-print.c (dump_probability): Add new argument.
10357 (dump_edge_probability): Dump both probability and count.
10358 (dump_gimple_label): Likewise.
10359 (dump_gimple_bb_header): Likewise.
10361 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
10364 * config/avr/avr-devices.c: Fix indentation.
10365 * config/avr/gen-avr-mmcu-specs.c: Dito.
10367 2017-06-13 Richard Biener <rguenther@suse.de>
10369 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
10370 instead get vector type from stmt_info.
10371 (vectorizable_reduction): Adjust. Remove dead code.
10373 2017-06-13 Richard Biener <rguenther@suse.de>
10375 PR middle-end/81065
10376 * fold-const.c (extract_muldiv_1): Remove bogus distribution
10377 case of C * (x * C2 + C3).
10378 (fold_addr_of_array_ref_difference): Properly fold index difference.
10380 2017-06-12 David S. Miller <davem@davemloft.net>
10383 * config/sparc/sparc.md (return expander): Emit frame blockage if
10384 function uses alloca.
10386 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
10388 * combine.c (make_field_assignment): Check len rather than the mode
10389 precision when calling force_to_mode.
10391 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
10393 Support multilibs and devices that see flash in RAM address range.
10396 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
10397 (avr_mcu_t) <flash_pm_offset>: New field.
10398 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
10399 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
10400 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
10401 (AVR_TINY_PM_OFFSET): Remove macro.
10402 * config/avr/avr.opt (-mshort-calls): New option.
10403 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
10404 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
10405 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
10406 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
10407 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
10408 instead of avr_arch->have_jmp_call.
10409 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
10410 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
10411 avr_arch->flash_pm_offset to define.
10412 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
10413 new field flash_pm_offset. Add entry for avrxmega3.
10414 (avr_texinfo): Add entry for avrxmega3.
10415 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
10416 attiny212, attiny214,
10417 attiny412, attiny414, attiny416, attiny417,
10418 attiny814, attiny816, attiny817,
10419 attiny1614, attiny1616, attiny1617,
10420 attiny3214, attiny3216, attiny3217.
10421 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
10422 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
10423 (avr_print_operand_address) [AVR_TINY]: Same.
10424 (avr_asm_init_sections) <readonly_data_section>: Only patch
10425 callback if avr_arch->flash_pm_offset = 0.
10426 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
10427 for rodata if avr_arch->flash_pm_offset != 0.
10428 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
10429 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
10430 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
10431 (m_options): Append opt_rcall.
10432 (m_dirnames): Append dir_rcall.
10433 * config/avr/t-multilib: Regenerate.
10435 * configure.ac [target=avr]: Check whether avrxmega3 default
10436 linker description file works as needed.
10437 * configure: Regenerate.
10438 * doc/avr-mmcu.texi: Regenerate.
10439 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
10440 <__AVR_ARCH__>: Document avrxmega3 and 103.
10441 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
10442 <__AVR_SHORT_CALLS__>: Document it.
10443 <__AVR_PM_BASE_ADDRESS__>: Document it.
10444 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
10445 (AVR Variable Attributes) <progmem>: Document this is
10446 not needed for avrxmega3.
10447 (AVR Named Address Spaces) <__flash>: Dito.
10449 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
10451 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
10453 2017-06-12 Doug Rupp <rupp@adacore.com>
10455 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
10456 Append vxworks-stdint.h to the tm_file list.
10457 * config/vxworks-stdint.h: New file.
10459 2017-06-12 Martin Liska <mliska@suse.cz>
10461 PR tree-optimization/81041
10462 * tree-profile.c (gimple_gen_ic_func_profiler):
10463 Create an extra BB in profile-generate
10464 (gimple_gen_time_profiler): Likewise.
10466 2017-06-12 Jakub Jelinek <jakub@redhat.com>
10468 PR tree-optimization/81003
10469 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
10470 (update_range_test): Use it instead of force_gimple_operand_gsi.
10472 2017-06-12 Richard Biener <rguenther@suse.de>
10474 PR tree-optimization/81053
10475 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
10476 with backedge value not defined in loop. Simplify def stmt
10479 2017-06-11 Tom de Vries <tom@codesourcery.com>
10482 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
10484 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
10485 nvptx_cannot_force_const_mem.
10487 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10489 * opts.c (finish_options): Move test for flag_split_stack after
10490 it has been initialized.
10492 2017-06-11 Jason Merrill <jason@redhat.com>
10494 * tree.h (id_equal): New.
10495 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
10496 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
10497 instead of strcmp of IDENTIFIER_POINTER.
10499 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10501 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
10502 (mark_all_inlined_calls_cdtor): Fix formating.
10503 (inline_transform): Rescale profile before inlining.
10505 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10507 * cgraph.h (cgraph_edge::clone): Update prototype.
10508 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
10509 (cgraph_node::create_clone): Update.
10510 (cgraph_node::create_version_clone): Update.
10511 * tree-inline.c (copy_bb): Update.
10512 (expand_call_inline): Update.
10514 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
10516 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
10517 factored out from ...
10518 (rs6000_emit_prologue): ... here.
10520 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
10522 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
10523 factored out from ...
10524 (rs6000_emit_prologue): ... here.
10526 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10528 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
10530 (handle_missing_profiles): Fix computation of tp_first_run.
10531 (counts_to_freqs): Do not touch freqs when count is 0.
10533 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10535 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
10538 2017-06-10 Tom de Vries <tom@codesourcery.com>
10540 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
10541 attributes): Document signal effective target.
10543 2017-06-10 Tom de Vries <tom@codesourcery.com>
10545 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
10546 Document effective target stack_size.
10548 2017-06-09 David Malcolm <dmalcolm@redhat.com>
10550 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
10551 to the edit_context if they can be auto-applied.
10553 2017-06-9 Ian Lance Taylor <iant@golang.org>
10555 * opts.c (finish_options): If -fsplit-stack, disable implicit
10556 -forder-blocks-and-partition.
10557 * doc/invoke.texi (Optimize Options): Document that when using
10558 -fsplit-stack -forder-blocks-and-partition is not implicitly
10561 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10563 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
10564 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
10565 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
10566 * builtins.def (abort, trap, unreachable): Declare cold.
10567 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
10568 * tree-core.h (ECF_COLD): New.
10569 * tree.c (set_call_expr_flags): Handle ECF_COLD.
10570 (build_common_builtin_nodes): Mark unreachable and abort as cold.
10572 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10574 * predict.c (unlikely_executed_stmt_p): Cleanup.
10576 2017-06-09 Richard Biener <rguenther@suse.de>
10578 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
10579 model if the ref is always written to.
10581 2017-06-09 Tamar Christina <tamar.christina@arm.com>
10583 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
10585 2017-06-09 Tamar Christina <tamar.christina@arm.com>
10587 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
10590 2017-06-09 Tom de Vries <tom@codesourcery.com>
10593 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
10594 "target cannot support label values" when encountering LABEL_REF.
10596 2017-06-09 Martin Liska <mliska@suse.cz>
10598 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
10599 (gimple_gen_ic_func_profiler): Emit direct comparison
10600 of __gcov_indirect_call_callee with NULL.
10601 (gimple_gen_time_profiler): Change probability from
10602 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
10604 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10606 * profile.c (edge_gcov_counts): Turn to pointer.
10607 (compute_branch_probabilities, compute_branch_probabilities): Update.
10608 (branch_prob): Do not clear edge_gcov_count.
10609 * profile.h (edge_gcov_counts): Turn to pointer.
10610 (edge_gcov_count): Update.
10612 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10614 * gimple.h (gimple_check_failed): Mark cold.
10616 2017-06-09 Richard Biener <rguenther@suse.de>
10618 PR tree-optimization/66623
10619 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
10620 refactor check_reduction into two parts, properly computing
10621 whether we have to check reduction validity for outer loop
10624 2017-06-09 Richard Biener <rguenther@suse.de>
10626 PR tree-optimization/79483
10627 * graphite-scop-detection.c (order): New global.
10628 (get_order): Compute bb to order mapping that satisfies code
10629 generation constraints.
10630 (cmp_pbbs): New helper.
10631 (build_scops): Start domwalk at entry block, sort generated
10634 2017-06-09 Richard Biener <rguenther@suse.de>
10636 PR middle-end/81007
10637 * ipa-polymorphic-call.c
10638 (ipa_polymorphic_call_context::restrict_to_inner_class):
10639 Skip FIELD_DECLs with error_mark_node type.
10640 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
10643 2017-06-09 Martin Liska <mliska@suse.cz>
10645 * predict.c (struct branch_predictor): New struct.
10646 (test_prediction_value_range): New test.
10647 (predict_c_tests): New function.
10648 * selftest-run-tests.c (selftest::run_tests): Run the function.
10649 * selftest.h: Declare new tests.
10651 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
10654 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
10655 gen_add3_insn did not fail.
10656 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
10657 r0, construct that number in a temporary reg and add that reg to r0.
10658 If asked to put the result in r0 as well, fail.
10660 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
10662 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10663 for early expansion of vec_eqv.
10665 2017-06-08 Jakub Jelinek <jakub@redhat.com>
10667 PR middle-end/81005
10668 * ubsan.c (instrument_null): Avoid pointless code temporary.
10669 (pass_ubsan::execute): Instrument aggregate arguments of calls.
10671 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
10675 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
10678 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
10681 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10683 * predict.c (unlikely_executed_edge_p): Move ahead.
10684 (probably_never_executed_edge_p): Use it.
10686 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
10688 PR middle-end/79988
10689 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
10690 gimple_call_builtin_p call.
10692 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10694 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
10695 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
10696 rtl_check_failed_type2, rtl_check_failed_code1,
10697 rtl_check_failed_code2, rtl_check_failed_code_mode,
10698 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
10699 rtvec_check_failed_bounds, rtl_check_failed_flag,
10700 _fatal_insn_not_found, _fatal_insn): Likewise.
10701 * tree.h (tree_contains_struct_check_failed,
10702 tree_check_failed, tree_not_check_failed,
10703 tree_class_check_failed, tree_range_check_failed,
10704 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
10705 tree_vec_elt_check_failed, phi_node_elt_check_failed,
10706 tree_operand_check_failed, omp_clause_check_failed,
10707 omp_clause_operand_check_failed, omp_clause_range_check_failed):
10710 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10712 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
10713 flag_branch_probabilities.
10714 * ipa-inline.c (edge_badness): Likewise.
10715 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
10716 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
10717 * predict.c (maybe_hot_frequency_p): Likewise.
10718 (probably_never_executed): Likewise.
10719 * sched-ebb.c (schedule_ebbs): Likewise.
10720 * sched-rgn.c (find_single_block_region): Likewise.
10721 * tracer.c (tail_duplicate): Likewise.
10723 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10725 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
10726 longer requires x_flag_profile_use.
10728 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10730 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
10731 instead of flag_reorder_blocks_and_partition.
10732 * dbxout.c (dbxout_function_end): Likewise.
10733 * dwarf2out.c (gen_subprogram_die): Likewise.
10734 * haifa-sched.c (sched_create_recovery_edges): Likewise.
10735 * hw-doloop.c (reorg_loops): Likewise.
10736 * varasm.c (assemble_start_function,
10737 assemble_end_function): Likewise.
10738 (decide_function_section): Do not check for
10739 flag_reorder_blocks_and_partition.
10741 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
10743 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
10745 (chkp_get_hard_register_fake_addr_expr): Ditto.
10746 (chkp_build_addr_expr): Add check for hard reg case.
10747 (chkp_parse_array_and_component_ref): Ditto.
10748 (chkp_find_bounds_1): Ditto.
10749 (chkp_process_stmt): Don't generate bounds store for
10752 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10754 * predict.c (maybe_hot_bb_p): Do not check profile status.
10755 (maybe_hot_edge_p): Likewise.
10756 (probably_never_executed): Check for zero counts even if profile
10758 (unlikely_executed_edge_p): New function.
10759 (unlikely_executed_stmt_p): New function.
10760 (unlikely_executed_bb_p): New function.
10761 (set_even_probabilities): Use unlikely predicates.
10762 (combine_predictions_for_bb): Likewise.
10763 (predict_paths_for_bb): Likewise.
10764 (predict_paths_leading_to_edge): Likewise.
10765 (determine_unlikely_bbs): New function.
10766 (estimate_bb_frequencies): Use it.
10767 (compute_function_frequency): Use zero counts even if profile is
10769 * profile-count.h: Fix typo.
10771 2017-08-08 Julia Koval <julia.koval@intel.com>
10773 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
10774 _mm512_mask_cvtsepi16_storeu_epi8,
10775 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
10776 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
10777 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
10778 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
10779 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
10780 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
10781 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
10782 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
10783 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
10784 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
10785 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
10786 __builtin_ia32_pmovuswb256mem_mask,
10787 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
10788 __builtin_ia32_pmovwb512mem_mask): New builtins.
10790 2017-08-08 Julia Koval <julia.koval@intel.com>
10792 PR target/73350,80862
10793 * config/i386/subst.md (round): Fix round pattern.
10794 * config/i386/i386.c (ix86_erase_embedded_rounding):
10795 Fix erasing rounding for the fixed pattern.
10797 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10799 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
10801 2017-06-08 Martin Liska <mliska@suse.cz>
10803 PR gcov-profile/80911
10804 * gcov.c (block_info::block_info): New constructor.
10806 2017-06-07 Carl Love <cel@us.ibm.com>
10808 * config/rs6000/rs6000-c: The return type of the following
10809 built-in functions was implemented as int not long long. Fix sign
10810 of return value for the unsigned version of vec_mulo and vec_mule.
10811 vector unsigned long long vec_bperm (vector unsigned long long,
10812 vector unsigned char)
10813 vector signed long long vec_mule (vector signed int,
10815 vector unsigned long long vec_mule (vector unsigned int,
10816 vector unsigned int)
10817 vector signed long long vec_mulo (vector signed int,
10819 vector unsigned long long vec_mulo (vector unsigned int,
10820 vector unsigned int)
10821 * doc/extend.texi: Fix the documentation for the built-in
10824 2017-06-07 Carl Love <cel@us.ibm.com>
10827 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
10830 2017-06-07 Carl Love <cel@us.ibm.com>
10832 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
10833 support, Generate doublehv for signed int/float for BE case only.
10835 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
10837 * doc/invoke.texi (mcx16): Rewrite.
10839 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10841 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
10842 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
10843 *mov<mode>_softfloat, and an anonymous splitter): Use
10844 nonimmediate_operand instead of rs6000_nonimmediate_operand.
10846 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10848 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
10850 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
10851 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
10852 (rs6000_debug_reg_global): Adjust.
10853 (rs6000_init_hard_regno_mode_ok): Adjust.
10854 (rs6000_dbx_register_number): Adjust.
10855 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
10856 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
10857 Remove SPE_ACC and SPEFSCR.
10858 (REG_ALLOC_ORDER): Ditto.
10859 (FRAME_POINTER_REGNUM): Change to 111.
10860 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
10861 (REG_CLASS_NAMES): Ditto.
10862 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
10863 (REGISTER_NAMES): Ditto.
10864 (ADDITIONAL_REG_NAMES): Ditto.
10865 (rs6000_reg_names): Ditto.
10866 * config/rs6000/rs6000.md: Renumber some register number
10869 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10871 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
10873 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
10874 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
10876 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
10877 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
10878 Delete the SPE high registers.
10879 (REG_ALLOC_ORDER): Ditto.
10880 (enum reg_class): Remove SPE_HIGH_REGS.
10881 (REG_CLASS_NAMES): Ditto.
10882 (REG_CLASS_CONTENTS): Delete the SPE high registers.
10883 (REGISTER_NAMES): Ditto.
10884 (rs6000_reg_names): Ditto.
10885 * doc/tm.texi.in: Remove SPE as example.
10886 * doc/tm.texi: Regenerate.
10888 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10890 * config/rs6000/8540.md (ppc8540_brinc): Delete.
10891 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
10892 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
10893 * config/rs6000/rs6000.md (type): Remove "brinc".
10895 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10897 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
10898 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
10899 * config/rs6000/linuxspe.h: Delete file.
10900 * config/rs6000/rs6000.md: Don't include spe.md.
10901 * config/rs6000/spe.h: Delete file.
10902 * config/rs6000/spe.md: Delete file.
10903 * config/rs6000/t-rs6000: Remove spe.md.
10905 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10907 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
10908 (reg_or_none500mem_operand): Delete.
10909 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
10910 instead of reg_or_none500mem_operand.
10912 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10914 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
10915 handling of SPE flags.
10916 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
10918 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10920 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
10922 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
10923 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
10924 paired_divv2sf3): Similar.
10925 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
10926 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
10927 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
10929 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
10930 Rename the paired_* instruction patterns.
10931 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
10933 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
10934 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
10935 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
10936 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
10937 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
10938 PAIRED_VECTOR_MODE.
10939 (struct machine_function): Delete field spe_insn_chain_scanned_p.
10940 (spe_func_has_64bit_regs_p): Delete.
10941 (spe_expand_predicate_builtin): Delete.
10942 (spe_expand_evsel_builtin): Delete.
10943 (TARGET_DWARF_REGISTER_SPAN): Do not define.
10944 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
10945 (invalid_e500_subreg): Delete.
10946 (rs6000_legitimize_address): Always force_reg op2 as well, for
10947 paired single memory accesses.
10948 (rs6000_member_type_forces_blk): Delete.
10949 (rs6000_spe_function_arg): Delete.
10950 (rs6000_expand_unop_builtin): Delete SPE handling.
10951 (rs6000_expand_binop_builtin): Ditto.
10952 (spe_expand_stv_builtin): Delete.
10953 (bdesc_2arg_spe): Delete.
10954 (spe_expand_builtin): Delete.
10955 (spe_expand_predicate_builtin): Delete.
10956 (spe_expand_evsel_builtin): Delete.
10957 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
10958 (spe_init_builtins): Delete.
10959 (spe_func_has_64bit_regs_p): Delete.
10960 (savres_routine_name): Delete "info" parameter. Adjust callers.
10961 (rs6000_emit_stack_reset): Ditto.
10962 (rs6000_dwarf_register_span): Delete.
10963 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
10964 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
10965 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
10967 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
10969 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
10970 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
10971 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
10972 mulv2sf3, divv2sf3): Delete expanders.
10974 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10976 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
10978 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10980 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
10981 * config/rs6000/rs6000.c: Ditto.
10983 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10985 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
10986 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
10987 comparison_operator.
10989 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10991 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
10992 * config/rs6000/rs6000.opt: Ditto.
10993 * config/rs6000/t-rtems: Ditto.
10995 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10997 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
10998 TARGET_E500_SINGLE by 0, simplify.
10999 * config/rs6000/rs6000.c: Ditto.
11000 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
11001 (spe_build_register_parallel): Delete.
11002 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
11003 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
11004 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
11005 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
11006 (E500_CONVERT): Delete.
11007 * config/rs6000/spe.md: Remove many patterns and all define_constants.
11009 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11011 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
11012 * config/rs6000/dfp.md: Ditto.
11013 (negdd2, *negdd2_fpr): Merge.
11014 (absdd2, *absdd2_fpr): Merge.
11015 (negtd2, *negtd2_fpr): Merge.
11016 (abstd2, *abstd2_fpr): Merge.
11017 * config/rs6000/e500.h: Delete file.
11018 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
11019 TARGET_FPRS by 1 and simplify.
11020 * config/rs6000/rs6000-c.c: Ditto.
11021 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
11022 TARGET_DF_SPE by 0.
11023 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
11025 * config/rs6000/rs6000.md: Ditto.
11026 (floatdidf2, *floatdidf2_fpr): Merge.
11027 (move_from_CR_gt_bit): Delete.
11028 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
11029 (E500_CR_IOR_COMPARE): Delete.
11030 (All patterns that require !TARGET_FPRS): Delete.
11031 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
11033 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11035 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
11037 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11039 * graphds.c (add_edge): Intitialize edge's attached data.
11040 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
11041 pointer parameter. Call pointed function on each edge during
11042 graph traversing. Skip traversing the edge when the function
11044 (graphds_dfs, graphds_scc): Ditto.
11045 (for_each_edge): New parameter. Pass the new parameter to callback
11047 * graphds.h (skip_edge_callback): New function pointer type.
11048 (graphds_dfs, graphds_scc): New function pointer parameter.
11049 (graphds_edge_callback, for_each_edge): New parameter.
11051 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11053 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
11054 out code checking if runtime alias check is possible to below ...
11055 Call the new function.
11056 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
11057 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
11059 2017-06-07 Marek Polacek <polacek@redhat.com>
11062 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
11063 TYPE_OVERFLOW_WRAPS check.
11065 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11067 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
11068 if versioning is required.
11069 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
11070 peeling with the check for versioning.
11072 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11074 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
11075 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
11076 Set true to new parameter if new ssa variable is defined.
11077 (vect_gen_vector_loop_niters): Refactor. Set range information
11078 for the new vector loop bound variable.
11079 (vect_do_peeling): Ditto.
11081 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11083 * tree-affine.c (ssa.h): Include header file.
11084 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
11085 has wrapping overflow behavior.
11087 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11089 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
11091 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11093 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
11094 (tree_to_aff_combination): ... here.
11096 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11098 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
11099 reg_pressure model function.
11100 (ivopts_global_cost_for_size): Delete.
11101 (determine_set_costs, iv_ca_recount_cost): Call new model function
11102 ivopts_estimate_reg_pressure.
11104 2017-06-07 Tamar Christina <tamar.christina@arm.com>
11106 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
11107 expensive than udiv. Remove floating point cases from mod.
11109 2017-06-07 Tamar Christina <tamar.christina@arm.com>
11111 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
11112 Increase idiv cost.
11114 2017-06-07 Tamar Christina <tamar.christina@arm.com>
11116 * config/aarch64/aarch64.md
11117 (copysignsf3): Fix mask generation.
11119 2017-06-07 Jakub Jelinek <jakub@redhat.com>
11121 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
11123 (class dump_manager): Add register_dumps method.
11124 * dumpfile.c: Include langhooks.h.
11125 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
11126 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
11127 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
11128 (dump_manager::dump_register): Start with 512 entries instead of 32.
11129 (dump_manager::register_dumps): New method.
11130 * toplev.c (general_init): Instead of invoking register_dumps
11131 langhook, invoke register_dumps method on the dump manager.
11132 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
11135 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
11137 * doc/md.texi: Clarify the restrictions on a define_insn condition.
11138 Say that # requires an associated define_split to exist, and that
11139 the define_split must be suitable for use after register allocation.
11141 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11143 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
11144 (compute_outgoing_frequencies): Also initialize zero counts.
11145 (find_many_sub_basic_blocks): Do not produce uninitialized profile
11146 around loops; preserve more of profile when nothing changes.
11148 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
11150 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
11152 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
11153 * config/arm/arm-cpu-cdata.h: Regenerate.
11154 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
11155 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
11156 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
11157 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
11158 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
11160 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
11162 * config/arm/t-rmprofile: Likewise.
11163 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
11165 2017-06-06 David S. Miller <davem@davemloft.net>
11168 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
11169 blockage if function uses alloca.
11171 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11173 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
11174 New "uid" fields to hold pretty-print IDs of group and ref.
11175 Memory references are now identified as <group_id>:<ref_id>
11176 instead of using [random] addresses.
11177 (dump_mem_details): Simplify, no functional change.
11178 (dump_mem_ref): Simplify and make output more concise.
11179 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
11180 (find_or_create_group): Initialize group uid.
11181 (record_ref): Initialize ref uid. Improve debug output.
11182 (prune_group_by_reuse, should_issue_prefetch_p,)
11183 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
11184 (mark_nontemporal_store, determine_loop_nest_reuse):
11185 Improve debug output.
11187 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11189 * dbgcnt.def (prefetch): New debug counter.
11190 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
11191 (schedule_prefetches): Stop issueing prefetches if debug counter
11194 2017-06-06 Tom de Vries <tom@codesourcery.com>
11196 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
11197 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
11199 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11201 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
11202 Use aarch64_reg_or_zero predicate for operand 4.
11203 (aarch64_compare_and_swap<mode> define_insn_and_split):
11204 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
11205 (aarch64_store_exclusive<mode>): Likewise for operand 2.
11207 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
11209 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
11210 (arm_compute_save_core_reg_mask): This.
11211 (thumb1_compute_save_reg_mask): Rename into ...
11212 (thumb1_compute_save_core_reg_mask): This.
11213 (arm_compute_save_reg0_reg12_mask): Adapt comment.
11214 (arm_compute_frame_layout): Likewise.
11216 2017-06-06 Richard Biener <rguenther@suse.de>
11218 PR tree-optimization/80974
11219 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
11220 keep or clear leaders SSA info.
11222 2017-06-06 Tom de Vries <tom@codesourcery.com>
11224 * config/nvptx/nvptx.c (split_mode_p): New function.
11225 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
11227 2017-06-06 Tom de Vries <tom@codesourcery.com>
11229 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
11231 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11234 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
11237 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11239 * shrink-wrap.c (handle_simple_exit): Update profile.
11240 (try_shrink_wrapping): Upate profile.
11242 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11244 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
11245 (tree_guess_outgoing_edge_probabilities): New.
11246 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
11247 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
11249 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11251 * ipa-split.c (split_function): Initialize return bb profile.
11253 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11255 * profile.c (compute_branch_probabilities): Also initialize
11256 EXIT_BLOCK profile.
11258 2017-06-06 Richard Biener <rguenther@suse.de>
11260 PR tree-optimization/80928
11261 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
11262 (vect_analyze_loop_operations): Properly guard analysis for
11264 (vect_transform_loop): Likewise.
11265 (vect_analyze_loop_2): Also reset SLP type on PHIs.
11266 (vect_model_induction_cost): Do not cost for pure SLP.
11267 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
11268 of induction in inner loop vectorization.
11269 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
11270 (vect_get_and_check_slp_defs): Handle vect_induction_def.
11271 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
11273 (vect_analyze_slp_cost_1): Cost induction.
11274 (vect_detect_hybrid_slp_stmts): Handle PHIs.
11275 (vect_get_slp_vect_defs): Likewise.
11276 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
11277 (vect_transform_stmt): Handle SLP reductions.
11278 * tree-vectorizer.h (vectorizable_induction): Adjust.
11280 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
11282 * config/rs6000/rs6000.c (make_resolver_func): Update
11283 init_lowered_empty_function call.
11285 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
11287 * doc/invoke.texi: Document the -fprofile-abs-path option.
11288 * common.opt (fprofile-abs-path): New option.
11289 * gcov-io.h (gcov_write_filename): Declare.
11290 * gcov-io.c (gcov_write_filename): New function.
11291 * coverage.c (coverage_begin_function): Use gcov_write_filename.
11292 * profile.c (output_location): Likewise.
11294 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
11296 * shring-wrap.c: Revert accidental commit.
11298 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
11300 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
11302 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
11304 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
11306 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
11307 profile in callgraph edge.
11308 * profile-count.h (apply_probability): If THIS is 0, then result is 0
11309 (apply_scale): Likewise.
11310 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
11311 Also scale profile when inlining function with zero profile.
11312 (initialize_cfun): Update exit block profile even when it is zero.
11313 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
11314 when profile is read.
11316 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
11318 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
11319 (CLONE_*): New constants to define the processors we can generate
11320 code for with the target_clone attribute.
11321 (rs6000_clone_map): New array to identify which clone processors
11322 the current program is running on.
11323 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
11324 target_clone attribute.
11325 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
11326 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
11327 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
11328 (cpu_expand_builtin): Add support for target_clone attribute.
11329 (rs6000_valid_attribute_p): Allow "default" attribute.
11330 (get_decl_name): New debug function to simplify printing the
11331 current function name in debugging statements.
11332 (rs6000_clone_priority): New functions to support the target_clone
11333 attribute, and be able to generate code to switch between ISA 2.05
11334 through ISA 3.0 (power6 through power9).
11335 (rs6000_compare_version_priority): Likewise.
11336 (rs6000_get_function_versions_dispatcher): Likewise.
11337 (make_resolver_func): Likewise.
11338 (add_condition_to_bb): Likewise.
11339 (dispatch_function_versions): Likewise.
11340 (rs6000_generate_version_dispatcher_body): Likewise.
11341 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
11342 (fusion_gpr_load_p): Fix a spacing issue.
11343 * doc/extend.texi (Common Function Attributes): Document that the
11344 PowerPC supports the target_clone attribute.
11346 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
11348 * config/arm/arm.h: explain F symbol found in description of ARM
11349 register allocation in its legend.
11351 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
11353 * config/mips/frame-header-opt.c: Include profile-count.h.
11354 * config/riscv/riscv.c: Include profile-count.h
11356 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
11358 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
11360 (sm_set_flag_if_changed): Add bbs field.
11361 (execute_sm_if_changed_flag_set): Pass BBS.
11362 (execute_sm): Update.
11364 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11366 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
11369 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11371 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
11372 (peephole2): New peephole2 to emit the above.
11373 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
11375 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11377 * config/aarch64/aarch64.c (define_peephole2 above
11378 *sub_<shift>_<mode>): New peephole.
11380 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11382 * config/i386/i386.c (make_resolver_func): Update.
11383 * Makefile.in: Add profile-count.h and profile-count.o
11384 * auto-profile.c (afdo_indirect_call): Update to new API.
11385 (afdo_set_bb_count): Update.
11386 (afdo_propagate_edge): Update.
11387 (afdo_propagate_circuit): Update.
11388 (afdo_calculate_branch_prob): Update.
11389 (afdo_annotate_cfg): Update.
11390 * basic-block.h: Include profile-count.h
11391 (struct edge_def): Turn count to profile_count.
11392 (struct basic_block_def): Likewie.
11393 (REG_BR_PROB_BASE): Move to profile-count.h
11394 (RDIV): Move to profile-count.h
11395 * bb-reorder.c (max_entry_count): Turn to profile_count.
11396 (find_traces): Update.
11397 (rotate_loop):Update.
11398 (connect_traces):Update.
11399 (sanitize_hot_paths):Update.
11400 * bt-load.c (migrate_btr_defs): Update.
11401 * cfg.c (RDIV): Remove.
11402 (init_flow): Use alloc_block.
11403 (alloc_block): Uninitialize count.
11404 (unchecked_make_edge): Uninitialize count.
11405 (check_bb_profile): Update.
11406 (dump_edge_info): Update.
11407 (dump_bb_info): Update.
11408 (update_bb_profile_for_threading): Update.
11409 (scale_bbs_frequencies_int): Update.
11410 (scale_bbs_frequencies_gcov_type): Update.
11411 (scale_bbs_frequencies_profile_count): New.
11412 * cfg.h (update_bb_profile_for_threading): Update.
11413 (scale_bbs_frequencies_profile_count): Declare.
11414 * cfgbuild.c (compute_outgoing_frequencies): Update.
11415 (find_many_sub_basic_blocks): Update.
11416 * cfgcleanup.c (try_forward_edges): Update.
11417 (try_crossjump_to_edge): Update.
11418 * cfgexpand.c (expand_gimple_tailcall): Update.
11419 (construct_exit_block): Update.
11420 * cfghooks.c (verify_flow_info): Update.
11421 (dump_bb_for_graph): Update.
11422 (split_edge): Update.
11423 (make_forwarder_block): Update.
11424 (duplicate_block): Update.
11425 (account_profile_record): Update.
11426 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
11427 (get_estimated_loop_iterations): Update.
11428 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
11429 (single_likely_exit): Update.
11430 * cfgloopmanip.c (scale_loop_profile): Update.
11432 (set_zero_probability): Update.
11433 (lv_adjust_loop_entry_edge): Update.
11434 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
11435 (purge_dead_edges): Update.
11436 (rtl_account_profile_record): Update.
11437 * cgraph.c (cgraph_node::create): Uninitialize count.
11438 (symbol_table::create_edge): Uninitialize count.
11439 (cgraph_update_edges_for_call_stmt_node): Update.
11440 (cgraph_edge::dump_edge_flags): Update.
11441 (cgraph_node::dump): Update.
11442 (cgraph_edge::maybe_hot_p): Update.
11443 * cgraph.h: Include profile-count.h
11444 (create_clone), create_edge, create_indirect_edge): Update.
11445 (cgraph_node): Turn count to profile_count.
11446 (cgraph_edge0: Likewise.
11447 (make_speculative, clone): Update.
11448 (create_edge): Update.
11449 (init_lowered_empty_function): Update.
11450 * cgraphclones.c (cgraph_edge::clone): Update.
11451 (duplicate_thunk_for_node): Update.
11452 (cgraph_node::create_clone): Update.
11453 * cgraphunit.c (cgraph_node::analyze): Update.
11454 (cgraph_node::expand_thunk): Update.
11455 * final.c (dump_basic_block_info): Update.
11456 * gimple-streamer-in.c (input_bb): Update.
11457 * gimple-streamer-out.c (output_bb): Update.
11458 * graphite.c (print_global_statistics): Update.
11459 (print_graphite_scop_statistics): Update.
11460 * hsa-brig.c: Include basic-block.h.
11461 * hsa-dump.c: Include basic-block.h.
11462 * hsa-gen.c (T sum_slice): Update.
11463 (convert_switch_statements):Update.
11464 * hsa-regalloc.c: Include basic-block.h.
11465 * ipa-chkp.c (chkp_produce_thunks): Update.
11466 * ipa-cp.c (struct caller_statistics): Update.
11467 (init_caller_stats): Update.
11468 (gather_caller_stats): Update.
11469 (ipcp_cloning_candidate_p): Update.
11470 (good_cloning_opportunity_p): Update.
11471 (get_info_about_necessary_edges): Update.
11472 (dump_profile_updates): Update.
11473 (update_profiling_info): Update.
11474 (update_specialized_profile): Update.
11475 (perhaps_add_new_callers): Update.
11476 (decide_about_value): Update.
11477 (ipa_cp_c_finalize): Update.
11478 * ipa-devirt.c (struct odr_type_warn_count): Update.
11479 (struct decl_warn_count): Update.
11480 (struct final_warning_record): Update.
11481 (possible_polymorphic_call_targets): Update.
11482 (ipa_devirt): Update.
11483 * ipa-fnsummary.c (redirect_to_unreachable): Update.
11484 * ipa-icf.c (sem_function::merge): Update.
11485 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
11486 * ipa-inline.c (compute_uninlined_call_time): Update.
11487 (compute_inlined_call_time): Update.
11488 (want_inline_small_function_p): Update.
11489 (want_inline_self_recursive_call_p): Update.
11490 (edge_badness): Update.
11491 (lookup_recursive_calls): Update.
11492 (recursive_inlining): Update.
11493 (inline_small_functions): Update.
11494 (dump_overall_stats): Update.
11495 (dump_inline_stats): Update.
11496 * ipa-profile.c (ipa_profile_generate_summary): Update.
11497 (ipa_propagate_frequency): Update.
11498 (ipa_profile): Update.
11499 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
11500 * ipa-utils.c (ipa_merge_profiles): Update.
11501 * loop-doloop.c (doloop_modify): Update.
11502 * loop-unroll.c (report_unroll): Update.
11503 (unroll_loop_runtime_iterations): Update.
11504 * lto-cgraph.c (lto_output_edge): Update.
11505 (lto_output_node): Update.
11506 (input_node): Update.
11507 (input_edge): Update.
11508 (merge_profile_summaries): Update.
11509 * lto-streamer-in.c (input_cfg): Update.
11510 * lto-streamer-out.c (output_cfg): Update.
11511 * mcf.c (create_fixup_graph): Update.
11512 (adjust_cfg_counts): Update.
11513 (sum_edge_counts): Update.
11514 * modulo-sched.c (sms_schedule): Update.
11515 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
11516 * predict.c (maybe_hot_count_p): Update.
11517 (probably_never_executed): Update.
11518 (dump_prediction): Update.
11519 (combine_predictions_for_bb): Update.
11520 (propagate_freq): Update.
11521 (handle_missing_profiles): Update.
11522 (counts_to_freqs): Update.
11523 (rebuild_frequencies): Update.
11524 (force_edge_cold): Update.
11525 * predict.h: Include profile-count.h
11526 (maybe_hot_count_p, counts_to_freqs): UPdate.
11527 * print-rtl-function.c: Do not include cfg.h
11528 * print-rtl.c: Include basic-block.h
11529 * profile-count.c: New file.
11530 * profile-count.h: New file.
11531 * profile.c (is_edge_inconsistent): Update.
11532 (correct_negative_edge_counts): Update.
11533 (is_inconsistent): Update.
11534 (set_bb_counts): Update.
11535 (read_profile_edge_counts): Update.
11536 (compute_frequency_overlap): Update.
11537 (compute_branch_probabilities): Update; Initialize and deinitialize
11539 (branch_prob): Update.
11540 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
11541 (edge_gcov_count): New.
11542 (bb_gcov_count): New.
11543 * shrink-wrap.c (try_shrink_wrapping): Update.
11544 * tracer.c (better_p): Update.
11545 * trans-mem.c (expand_transaction): Update.
11546 (ipa_tm_insert_irr_call): Update.
11547 (ipa_tm_insert_gettmclone_call): Update.
11548 * tree-call-cdce.c: Update.
11549 * tree-cfg.c (gimple_duplicate_sese_region): Update.
11550 (gimple_duplicate_sese_tail): Update.
11551 (gimple_account_profile_record): Update.
11552 (execute_fixup_cfg): Update.
11553 * tree-inline.c (copy_bb): Update.
11554 (copy_edges_for_bb): Update.
11555 (initialize_cfun): Update.
11556 (freqs_to_counts): Update.
11557 (copy_cfg_body): Update.
11558 (expand_call_inline): Update.
11559 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11560 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
11561 (try_unroll_loop_completely): Update.
11562 (try_peel_loop): Update.
11563 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11564 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
11565 * tree-ssa-loop-split.c (connect_loops): Update.
11566 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
11567 * tree-ssa-reassoc.c (branch_fixup): Update.
11568 * tree-ssa-tail-merge.c (replace_block_by): Update.
11569 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
11570 (compute_path_counts): Update.
11571 (update_profile): Update.
11572 (recompute_probabilities): Update.
11573 (update_joiner_offpath_counts): Update.
11574 (estimated_freqs_path): Update.
11575 (freqs_to_counts_path): Update.
11576 (clear_counts_path): Update.
11577 (ssa_fix_duplicate_block_edges): Update.
11578 (duplicate_thread_path): Update.
11579 * tree-switch-conversion.c (case_bit_test_cmp): Update.
11580 (struct switch_conv_info): Update.
11581 * tree-tailcall.c (decrease_profile): Update.
11582 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11583 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11584 * value-prof.c (check_counter): Update.
11585 (gimple_divmod_fixed_value): Update.
11586 (gimple_mod_pow2): Update.
11587 (gimple_mod_subtract): Update.
11588 (gimple_ic_transform): Update.
11589 (gimple_stringop_fixed_value): Update.
11590 * value-prof.h (gimple_ic): Update.
11592 2017-06-02 Carl Love <cel@us.ibm.com>
11594 * config/rs6000/rs6000-c: Add support for built-in functions
11595 vector double vec_doublee (vector signed int);
11596 vector double vec_doublee (vector unsigned int);
11597 vector double vec_doublee (vector float);
11598 vector double vec_doubleh (vector signed int);
11599 vector double vec_doubleh (vector unsigned int);
11600 vector double vec_doubleh (vector float);
11601 vector double vec_doublel (vector signed int);
11602 vector double vec_doublel (vector unsigned int);
11603 vector double vec_doublel (vector float);
11604 vector double vec_doubleo (vector signed int);
11605 vector double vec_doubleo (vector unsigned int);
11606 vector double vec_doubleo (vector float);.
11607 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
11608 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
11610 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
11611 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
11612 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
11614 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
11615 vec_doublel, vec_doubleh.
11616 * doc/extend.texi: Update the built-in documentation file for the
11617 new built-in functions.
11619 2017-06-02 David Malcolm <dmalcolm@redhat.com>
11622 * ipa-inline-analysis.c (free_growth_caches): Set
11623 edge_removal_hook_holder to NULL after removing it.
11625 2017-06-02 Sudakshina Das <sudi.das@arm.com>
11627 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
11628 comparision with zero.
11630 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
11631 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11632 for early expansion of vec_min and vec_max builtins.
11633 (builtin_function_type): Add min/max unsigned variants to those
11634 identified as having unsigned arguments.
11636 2017-06-02 Olivier Hainque <hainque@adacore.com>
11638 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
11640 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11642 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
11643 Use VALL_F16 iterator rather than VALL.
11645 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11647 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
11648 Emit CBNZ inside loop when doing a strong exchange and comparing
11649 against zero. Generate the CC flags after the loop.
11651 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
11653 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
11654 (dl_section_ref): New.
11655 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
11656 On AIX, append an expression to subtract the size of the
11657 section length to dl_section_ref.
11659 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
11661 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11662 for early expansion of vector absolute builtins.
11664 2017-06-02 Richard Biener <rguenther@suse.de>
11666 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
11667 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
11669 2017-06-02 Richard Biener <rguenther@suse.de>
11671 PR tree-optimization/80948
11672 * tree-tailcall.c (find_tail_calls): Track stmts to move in
11673 stmt order as well.
11675 2017-06-02 Richard Biener <rguenther@suse.de>
11677 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
11679 * tree-vect-stmts.c (process_use): Do not mark backedge defs
11680 for inductions as relevant.
11682 2017-06-02 Richard Biener <rguenther@suse.de>
11684 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
11685 (vectorizable_induction): ... this. Remove dead code.
11687 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
11689 * builtins. (expand_builtin_alloca): Remove second parameter and
11690 infer its value from the first parameter instead.
11691 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
11693 2017-06-02 Jakub Jelinek <jakub@redhat.com>
11695 PR rtl-optimization/80903
11696 * loop-doloop.c (add_test): Unshare sequence.
11698 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
11700 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
11702 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
11704 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
11706 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
11707 xlogue_layout::get_instance, logue_layout::xlogue_layout,
11708 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
11709 (xlogue_layout::get_stub_rtx): Make static.
11710 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
11711 (xlogue_layout::compute_stub_managed_regs): Rename to...
11712 (xlogue_layout::count_stub_managed_regs): ...this.
11713 (xlogue_layout::is_stub_managed_reg): New function.
11714 (xlogue_layout::m_stub_names): Rename to...
11715 (xlogue_layout::s_stub_names): ...this, make static.
11716 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
11717 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
11718 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
11719 xlogue_layout::s_stub_names): Instantiate statics.
11720 (stub_managed_regs): Remove.
11721 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
11722 (disable_call_ms2sysv_xlogues): Rename to...
11723 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
11724 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
11726 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
11727 change after reload_completed.
11728 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
11730 (ix86_expand_prologue): Likewise.
11731 (ix86_expand_epilogue): Likewise.
11732 (ix86_expand_split_stack_prologue): Likewise.
11733 (ix86_compute_frame_layout): Remove frame parameter ...
11734 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
11735 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
11737 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
11738 (ix86_frame): Move from here ...
11739 * config/i386/i386.h (ix86_frame): ... to here.
11740 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
11741 complete ix86_frame data structure instead. Remove some_ld_name.
11743 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
11745 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
11746 symbols that hold a DECL_VALUE_EXPR.
11748 2017-06-01 Martin Jambor <mjambor@suse.cz>
11750 PR tree-optimization/80898
11751 * tree-sra.c (process_subtree_disqualification): Removed.
11752 (disqualify_candidate): Do not acll
11753 process_subtree_disqualification.
11754 (subtree_mark_written_and_enqueue): New function.
11755 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
11756 RHS has been disqualified and re-queue LHS if necessary. Apart
11757 from that, ignore disqualified RHS.
11759 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11761 * config/s390/s390.c (s390_emit_epilogue): Disable early return
11762 address fetch for z10 or later.
11764 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11766 * config/arc/arc.md (tst_movb): Add guard when splitting.
11768 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11770 * config/arc/arc.c (arc_can_eliminate): Test against
11771 arc_frame_pointer_needed.
11773 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11775 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
11776 to prevent store reordering.
11777 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
11778 (type): Add block type.
11779 (stack_tie): Define special instruction to be used in
11782 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11784 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
11785 constraint. It is not valid for the pattern.
11786 (noncommutative_binary_comparison): Likewise.
11788 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11790 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
11793 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11795 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
11796 be used by the reg-alloc.
11798 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11800 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
11801 reg-alloc when having mul64 or mul32x16 instructions.
11802 (mulsidi3): Likewise.
11803 (umulsidi3): Likewise.
11804 (mulsi32x16): New pattern.
11805 (mulsi64): Likewise.
11806 (mulsidi64): Likewise.
11807 (umulsidi64): Likewise.
11808 (MUL32x16_REG): Define.
11809 (mul64_600): Use MUL32x16_REG.
11810 (mac64_600): Likewise.
11811 (umul64_600): Likewise.
11812 (umac64_600): Likewise.
11814 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11816 * config/arc/arc.md (mulsi3_700): Make it commutative.
11818 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
11820 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
11822 (*sign_extendsidi2_insn): Likewise for movstosw.
11824 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
11826 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
11827 the type of the input discriminant value. Convert the
11828 discriminant value of signedness vary.
11830 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
11832 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
11833 Add to -Wall section.
11835 2017-06-01 Richard Biener <rguenther@suse.de>
11837 PR middle-end/66313
11838 * fold-const.c (fold_plusminus_mult_expr): If the factored
11839 factor may be zero use a wrapping type for the inner operation.
11840 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
11841 and handle moved defs.
11842 (process_assignment): Properly guard the unary op case. Return a
11843 tri-state indicating that moving the stmt before the call may allow
11844 to continue. Pass through to_move.
11845 (find_tail_calls): Handle moving unrelated defs before
11848 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
11851 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
11852 splitter result in the canonical way.
11854 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
11856 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
11857 also for 32bit target. Update insn attributes.
11858 (zero-extendsidi2 splitter): Allow all registers for operand 1.
11860 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
11862 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
11863 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
11864 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
11865 (_mm_maskz_min_ss): New intrinsics.
11867 2017-05-31 Martin Liska <mliska@suse.cz>
11869 * tree-vect-loop.c (vect_create_epilog_for_reduction):
11870 Change comment style to one we normally use.
11871 (vectorizable_reduction): Likewise.
11872 (vectorizable_induction): Likewise.
11873 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11874 (vectorizable_call): Likewise.
11875 (vectorizable_simd_clone_call): Likewise.
11876 (vectorizable_conversion): Likewise.
11877 (vectorizable_assignment): Likewise.
11878 (vectorizable_shift): Likewise.
11879 (vectorizable_operation): Likewise.
11880 (vectorizable_store): Likewise.
11881 (vectorizable_load): Likewise.
11882 * tree-vectorizer.h: Likewise.
11884 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
11886 * passes.c (emergency_dump_function): New.
11887 * tree-pass.h (emergency_dump_function): Declare.
11888 * plugin.c (plugins_internal_error_function): Remove.
11889 * plugin.h (plugins_internal_error_function): Remove declaration.
11890 * toplev.c (internal_error_function): New static function. Use it...
11891 (general_init): ...here.
11893 2017-05-31 Graham Markall <graham.markall@embecosm.com>
11895 * config/arc/arc.c (arc_print_operand): Handle constant operands.
11896 (arc_rtx_costs): Add costs for new patterns.
11897 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
11898 * config/arc/predicates.md: Add _1_2_3_operand predicate.
11900 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
11902 * tree-ssa-strlen.c (get_next_strinfo): New function.
11903 (get_stridx_plus_constant): Use it.
11904 (zero_length_string): Likewise.
11905 (adjust_related_strinfos): Likewise.
11906 (adjust_last_stmt): Likewise.
11908 2017-05-31 Richard Biener <rguenther@suse.de>
11911 * config/i386/i386.c (ix86_expand_builtin): Remove assert
11912 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
11914 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
11916 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
11917 loop_vinfo argument and use of dependence distance vectors.
11918 Check instead whether the two references differ only in their
11919 initial value and assume that they have the same alignment if the
11920 difference is a multiple of the vector alignment.
11921 (vect_analyze_data_refs_alignment): Update call accordingly.
11923 2017-05-31 Martin Liska <mliska@suse.cz>
11926 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
11928 2017-05-31 Bin Cheng <bin.cheng@arm.com>
11930 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
11931 (create_intersect_range_checks): Move from ...
11932 * tree-data-ref.c (create_intersect_range_checks_index)
11933 (create_intersect_range_checks): ... to here.
11934 (create_runtime_alias_checks): New function factored from ...
11935 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
11936 here. Call above function.
11937 * tree-data-ref.h (create_runtime_alias_checks): New function.
11939 2017-05-31 Bin Cheng <bin.cheng@arm.com>
11941 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
11942 segment length for dr_b and compute it in wide_int.
11944 2017-05-31 Richard Biener <rguenther@suse.de>
11946 PR tree-optimization/80906
11947 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
11948 and pass through iv_map.
11949 (copy_bb_and_scalar_dependences): Adjust.
11950 (translate_pending_phi_nodes): Likewise.
11951 (copy_loop_close_phi_args): Handle code-generating IVs instead
11954 2017-05-30 David Malcolm <dmalcolm@redhat.com>
11956 * diagnostic-color.c (color_dict): Add "type-diff".
11957 (parse_gcc_colors): Update comment.
11958 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
11959 -fdiagnostics-show-template-tree and -fno-elide-type.
11960 (GCC_COLORS): Add type-diff to example.
11962 (-fdiagnostics-show-template-tree): New.
11963 (-fno-elide-type): New.
11964 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
11965 the pp_format_decoder callback. Call any m_format_postprocessor's
11967 (pretty_printer::pretty_printer): Initialize
11968 m_format_postprocessor.
11969 (pretty_printer::~pretty_printer): Delete any
11970 m_format_postprocessor.
11971 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
11972 (class format_postprocessor): New class.
11973 (struct pretty_printer::format_decoder): Document the new parameters.
11974 (struct pretty_printer::m_format_postprocessor): New field.
11975 * tree-diagnostic.c (default_tree_printer): Update for new
11976 bool and const char ** params.
11977 * tree-diagnostic.h (default_tree_printer): Likewise.
11979 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
11981 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
11982 (lwa_operand): Delete rs6000_gen_cell_microcode test.
11983 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
11984 rs6000_gen_cell_microcode code.
11985 (rs6000_final_prescan_insn): Delete.
11986 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
11987 "warn-cell-microcode" entries.
11988 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
11989 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
11990 throughout. Change cc_reg_not_micro_cr0_operand to
11991 cc_reg_not_cr0_operand throughout.
11992 (*extendhi<mode>2_noload): Delete.
11993 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
11994 (mwarn-cell-microcode): Delete.
11995 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
11996 -mgen-cell-microcode and -mwarn-cell-microcode.
11998 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
12001 * config/i386/constraints.md (Yd): New constraint.
12003 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
12004 and (?Yd, r) alternatives. Update insn attributes.
12005 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
12006 and (?*Yd, r) alternatives. Update insn attributes.
12007 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
12009 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
12011 * gimplify.c (gimplify_modify_expr): Don't create a
12012 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
12015 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
12017 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
12019 2017-05-30 Richard Biener <rguenther@suse.de>
12021 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
12022 and reduc_def fields.
12023 (STMT_VINFO_REDUC_TYPE): New define.
12024 (STMT_VINFO_REDUC_DEF): Likewise.
12025 (vect_force_simple_reduction): Adjust prototype.
12026 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
12027 (vect_is_simple_reduction): Remove check_reduction argument.
12028 (vect_force_simple_reduction): Adjust and set
12029 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
12030 (vectorizable_reduction): Do not re-do reduction analysis
12031 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
12032 * tree-parloops.c (gather_scalar_reductions): Adjust.
12034 2017-05-30 Richard Biener <rguenther@suse.de>
12036 PR middle-end/80901
12037 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
12040 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
12042 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
12043 Introduce unknown_misalignment parameter and remove vf.
12044 (vect_peeling_hash_get_lowest_cost):
12045 Pass unknown_misalignment parameter.
12046 (vect_enhance_data_refs_alignment):
12047 Fix unsupportable data ref treatment.
12049 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
12051 * tree-vect-data-refs.c (vect_get_data_access_cost):
12052 Workaround for SLP handling.
12053 (vect_enhance_data_refs_alignment):
12054 Compute costs for doing no peeling at all, compare to the best
12055 peeling costs so far and avoid peeling if cheaper.
12057 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
12059 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
12060 Return peeling info and set costs to zero for unlimited cost
12062 (vect_enhance_data_refs_alignment): Also inspect all datarefs
12063 with unknown misalignment. Compute and costs for unknown
12064 misalignment, compare them to the costs for known misalignment
12065 and choose the cheapest for peeling.
12067 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
12069 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
12070 (vect_get_peeling_costs_all_drs): Create function.
12071 (vect_peeling_hash_get_lowest_cost):
12072 Use vect_get_peeling_costs_all_drs.
12073 (vect_peeling_supportable): Create function.
12074 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
12076 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
12078 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
12079 DR_HAS_NEGATIVE_STEP.
12080 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
12081 (vect_enhance_data_refs_alignment): Use.
12082 (vect_duplicate_ssa_name_ptr_info): Use.
12083 * tree-vectorizer.h (dr_misalignment): Use.
12084 (known_alignment_for_access_p): Use.
12086 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12089 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
12091 (has_section_name): New function.
12093 2017-05-30 Martin Liska <mliska@suse.cz>
12096 * auto-profile.c (get_function_decl_from_block): Fix
12099 2017-05-30 Richard Biener <rguenther@suse.de>
12101 PR middle-end/80876
12102 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
12104 2017-05-30 Martin Liska <mliska@suse.cz>
12106 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
12107 * dumpfile.h (struct dump_file_info): Remove ctors.
12109 2017-05-30 Martin Liska <mliska@suse.cz>
12111 * predict.def: Fix GNU coding style.
12113 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
12115 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
12116 Mark 'to' argument with ATTRIBUTE_UNUSED.
12118 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
12120 * config/xtensa/xtensa.c (xtensa_emit_call): Use
12121 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
12122 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
12125 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
12127 * doc/install.texi (Options specification): Restore entry of
12128 --enable-sjlj-exceptions.
12130 2017-05-27 Michael Eager <eager@eagercon.com>
12133 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
12135 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
12137 * config/microblaze/microblaze.h
12138 (FIXED_REGISTERS): Update in macro.
12139 (CALL_USED_REGISTERS): Update in macro.
12141 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
12143 * doc/install.texi: Add links to macOS binary distributions.
12145 2017-05-27 Jakub Jelinek <jakub@redhat.com>
12149 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
12151 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
12153 2017-05-26 Martin Liska <mliska@suse.cz>
12155 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
12157 2017-05-26 Martin Liska <mliska@suse.cz>
12159 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
12160 always leading ';; '.
12161 (dump_bb_info): Likewise.
12162 (brief_dump_cfg): Likewise.
12163 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
12164 * dumpfile.c: Remove usage of TDF_VERBOSE.
12165 * dumpfile.h (enum dump_kind): Likewise.
12166 (dump_gimple_bb_header): Do not use TDF_COMMENT.
12167 * print-tree.c (debug_verbose): Remove.
12168 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
12169 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
12170 * tree-diagnostic.c (default_tree_printer): Replace
12171 TDF_DIAGNOSTIC with TDF_SLIM.
12173 2017-05-26 Bin Cheng <bin.cheng@arm.com>
12175 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
12176 in parameter loop, rather than loop_vinfo.
12177 (create_intersect_range_checks): Ditto.
12178 (vect_create_cond_for_alias_checks): Update call to above functions.
12180 2017-05-26 Bin Cheng <bin.cheng@arm.com>
12182 PR tree-optimization/80815
12183 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
12184 for merging runtime alias checks. Handle negative DR_STEPs.
12186 2017-05-26 Bin Cheng <bin.cheng@arm.com>
12188 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
12190 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
12191 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
12192 out code pruning runtime alias checks.
12193 * tree-data-ref.c (prune_runtime_alias_test_list): New function
12194 factored out from above.
12195 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
12197 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
12199 (prune_runtime_alias_test_list): New decalaration.
12201 2017-05-26 Bin Cheng <bin.cheng@arm.com>
12203 * tree-vect-data-refs.c (compare_tree): Rename and move ...
12204 * tree-data-ref.c (data_ref_compare_tree): ... to here.
12205 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
12206 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
12207 (operator==, comp_dr_with_seg_len_pair): Ditto.
12208 (vect_prune_runtime_alias_test_list): Ditto.
12210 2017-05-26 Martin Liska <mliska@suse.cz>
12213 * params.def: Bound partial-inlining-entry-probability param.
12215 2017-05-26 Marek Polacek <polacek@redhat.com>
12218 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
12221 2017-05-26 Richard Biener <rguenther@suse.de>
12223 PR tree-optimization/80842
12224 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
12227 2017-05-26 Richard Biener <rguenther@suse.de>
12229 PR tree-optimization/80844
12230 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
12232 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
12234 * doc/md.texi (Machine Constraints): Update x86 family
12235 machine constraints section to match 'config/i386/constraints.md'.
12237 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
12239 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
12241 2017-05-25 Nathan Sidwell <nathan@acm.org>
12243 * doc/invoke.texi (--enable-languages): Update documentation.
12245 2017-05-25 Martin Liska <mliska@suse.cz>
12247 * dumpfile.c: Add TDF_FOLDING.
12248 * dumpfile.h (enum dump_kind): Likewise.
12249 * genmatch.c (dt_simplify::gen_1): Use it.
12251 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
12253 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
12255 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
12257 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
12258 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
12260 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
12262 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
12263 * match.pd (X == C): Rewrite it here.
12264 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
12265 with_certain_nonzero_bits2): New predicates.
12266 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
12268 2017-05-24 Nathan Sidwell <nathan@acm.org>
12270 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
12273 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
12276 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
12278 * config/powerpcspe: New port. Files are copied from the rs6000
12279 port, with "rs6000" in filenames replaced by "powerpcspe".
12281 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
12283 PR rtl-optimization/80754
12284 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
12286 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
12288 * config/sparc/sparc.md (length): Return the correct value for -mflat
12289 sibcalls to match output_sibcall.
12291 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
12295 * config/rs6000/rs6000.c (struct machine_function): Add new field
12297 (rs6000_get_separate_components): Init that field, use it.
12298 (rs6000_components_for_bb): Use the field.
12300 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
12302 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
12304 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
12306 PR middle-end/80823
12307 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
12309 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12312 * config/s390/s390.c (s390_check_qrst_address): Check incoming
12313 address against address_operand predicate.
12314 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
12316 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
12318 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
12319 parameters passed indirectly.
12321 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
12323 * config/i386/i386.md (*movdi_internal): Remove SSE4
12324 alternative 18 (?r, *v). Update insn attributes.
12325 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
12326 Update insn attributes.
12327 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
12328 Update insn attributes.
12329 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
12330 alternative 1 (r, v). Remove isa attribute.
12331 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
12332 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
12333 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
12335 2017-05-23 Tom de Vries <tom@codesourcery.com>
12337 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
12340 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12342 * cgraphunit.c (symbol_table::process_new_functions): Update.
12343 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
12344 (inline_generate_summary): Rename to ...
12345 (ipa_fn_summary_generate): ... this one.
12346 (inline_read_summary): Rename to ...
12347 (ipa_fn_summary_read): ... this one.
12348 (inline_write_summary): Rename to ...
12349 (ipa_fn_summary_write): ... this one.
12350 (inline_free_summary): Rename to ...
12351 (ipa_free_fn_summary): ... this one.
12352 (pass_data_local_fn_summary, pass_local_fn_summary,
12353 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
12354 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
12355 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
12356 make_pass_ipa_fn_summary): New.
12357 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
12358 inline_write_summary, inline_free_summary): Remove.
12359 (ipa_free_fn_summary) : New.
12360 * ipa-inline.c (ipa_inline): Update.
12361 (pass_ipa_inline): Do not generate summaries.
12362 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
12364 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
12365 and add pass_ipa_fn_summary.
12366 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
12368 (make_pass_inline_parameters): Remove.
12370 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
12372 * omp-low.c (struct omp_context): Remove "default_kind" member.
12375 * omp-offload.c (execute_oacc_device_lower): Remove the
12376 parallelism dimensions function attributes for unparallelized
12377 OpenACC kernels constructs.
12379 2017-05-23 Martin Liska <mliska@suse.cz>
12381 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
12383 (cgraph_edge::make_speculative): Likewise.
12384 (cgraph_edge::resolve_speculation): Likewise.
12385 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
12386 (cgraph_node::dump): Likewise.
12387 * cgraph.h: Likewise.
12388 * cgraphunit.c (analyze_functions): Likewise.
12389 (symbol_table::compile): Likewise.
12390 * ipa-cp.c (print_all_lattices): Likewise.
12391 (determine_versionability): Likewise.
12392 (initialize_node_lattices): Likewise.
12393 (ipcp_verify_propagated_values): Likewise.
12394 (estimate_local_effects): Likewise.
12395 (update_profiling_info): Likewise.
12396 (create_specialized_node): Likewise.
12397 (perhaps_add_new_callers): Likewise.
12398 (decide_about_value): Likewise.
12399 (decide_whether_version_node): Likewise.
12400 (identify_dead_nodes): Likewise.
12401 (ipcp_store_bits_results): Likewise.
12402 * ipa-devirt.c (dump_targets): Likewise.
12403 (ipa_devirt): Likewise.
12404 * ipa-icf.c (sem_item::dump): Likewise.
12405 (sem_function::equals): Likewise.
12406 (sem_variable::equals): Likewise.
12407 (sem_item_optimizer::read_section): Likewise.
12408 (sem_item_optimizer::execute): Likewise.
12409 (congruence_class::dump): Likewise.
12410 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
12411 (dump_inline_summary): Likewise.
12412 (estimate_node_size_and_time): Likewise.
12413 (inline_analyze_function): Likewise.
12414 * ipa-inline-transform.c (inline_call): Likewise.
12415 * ipa-inline.c (report_inline_failed_reason): Likewise.
12416 (want_early_inline_function_p): Likewise.
12417 (edge_badness): Likewise.
12418 (update_edge_key): Likewise.
12419 (inline_small_functions): Likewise.
12420 * ipa-profile.c (ipa_profile): Likewise.
12421 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
12422 (ipa_make_edge_direct_to_target): Likewise.
12423 (remove_described_reference): Likewise.
12424 (ipa_impossible_devirt_target): Likewise.
12425 (propagate_controlled_uses): Likewise.
12426 (ipa_print_node_params): Likewise.
12427 (ipcp_transform_function): Likewise.
12428 * ipa-pure-const.c (pure_const_read_summary): Likewise.
12429 (propagate_pure_const): Likewise.
12430 * ipa-reference.c (generate_summary): Likewise.
12431 (read_write_all_from_decl): Likewise.
12432 (propagate): Likewise.
12433 (ipa_reference_read_optimization_summary): Likewise.
12434 * ipa-utils.c (ipa_merge_profiles): Likewise.
12435 * ipa.c (walk_polymorphic_call_targets): Likewise.
12436 (symbol_table::remove_unreachable_nodes): Likewise.
12437 (ipa_single_use): Likewise.
12438 * passes.c (execute_todo): Likewise.
12439 * predict.c (drop_profile): Likewise.
12440 * symtab.c (symtab_node::get_dump_name): New function.
12441 (symtab_node::dump_name): Likewise.
12442 (symtab_node::dump_asm_name): Likewise.
12443 (symtab_node::dump_references): Likewise.
12444 (symtab_node::dump_referring): Likewise.
12445 (symtab_node::dump_base): Likewise.
12446 (symtab_node::debug_symtab): Likewise.
12447 * tree-sra.c (convert_callers_for_node): Likewise.
12448 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
12449 * value-prof.c (init_node_map): Likewise.
12451 2017-05-23 Martin Liska <mliska@suse.cz>
12453 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
12454 and symtab_node::debug_symtab to symbol_table::debug.
12455 * cgraphunit.c (analyze_functions): Use the renamed function.
12456 (symbol_table::compile): Likewise.
12457 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
12458 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
12459 * passes.c (execute_todo): Likewise.
12460 * symtab.c (symbol_table::dump): New function.
12461 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
12463 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12465 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
12466 that nonconst implies exec.
12468 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12470 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
12471 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
12472 (inline_edge_summary_vec): Turn into ...
12473 (ipa_call_summaries): ... this one.
12474 (redirect_to_unreachable, edge_set_predicate,
12475 evaluate_properties_for_edge, inline_summary_alloc,
12476 reset_ipa_call_summary, reset_inline_summary,
12477 inline_summary_t::duplicate): Update.
12478 (inline_edge_duplication_hook): Turn to ...
12479 (ipa_call_summary_t::duplicate): ... this one.
12480 (inline_edge_removal_hook): Turn to ...
12481 (ipa_call_summary_t::remove): ... this one.
12482 (dump_inline_edge_summary): Turn to ...
12483 (dump_ipa_call_summary): ... this one.
12484 (estimate_function_body_sizes): Update.
12485 (inline_update_callee_summaries): Update.
12486 (remap_edge_change_prob): Update.
12487 (remap_edge_summaries): Update.
12488 (inline_merge_summary): Update.
12489 (do_estimate_edge_time): Update.
12490 (inline_generate_summary): Update.
12491 (inline_read_section): Update.
12492 (inline_read_summary): Update.
12493 (inline_free_summary): Update.
12494 * ipa-inline.c (can_inline_edge_p): Update.
12495 (compute_inlined_call_time): Update.
12496 (want_inline_small_function_p): Update.
12497 (edge_badness): Update.
12498 (early_inliner): Update.
12499 * ipa-inline.h (inline_edge_summary): Turn to ...
12500 (ipa_call_summary): ... this one.
12501 (ipa_call_summary_t): New class.
12502 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
12503 (ipa_call_summaries): New.
12504 (inline_edge_summary): Remove.
12505 (estimate_edge_growth): Update.
12506 * ipa-profile.c (ipa_propagate_frequency_1): Update.
12507 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
12508 * ipa-split.c (execute_split_functions): Update.
12509 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
12511 2017-05-23 Tom de Vries <tom@codesourcery.com>
12513 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
12514 attributes): Document rdrand effective target.
12516 2017-05-23 Tom de Vries <tom@codesourcery.com>
12518 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
12519 attributes): Sort alphabetically.
12521 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
12523 * config/avr/genmultilib.awk: Use gsub instead of gensub.
12525 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
12528 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
12529 V2DF/V2DI splat into two separate patterns, one that handles
12530 registers, and the other that only handles memory. Drop support
12531 for splatting from a GPR on ISA 2.07 and then splitting the
12532 splat into direct move and splat.
12533 (vsx_splat_<mode>_reg): Likewise.
12534 (vsx_splat_<mode>_mem): Likewise.
12536 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
12538 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
12540 2017-05-22 Jakub Jelinek <jakub@redhat.com>
12542 PR middle-end/80809
12543 * omp-low.c (finish_taskreg_remap): New function.
12544 (finish_taskreg_scan): If unit size of ctx->record_type
12545 is non-constant, unshare the size expression and replace
12546 decls in it with possible outer var refs.
12548 PR middle-end/80809
12549 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
12550 GOVD_SHARED rather than GOVD_PRIVATE with it.
12551 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
12552 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
12554 PR middle-end/80853
12555 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
12556 as last argument to build_outer_var_ref for pointer bases of array
12557 section reductions.
12559 2017-05-19 Martin Sebor <msebor@redhat.com>
12561 * print-tree.c (print_node): Print DECL_READ_P flag.
12563 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12565 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
12566 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
12567 * cgraph.c: Likewise.
12568 * cgraphunit.c: Likewise.
12569 * gengtype.c: Likewise.
12570 * ipa-cp.c: Likewise.
12571 * ipa-devirt.c: Likewise.
12572 * ipa-icf.c: Likewise.
12573 * ipa-predicate.c: Likewise.
12574 * ipa-profile.c: Likewise.
12575 * ipa-prop.c: Likewise.
12576 * ipa-split.c: Likewise.
12578 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
12579 edge_predicate_pool, dump_inline_hints,
12580 inline_summary::account_size_time, redirect_to_unreachable,
12581 edge_set_predicate, set_hint_predicate,
12582 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
12583 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
12584 inline_summary_t::remove, remap_hint_predicate_after_duplication,
12585 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
12586 ipa_call_summary_t::remove, initialize_growth_caches,
12587 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
12588 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
12589 mark_modified, unmodified_parm_1, unmodified_parm,
12590 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
12591 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
12592 compute_bb_predicates, will_be_nonconstant_expr_predicate,
12593 will_be_nonconstant_predicate, record_modified_bb_info,
12594 get_minimal_bb, record_modified, param_change_prob,
12595 phi_result_unknown_predicate, predicate_for_phi_result,
12596 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
12597 estimate_function_body_sizes, compute_inline_parameters,
12598 compute_inline_parameters_for_curren, pass_data_inline_parameters,
12599 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
12600 inline_update_callee_summaries, remap_edge_change_prob,
12601 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
12602 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
12603 inline_analyze_function, inline_summary_t::insert,
12604 inline_generate_summary, read_ipa_call_summary, inline_read_section,
12605 inline_read_summary, write_ipa_call_summary, inline_write_summary,
12606 inline_free_summary): Move to ipa-fnsummary.h
12607 (predicate_t): Remove.
12608 * ipa-fnsummary.c: New file.
12609 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
12610 (enum inline_hints_vals, inline_hints, agg_position_info,
12611 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
12612 inline_summaries, ipa_call_summary, ipa_call_summary_t,
12613 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
12614 dump_inline_summary, dump_inline_hints, inline_generate_summary,
12615 inline_read_summary, inline_write_summary, inline_free_summary,
12616 inline_analyze_function, initialize_inline_failed,
12617 inline_merge_summary, inline_update_overall_summary,
12618 compute_inline_parameters): Move to ipa-fnsummary.h
12619 * ipa-fnsummary.h: New file.
12620 * ipa-inline-transform.h: Include ipa-inline.h.
12621 * ipa-inline.c: LIkewise.
12623 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12625 * ipa-inline.c (edge_badness): Use inlined_time instead of
12626 inline_summaries->get.
12628 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12630 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
12632 2017-05-22 Nathan Sidwell <nathan@acm.org>
12634 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
12635 (fdump-lang): Document 'raw' option.
12636 * dumpfile.h (TDI_tu): Delete.
12637 * dumpfile.c (dump_files): Remove translation-unit.
12638 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
12640 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
12642 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
12643 command option from $(AWK) call.
12644 * config/avr/genmultilib.awk: Simplify and rewrite so that it
12645 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
12646 [FORMAT]: Remove handling of variable.
12647 * config/avr/t-multilib: Regenerate.
12649 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12651 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
12653 (dump_inline_summary): Do not print self_time.
12654 (estimate_function_body_sizes): Do not set self_time.
12655 (compute_inline_parameters): Likewise.
12656 (inline_read_section, inline_write_summary): Do not stream self_time.
12657 * ipa-inline.h (inline_summary): Drop self_time.
12659 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12661 * ipa-inline-analysis.c (account_size_time): Rename to ...
12662 (inline_summary::account_size_time): ... this one.
12663 (reset_ipa_call_summary): Turn to ...
12664 (ipa_call_summary::reset): ... this one.
12665 (reset_inline_summary): Turn to ...
12666 (inline_summary::reset): ... this one.
12667 (inline_summary_t::remove): Update.
12668 (inline_summary_t::duplicate): Update.
12669 (ipa_call_summary_t::remove): Update.
12670 (dump_inline_summary): Update.
12671 (estimate_function_body_sizes): Update.
12672 (compute_inline_parameters): Update.
12673 (estimate_node_size_and_time): Update.
12674 (inline_merge_summary): Update.
12675 (inline_update_overall_summary): Update.
12676 (inline_read_section): Update.
12677 (inline_write_summary): Update.
12678 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
12679 add account_size_time and reset member functions.
12680 (ipa_call_summary): Add reset function.
12681 * ipa-predicate.h (predicate::operator &): Constify.
12683 2017-05-22 Richard Biener <rguenther@suse.de>
12685 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
12687 2017-05-19 Jason Merrill <jason@redhat.com>
12689 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
12691 2017-05-19 Marek Polacek <polacek@redhat.com>
12694 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
12695 TYPE_OVERFLOW_WRAPS checks.
12697 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
12699 * tree-core.h (enum omp_clause_default_kind): Add
12700 "OMP_CLAUSE_DEFAULT_PRESENT".
12701 * tree-pretty-print.c (dump_omp_clause): Handle it.
12702 * gimplify.c (enum gimplify_omp_var_data): Add
12703 "GOVD_MAP_FORCE_PRESENT".
12704 (gimplify_adjust_omp_clauses_1): Map it to
12705 "GOMP_MAP_FORCE_PRESENT".
12706 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
12708 * gimplify.c (oacc_default_clause): Clarify.
12710 2017-05-19 Nathan Sidwell <nathan@acm.org>
12712 LANG_HOOK_REGISTER_DUMPS
12713 * toplev.c (general_init): Call register dump lang hook.
12714 * doc/invoke.texi: Document -fdump-lang option family.
12715 * dumpfile.c (dump_files): Remove class dump here.
12716 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
12717 * dumpfile.h (tree_dump_index): Remove TDI_class.
12718 * langhooks-def.h (lhd_register_dumps): Declare.
12719 (LANG_HOOKS_REGISTER_DUMPS): Define.
12720 (LANG_HOOKS_INITIALIZER): Add it.
12721 * langhooks.c (lhd_register_dumps): Define.
12722 * langhooks.h (struct lang_hooks): Add register_dumps.
12724 2017-05-19 Nathan Sidwell <nathan@acm.org>
12726 * context.h (context::set_passes): New.
12727 * context.c (context::context): Do not create pass manager.
12728 * toplev.c (general_init): Create pass manager here.
12730 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
12732 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
12733 use this splitter if two add or or instructions would also work for
12734 the constant we want to generate.
12736 2017-05-19 Richard Biener <rguenther@suse.de>
12739 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
12740 predicate evaluation.
12742 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
12744 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
12746 * ipa-inline.c (want_inline_small_function_p): Do not cast to
12749 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
12751 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
12752 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
12753 (inline_edge_summary_vec): Turn into ...
12754 (ipa_call_summaries): ... this one.
12755 (redirect_to_unreachable, edge_set_predicate,
12756 evaluate_properties_for_edge, inline_summary_alloc,
12757 reset_ipa_call_summary, reset_inline_summary,
12758 inline_summary_t::duplicate): Update.
12759 (inline_edge_duplication_hook): Turn to ...
12760 (ipa_call_summary_t::duplicate): ... this one.
12761 (inline_edge_removal_hook): Turn to ...
12762 (ipa_call_summary_t::remove): ... this one.
12763 (dump_inline_edge_summary): Turn to ...
12764 (dump_ipa_call_summary): ... this one.
12765 (estimate_function_body_sizes): Update.
12766 (inline_update_callee_summaries): Update.
12767 (remap_edge_change_prob): Update.
12768 (remap_edge_summaries): Update.
12769 (inline_merge_summary): Update.
12770 (do_estimate_edge_time): Update.
12771 (inline_generate_summary): Update.
12772 (inline_read_section): Update.
12773 (inline_read_summary): Update.
12774 (inline_free_summary): Update.
12775 * ipa-inline.c (can_inline_edge_p): Update.
12776 (compute_inlined_call_time): Update.
12777 (want_inline_small_function_p): Update.
12778 (edge_badness): Update.
12779 (early_inliner): Update.
12780 * ipa-inline.h (inline_edge_summary): Turn to ...
12781 (ipa_call_summary): ... this one.
12782 (ipa_call_summary_t): New class.
12783 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
12784 (ipa_call_summaries): New.
12785 (inline_edge_summary): Remove.
12786 (estimate_edge_growth): Update.
12787 * ipa-profile.c (ipa_propagate_frequency_1): Update.
12788 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
12789 * ipa-split.c (execute_split_functions): Update.
12790 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
12792 2017-05-19 Richard Biener <rguenther@suse.de>
12794 PR middle-end/80764
12795 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
12797 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
12799 * config/rs6000/rs6000.c (struct machine_function): Add field
12800 fpr_is_wrapped_separately.
12801 (rs6000_get_separate_components): Use 64 components. Handle the
12802 new FPR components.
12803 (rs6000_components_for_bb): Handle the FPR components.
12804 (rs6000_emit_prologue_components): Handle the FPR components.
12805 (rs6000_emit_epilogue_components): Handle the FPR components.
12806 (rs6000_set_handled_components): Handle the FPR components.
12807 (rs6000_emit_prologue): Don't output prologue code for those FPRs
12808 that are already separately shrink-wrapped.
12809 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
12810 that are already separately shrink-wrapped.
12812 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
12815 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
12818 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
12819 (define_peephole2 for Altivec d-form load): Add peepholes to catch
12820 cases where the register allocator uses a move and an offsettable
12821 memory operation to/from a FPR register on ISA 2.06/2.07.
12822 (define_peephole2 for Altivec d-form store): Likewise.
12824 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
12827 * config/i386/mmx.md (*mov<mode>_internal): Enable
12828 alternatives 11, 12, 13 and 14 also for 32bit targets.
12829 Remove alternatives 15, 16, 17 and 18.
12830 * config/i386/sse.md (vec_concatv2di): Change
12831 alternative (!x, *y) to (x, ?!*Yn).
12833 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
12835 * dumpfile.h (enum dump_kind): Remove stray comma.
12837 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
12839 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
12840 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
12841 predicate::num_conditions
12842 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
12843 (CHANGED): turn into predicate::changed.
12844 (agg_position_info): Move to ipa-predicate.h
12845 (add_condition, predicate::add_clause, predicate::operator &=,
12846 predicate::or_with, predicate::evaluate, predicate::probability,
12847 dump_condition, dump_clause, predicate::dump,
12848 predicate::remap_after_duplication, predicate::remap_after_inlining,
12849 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
12850 (evaluate_conditions_for_known_args): Update.
12851 (set_cond_stmt_execution_predicate): Update.
12852 * ipa-inline.h: Include ipa-predicate.h
12853 (condition, inline_param_summary, conditions, agg_position_info,
12854 predicate): Move to ipa-predicate.h
12855 * ipa-predicate.c: New file.
12856 * ipa-predicate.h: New file.
12858 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
12860 * final.c (leaf_function_p): Check we are not in a sequence.
12862 2017-05-18 Martin Liska <mliska@suse.cz>
12864 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
12865 * dumpfile.c (dump_register): Use new enum dump_kind.
12866 (get_dump_file_name): Likewise.
12867 (dump_enable_all): Likewise.
12868 (dump_switch_p_1): Likewise.
12869 (enable_rtl_dump_file): Remove usage of TDF_RTL.
12870 * dumpfile.h (enum dump_kind): New enum type.
12871 (struct dump_file_info): Create constructor and
12872 format fields and comments.
12873 * passes.c (pass_manager::register_one_dump_file):
12875 * statistics.c (statistics_early_init): Likewise.
12876 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
12877 TDF_TREE with TDF_SLIM.
12878 (gather_memory_references_ref): Likewise.
12880 2017-05-18 Martin Liska <mliska@suse.cz>
12882 * vec.h (struct vnull): Use it.
12884 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
12886 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
12887 (true_predicate, false_predicate, true_predicate_p,
12888 false_predicate_p): Remove.
12889 (single_cond_predicate, not_inlined_predicate): Turn to member function
12891 (add_condition): Update.
12892 (add_clause): Turn to...
12893 (predicate::add_clause): ... this one; update; allow passing NULL
12895 (and_predicates): Turn to ...
12896 (predicate::operator &=): ... this one.
12897 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
12898 (or_predicates): Turn to ...
12899 (predicate::or_with): ... this one.
12900 (evaluate_predicate): Turn to ...
12901 (predicate::evaluate): ... this one.
12902 (predicate_probability): Turn to ...
12903 (predicate::probability): ... this one.
12904 (dump_condition): Update.
12905 (dump_predicate): Turn to ...
12906 (predicate::dump): ... this one.
12907 (account_size_time): Update.
12908 (edge_set_predicate): Update.
12909 (set_hint_predicate): UPdate.
12910 (evaluate_conditions_for_known_args): Update.
12911 (evaluate_properties_for_edge): Update.
12912 (remap_predicate_after_duplication): Turn to...
12913 (predicate::remap_after_duplication): ... this one.
12914 (remap_hint_predicate_after_duplication): Update.
12915 (inline_summary_t::duplicate): UPdate.
12916 (dump_inline_edge_summary): Update.
12917 (dump_inline_summary): Update.
12918 (set_cond_stmt_execution_predicate): Update.
12919 (set_switch_stmt_execution_predicate): Update.
12920 (compute_bb_predicates): Update.
12921 (will_be_nonconstant_expr_predicate): Update.
12922 (will_be_nonconstant_predicate): Update.
12923 (phi_result_unknown_predicate): Update.
12924 (predicate_for_phi_result): Update.
12925 (array_index_predicate): Update.
12926 (estimate_function_body_sizes): Update.
12927 (estimate_node_size_and_time): Update.
12928 (estimate_ipcp_clone_size_and_time): Update.
12929 (remap_predicate): Rename to ...
12930 (predicate::remap_after_inlining): ... this one.
12931 (remap_hint_predicate): Update.
12932 (inline_merge_summary): Update.
12933 (inline_update_overall_summary): Update.
12934 (estimate_size_after_inlining): Update.
12935 (read_predicate): Rename to ...
12936 (predicate::stream_in): ... this one.
12937 (read_inline_edge_summary): Update.
12938 (write_predicate): Rename to ...
12939 (predicate::stream_out): ... this one.
12940 (write_inline_edge_summary): Update.
12941 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
12942 (clause_t): Turn to uint32_t
12943 (predicate): Turn to class; implement constructor and operators
12945 (size_time_entry): Update.
12946 (inline_summary): Update.
12947 (inline_edge_summary): Update.
12949 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
12951 * fold-const.c (fold_binary_loc): Move transformation...
12952 * match.pd (C - X CMP X): ... here.
12954 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
12956 * config/sparc/sparc.c (sparc_option_override): Set function
12957 alignment for -mcpu=niagara7 to 64 to match the I$ line.
12958 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
12960 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
12962 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
12964 2017-05-18 Marek Polacek <polacek@redhat.com>
12967 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
12968 (pass_ubsan::execute): Call gimple_assign_single_p instead of
12969 gimple_assign_load_p.
12971 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
12973 PR middle-end/80692
12974 * real.c (do_compare): Give decimal_do_compare preference over
12975 comparing just the signs.
12977 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
12979 * doc/md.texi (Canonicalization of Instructions): Describe the
12980 canonical form of instructions that inherently set a condition
12983 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
12985 PR middle-end/80775
12986 * tree-cfg.c: Move deletion of unreachable case statements to after
12987 the merging of consecutive case labels.
12989 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
12991 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
12992 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
12993 restoring of callee-saved registers.
12995 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
12997 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
12998 * config/visium/visium.c (single_set_and_flags): Likewise.
12999 * config/visium/visium.md (Substitutions): Likewise.
13001 2017-05-17 Martin Liska <mliska@suse.cz>
13003 * cfg.c: Introduce dump_flags_t type and
13004 use it instead of int type.
13006 * cfghooks.c: Likewise.
13007 * cfghooks.h (struct cfg_hooks): Likewise.
13008 * cfgrtl.c: Likewise.
13009 * cfgrtl.h: Likewise.
13010 * cgraph.c (cgraph_node::get_body): Likewise.
13011 * coretypes.h: Likewise.
13012 * domwalk.c: Likewise.
13013 * domwalk.h: Likewise.
13014 * dumpfile.c (struct dump_option_value_info): Likewise.
13015 (dump_enable_all): Likewise.
13016 (dump_switch_p_1): Likewise.
13017 (opt_info_switch_p): Likewise.
13018 * dumpfile.h (enum tree_dump_index): Likewise.
13019 (struct dump_file_info): Likewise.
13020 * genemit.c: Likewise.
13021 * generic-match-head.c: Likewise.
13022 * gengtype.c (open_base_files): Likewise.
13023 * gimple-pretty-print.c: Likewise.
13024 * gimple-pretty-print.h: Likewise.
13025 * graph.c (print_graph_cfg): Likewise.
13026 * graphite-scop-detection.c (dot_all_sese): Likewise.
13027 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
13028 * loop-unroll.c (report_unroll): Likewise.
13029 * passes.c (pass_manager::register_one_dump_file): Likewise.
13030 * print-tree.c: Likewise.
13031 * statistics.c: Likewise.
13032 * tree-cfg.c: Likewise.
13033 * tree-cfg.h: Likewise.
13034 * tree-dfa.c: Likewise.
13035 * tree-dfa.h: Likewise.
13036 * tree-dump.c (dump_function): Likewise.
13037 * tree-dump.h (struct dump_info): Likewise.
13038 * tree-pretty-print.c: Likewise.
13039 * tree-pretty-print.h: Likewise.
13040 * tree-ssa-live.c: Likewise.
13041 * tree-ssa-live.h: Likewise.
13042 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
13043 * tree-vect-loop.c: Likewise.
13044 * tree-vect-slp.c: Likewise.
13046 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
13047 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13049 PR tree-optimization/80457
13050 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
13051 of all arguments to a statement as scalar_to_vec operations.
13052 (vectorizable_call): Adjust call to vect_model_simple_cost for
13054 (vectorizable_conversion): Likewise.
13055 (vectorizable_assignment): Likewise.
13056 (vectorizable_shift): Likewise.
13057 (vectorizable_operation): Likewise.
13058 (vectorizable_comparison): Likewise.
13059 (vect_is_simple_cond): Record the def types for operands.
13060 (vectorizable_condition): Likewise, call vect_model_simple_cost.
13061 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
13062 for statement argument count.
13064 2017-05-16 Carl Love <cel@us.ibm.com>
13066 * config/rs6000/rs6000-c: Add support for built-in functions
13067 vector unsigned long long vec_bperm (vector unsigned long long,
13068 vector unsigned char)
13069 vector signed long long vec_mule (vector signed int,
13071 vector unsigned long long vec_mule (vector unsigned int,
13072 vector unsigned int)
13073 vector signed long long vec_mulo (vector signed int,
13075 vector unsigned long long vec_mulo (vector unsigned int,
13076 vector unsigned int)
13077 vector signed char vec_sldw (vector signed char,
13078 vector signed char,
13080 vector unsigned char vec_sldw (vector unsigned char,
13081 vector unsigned char,
13083 vector signed short vec_sldw (vector signed short,
13084 vector signed short,
13086 vector unsigned short vec_sldw (vector unsigned short,
13087 vector unsigned short,
13089 vector signed int vec_sldw (vector signed int,
13092 vector unsigned int vec_sldw (vector unsigned int,
13093 vector unsigned int,
13095 vector signed long long vec_sldw (vector signed long long,
13096 vector signed long long,
13098 vector unsigned long long vec_sldw (vector unsigned long long,
13099 vector unsigned long long,
13101 * config/rs6000/rs6000-c: Add support for built-in functions
13102 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
13103 * config/rs6000/altivec.h: Add defintion for vec_sldw.
13104 * doc/extend.texi: Update the built-in documentation for the
13105 new built-in functions.
13107 2017-05-16 Marek Polacek <polacek@redhat.com>
13111 * tree.c (save_expr): Don't fold the expression.
13113 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
13115 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
13116 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
13117 and (?*y,m). Update insn attributes.
13119 2017-05-16 Martin Liska <mliska@suse.cz>
13121 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
13122 flags argument of print_gimple_stmt, print_gimple_expr,
13123 print_generic_stmt and print_generic_expr.
13124 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
13125 * coretypes.h: Likewise.
13126 * except.c (dump_eh_tree): Likewise.
13127 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
13128 * gimple-pretty-print.h: Likewise.
13129 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
13130 (backprop::push_to_worklist): Likewise.
13131 (backprop::pop_from_worklist): Likewise.
13132 (backprop::process_use): Likewise.
13133 (backprop::intersect_uses): Likewise.
13134 (note_replacement): Likewise.
13135 * gimple-ssa-store-merging.c
13136 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
13137 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
13138 (pass_store_merging::execute): Likewise.
13139 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
13140 (ssa_base_cand_dump_callback): Likewise.
13141 (dump_incr_vec): Likewise.
13142 (replace_refs): Likewise.
13143 (replace_mult_candidate): Likewise.
13144 (create_add_on_incoming_edge): Likewise.
13145 (create_phi_basis): Likewise.
13146 (insert_initializers): Likewise.
13147 (all_phi_incrs_profitable): Likewise.
13148 (introduce_cast_before_cand): Likewise.
13149 (replace_one_candidate): Likewise.
13150 * gimplify.c (gimplify_expr): Likewise.
13151 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
13152 (set_rename): Likewise.
13153 (rename_uses): Likewise.
13154 (copy_loop_phi_nodes): Likewise.
13155 (add_close_phis_to_merge_points): Likewise.
13156 (copy_loop_close_phi_args): Likewise.
13157 (copy_cond_phi_args): Likewise.
13158 (graphite_copy_stmts_from_block): Likewise.
13159 (translate_pending_phi_nodes): Likewise.
13160 * graphite-poly.c (print_pdr): Likewise.
13161 (dump_gbb_cases): Likewise.
13162 (dump_gbb_conditions): Likewise.
13163 (print_scop_params): Likewise.
13164 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
13165 (build_cross_bb_scalars_use): Likewise.
13166 (gather_bbs::before_dom_children): Likewise.
13167 * hsa-dump.c (dump_hsa_immed): Likewise.
13168 * ipa-cp.c (print_ipcp_constant_value): Likewise.
13169 (get_replacement_map): Likewise.
13170 * ipa-inline-analysis.c (dump_condition): Likewise.
13171 (estimate_function_body_sizes): Likewise.
13172 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
13173 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
13174 * ipa-prop.c (ipa_dump_param): Likewise.
13175 (ipa_print_node_jump_functions_for_edge): Likewise.
13176 (ipa_modify_call_arguments): Likewise.
13177 (ipa_modify_expr): Likewise.
13178 (ipa_dump_param_adjustments): Likewise.
13179 (ipa_dump_agg_replacement_values): Likewise.
13180 (ipcp_modif_dom_walker::before_dom_children): Likewise.
13181 * ipa-pure-const.c (check_stmt): Likewise.
13182 (pass_nothrow::execute): Likewise.
13183 * ipa-split.c (execute_split_functions): Likewise.
13184 * omp-offload.c (dump_oacc_loop_part): Likewise.
13185 (dump_oacc_loop): Likewise.
13186 * trans-mem.c (tm_log_emit): Likewise.
13187 (tm_memopt_accumulate_memops): Likewise.
13188 (dump_tm_memopt_set): Likewise.
13189 (dump_tm_memopt_transform): Likewise.
13190 * tree-cfg.c (gimple_verify_flow_info): Likewise.
13191 (print_loop): Likewise.
13192 * tree-chkp-opt.c (chkp_print_addr): Likewise.
13193 (chkp_gather_checks_info): Likewise.
13194 (chkp_get_check_result): Likewise.
13195 (chkp_remove_check_if_pass): Likewise.
13196 (chkp_use_outer_bounds_if_possible): Likewise.
13197 (chkp_reduce_bounds_lifetime): Likewise.
13198 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
13199 (chkp_mark_completed_bounds): Likewise.
13200 (chkp_register_incomplete_bounds): Likewise.
13201 (chkp_mark_invalid_bounds): Likewise.
13202 (chkp_maybe_copy_and_register_bounds): Likewise.
13203 (chkp_build_returned_bound): Likewise.
13204 (chkp_get_bound_for_parm): Likewise.
13205 (chkp_build_bndldx): Likewise.
13206 (chkp_get_bounds_by_definition): Likewise.
13207 (chkp_generate_extern_var_bounds): Likewise.
13208 (chkp_get_bounds_for_decl_addr): Likewise.
13209 * tree-chrec.c (chrec_apply): Likewise.
13210 * tree-data-ref.c (dump_data_reference): Likewise.
13211 (dump_subscript): Likewise.
13212 (dump_data_dependence_relation): Likewise.
13213 (analyze_overlapping_iterations): Likewise.
13214 * tree-inline.c (expand_call_inline): Likewise.
13215 (tree_function_versioning): Likewise.
13216 * tree-into-ssa.c (dump_defs_stack): Likewise.
13217 (dump_currdefs): Likewise.
13218 (dump_names_replaced_by): Likewise.
13219 (dump_update_ssa): Likewise.
13220 (update_ssa): Likewise.
13221 * tree-object-size.c (pass_object_sizes::execute): Likewise.
13222 * tree-parloops.c (build_new_reduction): Likewise.
13223 (try_create_reduction_list): Likewise.
13224 (ref_conflicts_with_region): Likewise.
13225 (oacc_entry_exit_ok_1): Likewise.
13226 (oacc_entry_exit_single_gang): Likewise.
13227 * tree-pretty-print.h: Likewise.
13228 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
13229 (get_scalar_evolution): Likewise.
13230 (add_to_evolution): Likewise.
13231 (get_loop_exit_condition): Likewise.
13232 (analyze_evolution_in_loop): Likewise.
13233 (analyze_initial_condition): Likewise.
13234 (analyze_scalar_evolution): Likewise.
13235 (instantiate_scev): Likewise.
13236 (number_of_latch_executions): Likewise.
13237 (gather_chrec_stats): Likewise.
13238 (final_value_replacement_loop): Likewise.
13239 (scev_const_prop): Likewise.
13240 * tree-sra.c (dump_access): Likewise.
13241 (disqualify_candidate): Likewise.
13242 (create_access): Likewise.
13243 (reject): Likewise.
13244 (maybe_add_sra_candidate): Likewise.
13245 (create_access_replacement): Likewise.
13246 (analyze_access_subtree): Likewise.
13247 (analyze_all_variable_accesses): Likewise.
13248 (sra_modify_assign): Likewise.
13249 (initialize_constant_pool_replacements): Likewise.
13250 (find_param_candidates): Likewise.
13251 (decide_one_param_reduction): Likewise.
13252 (replace_removed_params_ssa_names): Likewise.
13253 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
13254 * tree-ssa-copy.c (dump_copy_of): Likewise.
13255 (copy_prop_visit_cond_stmt): Likewise.
13256 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
13257 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
13258 (record_equivalences_from_stmt): Likewise.
13259 * tree-ssa-dse.c (compute_trims): Likewise.
13260 (delete_dead_call): Likewise.
13261 (delete_dead_assignment): Likewise.
13262 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
13263 (forward_propagate_into_cond): Likewise.
13264 (pass_forwprop::execute): Likewise.
13265 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
13266 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
13268 (move_computations_worker): Likewise.
13269 (execute_sm): Likewise.
13270 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
13271 (remove_exits_and_undefined_stmts): Likewise.
13272 (remove_redundant_iv_tests): Likewise.
13273 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
13274 (adjust_iv_update_pos): Likewise.
13275 * tree-ssa-math-opts.c (bswap_replace): Likewise.
13276 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
13277 (value_replacement): Likewise.
13278 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
13279 * tree-ssa-pre.c (print_pre_expr): Likewise.
13280 (get_representative_for): Likewise.
13281 (create_expression_by_pieces): Likewise.
13282 (insert_into_preds_of_block): Likewise.
13283 (eliminate_insert): Likewise.
13284 (eliminate_dom_walker::before_dom_children): Likewise.
13285 (eliminate): Likewise.
13286 (remove_dead_inserted_code): Likewise.
13287 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
13288 * tree-ssa-reassoc.c (get_rank): Likewise.
13289 (eliminate_duplicate_pair): Likewise.
13290 (eliminate_plus_minus_pair): Likewise.
13291 (eliminate_not_pairs): Likewise.
13292 (undistribute_ops_list): Likewise.
13293 (eliminate_redundant_comparison): Likewise.
13294 (update_range_test): Likewise.
13295 (optimize_range_tests_var_bound): Likewise.
13296 (optimize_vec_cond_expr): Likewise.
13297 (rewrite_expr_tree): Likewise.
13298 (rewrite_expr_tree_parallel): Likewise.
13299 (linearize_expr): Likewise.
13300 (break_up_subtract): Likewise.
13301 (linearize_expr_tree): Likewise.
13302 (attempt_builtin_powi): Likewise.
13303 (attempt_builtin_copysign): Likewise.
13304 (transform_stmt_to_copy): Likewise.
13305 (transform_stmt_to_multiply): Likewise.
13306 (dump_ops_vector): Likewise.
13307 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
13308 (print_scc): Likewise.
13309 (set_ssa_val_to): Likewise.
13310 (visit_reference_op_store): Likewise.
13311 (visit_use): Likewise.
13312 (sccvn_dom_walker::before_dom_children): Likewise.
13313 (run_scc_vn): Likewise.
13314 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
13316 (expr_hash_elt::print): Likewise.
13317 (const_and_copies::pop_to_marker): Likewise.
13318 (const_and_copies::record_const_or_copy_raw): Likewise.
13319 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
13320 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
13321 (dump_predicates): Likewise.
13322 (find_uninit_use): Likewise.
13323 (warn_uninitialized_phi): Likewise.
13324 (pass_late_warn_uninitialized::execute): Likewise.
13325 * tree-ssa.c (verify_vssa): Likewise.
13326 (verify_ssa): Likewise.
13327 (maybe_optimize_var): Likewise.
13328 * tree-vrp.c (dump_value_range): Likewise.
13329 (dump_all_value_ranges): Likewise.
13330 (dump_asserts_for): Likewise.
13331 (register_edge_assert_for_2): Likewise.
13332 (vrp_visit_cond_stmt): Likewise.
13333 (vrp_visit_switch_stmt): Likewise.
13334 (vrp_visit_stmt): Likewise.
13335 (vrp_visit_phi_node): Likewise.
13336 (simplify_cond_using_ranges_1): Likewise.
13337 (fold_predicate_in): Likewise.
13338 (evrp_dom_walker::before_dom_children): Likewise.
13339 (evrp_dom_walker::push_value_range): Likewise.
13340 (evrp_dom_walker::pop_value_range): Likewise.
13341 (execute_early_vrp): Likewise.
13343 2017-05-16 Richard Biener <rguenther@suse.de>
13345 * dwarf2out.c (loc_list_from_tree_1): Do not create
13346 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
13348 2017-05-16 Richard Biener <rguenther@suse.de>
13350 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
13352 (note_variable_value_in_expr): If we resolved the decl ref
13353 do not push to the stack.
13355 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
13357 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
13358 operations in fast-math mode.
13359 (vaddq_f16): Likewise.
13360 (vmul_f16): Likewise.
13361 (vmulq_f16): Likewise.
13362 (vsub_f16): Likewise.
13363 (vsubq_f16): Likewise.
13364 * config/arm/neon.md (add<mode>3): New.
13366 (fma:<VH:mode>3): New. Also remove outdated comment.
13369 2017-05-16 Martin Liska <mliska@suse.cz>
13373 * ipa-devirt.c (warn_types_mismatch): Fix typo.
13374 (odr_types_equivalent_p): Likewise.
13376 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
13378 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
13380 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
13383 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
13384 non-interunit SSE move alternatives with '?'.
13385 (zero-extendsidi peephole2): New peephole to skip intermediate
13386 general register in SSE zero-extend sequence.
13388 2017-05-15 Jeff Law <law@redhat.com>
13390 * reorg.c (relax_delay_slots): Create a new variable to hold
13391 the temporary target rather than clobbering TARGET_LABEL.
13393 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
13394 missing argument to extract_bit_field call.
13395 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
13397 2017-05-15 Martin Liska <mliska@suse.cz>
13400 * gcc.c (process_command): Do not allow empty argument of -o option.
13402 2017-05-15 Renlin Li <renlin.li@arm.com>
13404 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
13405 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
13406 * config/aarch64/constraints.md (Usf): Add long call check.
13407 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
13408 (call_value): Likewise.
13409 (sibcall): Likewise.
13410 (sibcall_value): Likewise.
13412 (call_value_insn): New.
13413 (sibcall_insn): Update rtx pattern.
13414 (sibcall_value_insn): Likewise.
13415 (call_internal): Remove.
13416 (call_value_internal): Likewise.
13417 (sibcall_internal): Likewise.
13418 (sibcall_value_internal): Likewise.
13419 (call_reg): Likewise.
13420 (call_symbol): Likewise.
13421 (call_value_reg): Likewise.
13422 (call_value_symbol): Likewise.
13424 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
13427 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
13429 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
13431 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
13432 compatible with CCGOCmode and with CCZmode.
13434 2017-05-14 Martin Sebor <msebor@redhat.com>
13436 PR middle-end/77671
13437 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
13438 (gimple_fold_builtin_snprintf): Same.
13439 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
13440 (gimple_fold_builtin_snprintf): Same.
13441 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
13442 of character types.
13443 (is_call_safe): New function.
13444 (try_substitute_return_value): Call it.
13445 (try_simplify_call): New function.
13446 (pass_sprintf_length::handle_gimple_call): Call it.
13448 2017-05-14 Martin Sebor <msebor@redhat.com>
13450 PR middle-end/80669
13451 * builtins.c (expand_builtin_stpncpy): Simplify.
13453 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
13455 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
13456 * config/i386/i386.h
13457 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
13458 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
13459 (struct machine_function): Add new members call_ms2sysv,
13460 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
13461 (struct machine_frame_state): New fields sp_realigned and
13462 sp_realigned_offset.
13463 * config/i386/i386.c
13464 (enum xlogue_stub): New enum.
13465 (enum xlogue_stub_sets): New enum.
13466 (class xlogue_layout): New class.
13467 (struct ix86_frame): New fields stack_realign_allocate_offset,
13468 stack_realign_offset and outlined_save_offset. Modify comments to
13469 detail stack layout when using out-of-line stubs.
13470 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
13471 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
13472 -mcall-ms2sysv-xlogues.
13473 (stub_managed_regs): New static variable.
13474 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
13475 registers managed by out-of-line stub.
13476 (disable_call_ms2sysv_xlogues): New function.
13477 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
13478 m->call_ms2sysv when appropriate and compute frame layout for
13480 (sp_valid_at, fp_valid_at): New inline functions.
13481 (choose_basereg): New function.
13482 (choose_baseaddr): Add align parameter, use choose_basereg and modify
13484 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
13485 Use align parameter of choose_baseaddr to generated aligned SSE movs
13487 (pro_epilogue_adjust_stack): Modify to track
13488 machine_frame_state::sp_realigned.
13489 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
13490 (ix86_nsaved_sseregs): Likewise.
13491 (ix86_emit_save_regs): Likewise.
13492 (ix86_emit_save_regs_using_mov): Likewise.
13493 (ix86_emit_save_sse_regs_using_mov): Likewise.
13494 (get_scratch_register_on_entry): Likewise.
13495 (gen_frame_set): New function.
13496 (gen_frame_load): Likewise.
13497 (gen_frame_store): Likewise.
13498 (emit_outlined_ms2sysv_save): Likewise.
13499 (emit_outlined_ms2sysv_restore): Likewise.
13500 (ix86_expand_prologue): Modify stack re-alignment code and call
13501 emit_outlined_ms2sysv_save when appropriate.
13502 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
13503 parameter rtx_insn *insn, which allows the function to be used to only
13504 generate the notes.
13505 (ix86_expand_epilogue): Modify validity checks of frame and stack
13506 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
13507 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
13508 * config/i386/predicates.md
13509 (save_multiple): New predicate.
13510 (restore_multiple): Likewise.
13511 * config/i386/sse.md
13512 (save_multiple<mode>): New pattern.
13513 (save_multiple_realign<mode>): Likewise.
13514 (restore_multiple<mode>): Likewise.
13515 (restore_multiple_and_return<mode>): Likewise.
13516 (restore_multiple_leave_return<mode>): Likewise.
13517 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
13519 2017-05-14 Julia Koval <julia.koval@intel.com>
13521 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
13522 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
13523 (__builtin_ia32_xsetbv): New builtins.
13524 * config/i386/i386.c (ix86_expand_special_args_builtin):
13526 (ix86_expand_builtin): Special expand for new intrinsics.
13527 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
13528 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
13529 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
13531 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13533 * cfganal.c (inverted_post_order_compute): Change argument type
13535 * cfganal.h (inverted_post_order_compute): Adjust prototype.
13536 * df-core.c (rest_of_handle_df_initialize): Adjust.
13537 (rest_of_handle_df_finish): Likewise.
13538 (df_analyze_1): Likewise.
13539 (df_analyze): Likewise.
13540 (loop_inverted_post_order_compute): Change argument to be a vec *.
13541 (df_analyze_loop): Adjust.
13542 (df_get_n_blocks): Likewise.
13543 (df_get_postorder): Likewise.
13544 * df.h (struct df_d): Change field to be a vec.
13545 * lcm.c (compute_laterin): Adjust.
13546 (compute_available): Likewise.
13547 * lra-lives.c (lra_create_live_ranges_1): Likewise.
13548 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
13549 * tree-ssa-pre.c (compute_antic): Likewise.
13551 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13553 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
13554 (depth_first_search::depth_first_search): Change structure init
13555 function to this constructor.
13556 (depth_first_search::add_bb): Rename function to this member.
13557 (depth_first_search::execute): Likewise.
13558 (flow_dfs_compute_reverse_finish): Adjust.
13560 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13562 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
13563 (longest_simple_path): Likewise.
13564 * shrink-wrap.c (spread_components): Likewise.
13565 (disqualify_problematic_components): Likewise.
13566 (emit_common_heads_for_components): Likewise.
13567 (emit_common_tails_for_components): Likewise.
13568 (insert_prologue_epilogue_for_components): Likewise.
13570 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13572 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
13575 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13577 * df-core.c (df_set_blocks): Start using auto_bitmap.
13578 (df_compact_blocks): Likewise.
13579 * df-problems.c (df_rd_confluence_n): Likewise.
13580 * df-scan.c (df_insn_rescan_all): Likewise.
13581 (df_process_deferred_rescans): Likewise.
13582 (df_update_entry_block_defs): Likewise.
13583 (df_update_exit_block_uses): Likewise.
13584 (df_entry_block_bitmap_verify): Likewise.
13585 (df_exit_block_bitmap_verify): Likewise.
13586 (df_scan_verify): Likewise.
13587 * lra-constraints.c (lra_constraints): Likewise.
13588 (undo_optional_reloads): Likewise.
13589 (lra_undo_inheritance): Likewise.
13590 * lra-remat.c (calculate_gen_cands): Likewise.
13591 (do_remat): Likewise.
13592 * lra-spills.c (assign_spill_hard_regs): Likewise.
13593 (spill_pseudos): Likewise.
13594 * tree-ssa-pre.c (bitmap_set_and): Likewise.
13595 (bitmap_set_subtract_values): Likewise.
13597 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13599 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
13600 management with auto_bitmap.
13601 (fix_inter_tick): Likewise.
13602 (fix_recovery_deps): Likewise.
13603 * ira.c (add_store_equivs): Likewise.
13604 (find_moveable_pseudos): Likewise.
13605 (split_live_ranges_for_shrink_wrap): Likewise.
13606 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
13607 (rtx_reuse_manager::seen_def_p): Likewise.
13608 (rtx_reuse_manager::set_seen_def): Likewise.
13609 * print-rtl.h (class rtx_reuse_manager): Likewise.
13611 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13613 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
13615 (migrate_btr_def): Likewise.
13616 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
13617 * df-core.c (loop_post_order_compute): Likewise.
13618 (loop_inverted_post_order_compute): Likewise.
13619 * hsa-common.h: Likewise.
13620 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
13621 * init-regs.c (initialize_uninitialized_regs): Likewise.
13622 * ipa-inline.c (resolve_noninline_speculation): Likewise.
13623 (inline_small_functions): Likewise.
13624 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
13625 * ira.c (combine_and_move_insns): Likewise.
13626 (build_insn_chain): Likewise.
13627 * loop-invariant.c (find_invariants): Likewise.
13628 * lower-subreg.c (propagate_pseudo_copies): Likewise.
13629 * predict.c (tree_predict_by_opcode): Likewise.
13630 (predict_paths_leading_to): Likewise.
13631 (predict_paths_leading_to_edge): Likewise.
13632 (estimate_loops_at_level): Likewise.
13633 (estimate_loops): Likewise.
13634 * shrink-wrap.c (try_shrink_wrapping): Likewise.
13635 (spread_components): Likewise.
13636 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
13637 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
13638 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
13639 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
13640 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
13641 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
13642 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
13643 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
13644 (mark_threaded_blocks): Likewise.
13645 (thread_through_all_blocks): Likewise.
13646 * tree-ssa.c (verify_ssa): Likewise.
13647 (execute_update_addresses_taken): Likewise.
13648 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
13650 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13652 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
13654 (post_order_compute): Likewise.
13655 (inverted_post_order_compute): Likewise.
13656 (pre_and_rev_post_order_compute_fn): Likewise.
13658 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13660 * genrecog.c (int_set::int_set): Explicitly construct our
13661 auto_vec base class.
13662 * vec.h (auto_vec::auto_vec): New constructor.
13664 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13666 * bitmap.h (class auto_bitmap): New constructor taking
13667 bitmap_obstack * argument.
13669 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13671 * bitmap.h (class auto_bitmap): Change type of m_bits to
13672 bitmap_head, and adjust ctor / dtor and member operators.
13674 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
13676 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
13677 when returned register mode doesn't match original mode.
13679 2017-05-12 Jeff Law <law@redhat.com>
13680 Jakub Jelinek <jakub@redhat.com>
13682 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
13683 we look for cc setter after the compare-elim changes.
13684 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
13685 within the vector to match what compare-elim now expects.
13686 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
13687 (xorsi3_flags, one_cmplsi2_flags): Likewise.
13689 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
13690 after the compare-elim changes.
13691 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
13692 the vector to match what compare-elim now expects.
13693 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
13694 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
13695 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
13696 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
13697 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
13699 * config/visium/visium.c (single_set_and_flags): Fix where
13700 we look for cc setter after the compare-elim changes.
13701 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
13702 with the vector to match what compare-elim now expects.
13703 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
13704 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
13705 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
13706 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
13707 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
13708 (neg<mode>2_insn_set_overflow): Likewise.
13710 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
13712 PR middle-end/79794
13713 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
13714 maybe_expand_insn call, set ops[0].target. If still set after call,
13715 set alt_rtl. Add extra arg to recursive calls.
13716 (extract_bit_field): Add alt_rtl argument. Pass to
13718 * expmed.h (extract_bit_field): Fix prototype.
13719 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
13720 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
13721 to extract_bit_field_calls.
13722 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
13723 Pass alt_rtl to extract_bit_field calls.
13724 * calls.c (store_unaligned_arguments_into_psuedos)
13725 load_register_parameters): Pass extra NULL to extract_bit_field calls.
13726 * optabs.c (maybe_legitimize_operand): Clear op->target when call
13728 * optabs.h (struct expand_operand): Add target bitfield.
13730 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
13732 * compare-elim.c (try_eliminate_compare): Canonicalize
13733 operation with embedded compare to
13734 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
13735 (set (reg) (operation)].
13737 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
13739 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
13742 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
13743 cost of adding a carry flag for ADC instruction.
13744 [case MINUS]: Ignore the cost of subtracting a carry flag
13745 for SBB instruction.
13747 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
13749 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
13751 * config/rs6000/bmiintrin.h: New file.
13752 * config/rs6000/bmi2intrin.h: New file.
13753 * config/rs6000/x86intrin.h: New file.
13755 2017-05-12 Jeff Law <law@redhat.com>
13757 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
13760 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
13762 PR middle-end/80707
13763 * tree-cfg.c: Remove cfg edges of unreachable case statements.
13765 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13767 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
13768 early expansion of vector divide builtins.
13769 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
13770 builtins identified as having unsigned arguments.
13772 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13774 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
13775 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
13776 expansion of vector logical operations (and, andc, or, xor,
13779 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13781 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
13782 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
13784 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13786 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
13787 early GIMPLE expansion of vector multiplies.
13789 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13791 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
13792 TARGET_HAVE_MOVT conditional.
13793 (movt splitter): Likewise.
13795 2017-05-12 Richard Biener <rguenther@suse.de>
13797 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
13798 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13799 Fold all stmts not inplace.
13801 2017-05-12 Richard Biener <rguenther@suse.de>
13803 PR tree-optimization/80713
13804 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
13805 inserted_exprs bit for not removed stmts.
13807 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
13809 PR middle-end/69921
13810 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
13811 parallelized" attribute for parallelized OpenACC kernels.
13812 * omp-offload.c (execute_oacc_device_lower): Use it.
13814 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
13815 Set "oacc kernels" attribute.
13816 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
13817 parameter. Adjust all users.
13818 (oacc_fn_attrib_kernels_p): Remove function.
13819 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
13820 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
13821 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
13822 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
13823 assert "oacc kernels" attribute is set.
13825 2017-05-11 Carl Love <cel@us.ibm.com>
13827 * config/rs6000/rs6000-c: Add support for built-in functions
13828 vector unsigned char vec_popcnt (vector signed char)
13829 vector unsigned char vec_popcnt (vector unsigned char)
13830 vector unsigned short vec_popcnt (vector signed short)
13831 vector unsigned short vec_popcnt (vector unsigned short)
13832 vector unsigned int vec_popcnt (vector signed int)
13833 vector unsigned int vec_popcnt (vector unsigned int)
13834 vector unsigned long long vec_popcnt (vector signed long long)
13835 vector unsigned long long vec_popcnt (vector unsigned long long)
13836 vector signed long long vec_slo (vector signed long long,
13837 vector signed char)
13838 vector signed long long vec_slo (vector signed long long,
13839 vector unsigned char)
13840 vector unsigned long long vec_slo (vector unsigned long long,
13841 vector signed char)
13842 vector unsigned long long vec_slo (vector unsigned long long,
13843 vector unsigned char)
13844 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
13845 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
13846 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
13847 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
13848 * doc/extend.texi: Update the built-in documentation file for the
13849 new built-in functions.
13851 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
13853 * attribs.h (sorted_attr_string): Move machine independent
13854 functions for target clone support from the i386 port to common
13855 code. Rename ix86_function_versions to common_function_versions.
13856 Rename make_name to make_unique_name.
13857 (common_function_versions): Likewise.
13858 (make_unique_name): Likewise.
13859 (make_dispatcher_decl): Likewise.
13860 (is_function_default_version): Likewise.
13861 * attribs.c (attr_strcmp): Likewise.
13862 (sorted_attr_string): Likewise.
13863 (common_function_versions): Likewise.
13864 (make_unique_name): Likewise.
13865 (make_dispatcher_decl): Likewise.
13866 (is_function_default_version): Likewise.
13867 * config/i386/i386.c (attr_strcmp): Likewise.
13868 (sorted_attr_string): Likewise.
13869 (ix86_function_versions): Likewise.
13870 (make_name): Likewise.
13871 (make_dispatcher_decl): Likewise.
13872 (is_function_default_version): Likewise.
13873 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
13875 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13878 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
13879 Account for direct move costs for vec_construct of integer
13882 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
13885 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
13886 (UNSPEC_STX_ATOMIC): Ditto.
13887 (loaddi_via_sse): New insn.
13888 (storedi_via_sse): Ditto.
13889 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
13890 Update corresponding peephole2 patterns.
13891 (atomic_storedi_fpu): Ditto.
13893 2017-05-11 Julia Koval <julia.koval@intel.com>
13895 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
13896 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
13898 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
13899 (__builtin_ia32_rsqrt14ss_mask): New builtins.
13900 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
13902 2017-05-11 Nathan Sidwell <nathan@acm.org>
13904 * graphite-poly.c: Include dumpfile.h.
13906 * dumpfle.h (dump_function): Declare here ...
13907 * tree-dump.h (dump_function): ... not here.
13908 * dumpfile.c: #include tree-cfg.h.
13909 (dump_function): Move here from ...
13910 * tree-dump.c (dump_function): ... here.
13911 * gimplify.c: #include splay-tree.h, not tree-dump.h.
13912 * graphite-poly.c: Don't include tree-dump.h.
13913 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
13914 * print-tree.c: Likewise.
13915 * stor-layout.c: Likewise.
13916 * tree-nested.c: Likewise.
13918 * dumpfile.c (dump_start): Use TDF_FLAGS.
13919 (dump_enable_all): Fix TDF_KIND check thinko.
13921 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
13923 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13924 array entries to represent two legal parameterizations of the
13925 overloaded __builtin_cmpb function, as represented by the
13926 P6_OV_BUILTIN_CMPB constant.
13927 (altivec_resolve_overloaded_builtin): Add special case handling
13928 for the __builtin_cmpb function, as represented by the
13929 P6_OV_BUILTIN_CMPB constant.
13930 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
13931 (BU_P6_64BIT_2): New macro.
13932 (BU_P6_OVERLOAD_2): New macro
13933 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
13934 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
13935 (CMPB): Add overload support to represent both 32-bit and 64-bit
13936 compare-bytes function.
13937 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
13938 support for TARGET_CMPB.
13939 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
13940 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
13941 documentation of the __builtin_cmpb overloaded built-in function.
13943 2017-05-11 Richard Biener <rguenther@suse.de>
13945 PR tree-optimization/80705
13946 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
13947 bases are not vectorizable.
13949 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13951 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
13952 when counting register pressure.
13954 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13956 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
13957 (iv_ca_more_deps): Renamed to ...
13958 (iv_ca_compare_deps): ... this.
13959 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
13961 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13963 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
13965 (determine_group_iv_costs): ... here.
13966 (find_inv_vars_cb): Record inv var if it's not recorded before.
13968 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13970 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
13971 (get_shiftadd_cost): Ditto.
13973 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13975 * tree-ssa-address.c: Include header file.
13976 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
13978 (add_to_parts): Refactor.
13979 (addr_to_parts): New parameter. Update use of move_hint_to_base.
13980 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
13983 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13985 PR tree-optimization/53090
13986 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
13988 (extract_cond_operands): Detect condition with IV on both sides
13989 and return COMP_IV_EXPR_2.
13990 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
13991 (rewrite_use_compare): Simplify by removing call to function
13992 extract_cond_operands.
13994 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13996 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
13997 (extract_cond_operands): Detect condition comparing against non-
13998 invariant bound and return appropriate enum value.
13999 (find_interesting_uses_cond): Update use of extract_cond_operands.
14000 Handle its return value accordingly.
14001 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
14003 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14005 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
14006 nonlinear iv_use computation in loop invariant sensitive way.
14008 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14010 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
14011 (find_iv_candidates): Call relate_compare_use_with_all_cands.
14013 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14015 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
14016 (dump_cand): Support iv_cand.inv_exprs.
14017 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
14019 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
14022 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14024 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
14026 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
14027 as local function. Include necessary header files.
14028 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
14030 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14032 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
14034 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14036 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
14037 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
14038 RSHIFT_EXPR and BIT_NOT_EXPR.
14040 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14042 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
14043 (adjust_setup_cost): New parameter supporting round up adjustment.
14044 (struct address_cost_data): Delete.
14045 (force_expr_to_var_cost): Don't bound cost with spill_cost.
14046 (split_address_cost, ptr_difference_cost): Delete.
14047 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
14048 (struct ainc_cost_data): New struct.
14049 (get_address_cost_ainc): New function.
14050 (get_address_cost, get_computation_cost): Reimplement.
14051 (determine_group_iv_cost_address): Record inv_expr for all uses of
14053 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
14054 (iv_ca_has_deps): Reimplemented to ...
14055 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
14057 (iv_ca_extend): Call iv_ca_more_deps.
14059 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14061 * tree-ssa-address.c (struct mem_address): Move to header file.
14062 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
14063 * tree-ssa-address.h (struct mem_address): Move from C file.
14064 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
14066 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14068 * tree-affine.h (aff_combination_type): New interface.
14069 (aff_combination_zero_p): Remove static.
14070 (aff_combination_const_p): New interface.
14071 (aff_combination_singleton_var_p): New interfaces.
14073 2017-05-11 Richard Biener <rguenther@suse.de>
14075 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
14076 Skip unreachable blocks and destinations.
14077 (eliminate): Move stmt removal and fixup ...
14078 (fini_eliminate): ... here. Skip inserted exprs.
14079 (pass_pre::execute): Move fini_pre after fini_eliminate.
14080 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
14081 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
14082 PRE to get rid of dead code that has invalid SSA form and
14083 split critical edges again.
14085 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14087 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
14089 2017-05-11 Richard Biener <rguenther@suse.de>
14091 * passes.c (execute_function_todo): Verify loops if they are
14092 said to be up-to-date.
14093 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
14094 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
14096 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
14099 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
14100 handle calling assemble_external ourself.
14103 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
14104 modes with zero size. Enhance comment.
14106 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14108 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
14109 built-ins for vec_xl and vec_xst with short and char pointer
14112 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
14114 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
14115 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
14116 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
14117 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
14118 (_mm_maskz_min_round_ss): New intrinsics.
14119 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
14120 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
14121 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
14122 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
14123 (__builtin_ia32_minss_mask_round): New builtins.
14124 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14125 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
14126 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
14128 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
14129 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
14131 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
14134 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
14136 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
14137 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
14138 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
14139 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
14140 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
14141 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
14142 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
14143 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14144 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
14145 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
14146 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
14147 (__builtin_ia32_mulss_mask_round): New builtins.
14148 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14149 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
14150 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
14152 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
14153 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
14155 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
14158 2017-05-10 Julia Koval <julia.koval@intel.com>
14160 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
14161 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
14162 (_mm256_setr_m128i): New intrinsics.
14164 2017-05-10 Julia Koval <julia.koval@intel.com>
14166 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
14167 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
14168 (_mm_maskz_rcp14_ss): New intrinsics.
14169 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
14170 (__builtin_ia32_rcp14ss_mask): New builtins.
14171 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
14173 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
14175 PR tree-optimization/51513
14176 * tree-cfg.c (gimple_seq_unreachable_p): New function.
14177 (assert_unreachable_fallthru_edge_p): Use it.
14178 (group_case_labels_stmt): Likewise.
14179 * tree-cfg.h: Prototype it.
14180 * stmt.c: Include cfghooks.h and tree-cfg.h.
14181 (emit_case_dispatch_table) <gap_label>: New local variable.
14182 Use it to fill dispatch table gaps.
14183 Test for default_label before updating probabilities.
14184 (expand_case) <default_label>: Remove unneeded initialization.
14185 Test for unreachable default case statement and remove its edge.
14186 Set default_label accordingly.
14187 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
14189 2017-05-10 Carl Love <cel@us.ibm.com>
14191 * config/rs6000/rs6000-c: Add support for built-in functions
14192 vector signed char vec_neg (vector signed char)
14193 vector signed short int vec_neg (vector short int)
14194 vector signed int vec_neg (vector signed int)
14195 vector signed long long vec_neg (vector signed long long)
14196 vector float vec_neg (vector float)
14197 vector double vec_neg (vector double)
14198 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
14200 * config/rs6000/altivec.h: Add define for vec_neg
14201 * doc/extend.texi: Update the built-in documentation for the
14202 new built-in functions.
14204 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14206 PR tree-optimization/77644
14207 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
14209 2017-05-10 Nathan Sidwell <nathan@acm.org>
14211 * dumpfile.h (TDI_lang_all): New.
14212 (TDF_KIND): New. Renumber others
14213 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
14215 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
14217 (get_dump_file_name): Adjust suffix generation.
14218 (dump_enable_all): Use TDF_KIND.
14219 * doc/invoke.texi (-fdump-lang-all): Document.
14221 * dumpfile.h: Tabify.
14223 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
14226 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
14227 Move member access before delete.
14229 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
14231 * tree-inline.c (expand_call_inline): Split block at stmt
14234 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
14237 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
14238 are now unused after splitting mov{sf,sd}_hardfloat.
14239 (f32_lr2): Likewise.
14240 (f32_lm): Likewise.
14241 (f32_lm2): Likewise.
14242 (f32_li): Likewise.
14243 (f32_li2): Likewise.
14244 (f32_lv): Likewise.
14245 (f32_sr): Likewise.
14246 (f32_sr2): Likewise.
14247 (f32_sm): Likewise.
14248 (f32_sm2): Likewise.
14249 (f32_si): Likewise.
14250 (f32_si2): Likewise.
14251 (f32_sv): Likewise.
14252 (f32_dm): Likewise.
14253 (f32_vsx): Likewise.
14254 (f32_av): Likewise.
14255 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
14256 For movsf, order stores so the VSX stores occur before the GPR
14257 store which encourages the register allocator to use a traditional
14258 FPR instead of a GPR. For movsd, order the stores so that the GPR
14259 store comes before the VSX stores to allow the power6 to work.
14260 This is due to the power6 not having a 32-bit integer store
14261 instruction from a FPR.
14262 (movsf_hardfloat): Likewise.
14263 (movsd_hardfloat): Likewise.
14265 2017-05-09 Martin Sebor <msebor@redhat.com>
14267 PR translation/80280
14268 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
14271 PR translation/80280
14272 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
14273 data member added in r247778.
14274 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
14276 2017-05-09 Nathan Sidwell <nathan@acm.org>
14278 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
14280 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
14283 2017-05-09 Marek Polacek <polacek@redhat.com>
14285 * doc/invoke.texi: Fix typo.
14287 2017-05-09 Richard Biener <rguenther@suse.de>
14289 * tree-vrp.c (vrp_val_is_max): Adjust comment.
14290 (vrp_val_is_min): Likewise.
14291 (set_value_range_to_value): Likewise.
14292 (set_value_range_to_nonnegative): Likewise.
14293 (gimple_assign_nonzero_p): Likewise.
14294 (gimple_stmt_nonzero_p): Likewise.
14295 (vrp_int_const_binop): Likewise. Remove unreachable case.
14296 (adjust_range_with_scev): Adjust comments.
14297 (compare_range_with_value): Likewise.
14298 (extract_range_from_phi_node): Likewise.
14299 (test_for_singularity): Likewise.
14301 2017-05-09 Richard Biener <rguenther@suse.de>
14303 * tree-vrp.c (get_single_symbol): Add assert that we don't
14304 get overflowed constants as invariant part.
14305 (compare_values_warnv): Add comment before the TREE_NO_WARNING
14306 checks. Use wi::cmp instead of recursing for integer constants.
14307 (compare_values): Just ignore whether we assumed undefined
14308 overflow instead of failing the compare.
14309 (extract_range_for_var_from_comparison_expr): Add comment before the
14310 TREE_NO_WARNING sets.
14311 (test_for_singularity): Likewise.
14312 (extract_range_from_comparison): Do not disable optimization
14313 when we assumed undefined overflow.
14314 (extract_range_basic): Remove init of unused var.
14316 2017-05-09 Richard Biener <rguenther@suse.de>
14318 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
14319 (extract_range_from_multiplicative_op_1): Adjust.
14320 (extract_range_from_binary_expr_1): Use int_const_binop.
14322 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
14325 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
14326 rs6000_store_data_bypass_p in seven define_bypass directives and
14327 in several comments.
14328 * config/rs6000/rs6000-protos.h: Add prototype for
14329 rs6000_store_data_bypass_p function.
14330 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
14331 function implements slightly different (rs6000-specific) semantics
14332 than store_data_bypass_p, returning false rather than aborting
14333 with assertion error when arguments do not satisfy the
14334 requirements of store data bypass.
14335 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
14336 rs6000_store_data_bypass_p.
14338 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
14340 * config/xtensa/xtensa-protos.h
14341 (xtensa_initial_elimination_offset): New declaration.
14342 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
14343 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
14344 macro definition, add case for FRAME_POINTER_REGNUM when
14345 FRAME_GROWS_DOWNWARD.
14346 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
14347 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
14348 xtensa_initial_elimination_offset.
14350 2017-05-08 Nathan Sidwell <nathan@acm.org>
14352 * doc/invoke.texi: Alphabetize -fdump options.
14354 2017-05-08 Martin Sebor <msebor@redhat.com>
14356 PR translation/80280
14357 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
14359 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
14361 * target.def (compute_frame_layout): New optional target hook.
14362 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
14363 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
14364 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
14366 * reload1.c (verify_initial_elim_offsets): Likewise.
14367 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
14368 (use_simple_return_p): Call arm_compute_frame_layout if needed.
14369 (arm_get_frame_offsets): Split up into this ...
14370 (arm_compute_frame_layout): ... and this function.
14372 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
14374 * config/aarch64/constraints.md (Usa): New constraint.
14375 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
14377 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
14379 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
14380 with_multilib_list after it has been checked.
14382 2017-05-08 Richard Biener <rguenther@suse.de>
14384 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
14385 (bitmap_set_subtract_values): Likewise.
14387 2017-05-08 Richard Biener <rguenther@suse.de>
14389 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
14390 (gimple_assign_nonzero): ... this and remove strict_overflow_p
14392 (gimple_stmt_nonzero_warnv_p): Rename to ...
14393 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
14395 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
14396 (extract_range_basic): Adjust, do not disable propagation on
14397 strict overflow sensitive simplification.
14398 (vrp_visit_cond_stmt): Likewise.
14400 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
14402 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
14403 body size unconditionally.
14405 2017-05-07 Jeff Law <law@redhat.com>
14408 2017-05-06 Jeff Law <law@redhat.com>
14409 PR tree-optimization/78496
14410 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
14413 PR tree-optimization/78496
14414 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
14415 (simplify_stmt_using_ranges): Call it.
14416 (vrp_dom_walker::before_dom_children): Extract equivalences
14417 from an ASSERT_EXPR with an equality comparison against a
14420 2017-05-06 Jeff Law <law@redhat.com>
14422 PR tree-optimization/78496
14423 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
14426 PR tree-optimization/78496
14427 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
14428 (simplify_stmt_using_ranges): Call it.
14429 (vrp_dom_walker::before_dom_children): Extract equivalences
14430 from an ASSERT_EXPR with an equality comparison against a
14433 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
14435 * lra-constraints.c (lra_copy_reg_equiv): New function.
14436 (split_reg): Use it to copy equivalence information from the
14437 original register to the spill register.
14439 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
14441 PR rtl-optimization/75964
14442 * simplify-rtx.c (simplify_const_relational_operation): Remove
14443 invalid handling of comparisons of integer ABS.
14445 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
14447 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
14448 initialize to zero.
14449 (init_regs): Remove declaration.
14450 (function_arg_advance_32): Initialize error_p as boolean variable.
14452 2017-05-05 Nathan Sidwell <nathan@acm.org>
14454 * store-motion.c (remove_reachable_equiv_notes): Reformat long
14455 lines. Use for (;;).
14457 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14459 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
14460 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
14461 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
14462 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
14463 VF=2 that require versioning.
14465 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14467 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
14470 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14472 * diagnostic.h (diagnostic_override_option_index): Convert from
14473 macro to inline function.
14475 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14477 * diagnostic.c (last_module_changed_p): New function.
14478 (set_last_module): New function.
14479 (diagnostic_report_current_module): Convert macro usage to
14480 the above functions.
14481 * diagnostic.h (diagnostic_context::last_module): Strengthen
14482 from const line_map * to const line_map_ordinary *.
14483 (diagnostic_last_module_changed): Delete macro.
14484 (diagnostic_set_last_module): Delete macro.
14486 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14488 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
14489 with diagnostic_report_diagnostic.
14490 (diagnostic_n_impl_richloc): Likewise.
14491 * diagnostic.h (report_diagnostic): Delete macro.
14492 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
14493 with diagnostic_report_diagnostic.
14494 * substring-locations.c (format_warning_va): Likewise.
14496 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14498 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
14499 save/restor of format_spec. Move option-printing code to...
14500 (print_option_information): ...this new function, and
14501 reimplement by simply printing to the pretty_printer,
14502 rather than appending to the format string.
14504 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14506 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
14507 handling logic into...
14508 (update_effective_level_from_pragmas): ...this new function.
14510 2017-05-04 Andrew Waterman <andrew@sifive.com>
14512 * config/riscv/riscv.opt (mstrict-align): New option.
14513 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
14514 (SLOW_UNALIGNED_ACCESS): Define.
14515 (riscv_slow_unaligned_access): Declare.
14516 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
14518 (riscv_slow_unaligned_access): New variable.
14519 (rocket_tune_info): Set slow_unaligned_access to true.
14520 (optimize_size_tune_info): Set slow_unaligned_access to false.
14521 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
14522 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
14523 (riscv_option_override): Set riscv_slow_unaligned_access.
14524 * doc/invoke.texi: Add -mstrict-align to RISC-V.
14526 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
14528 * config/riscv/riscv.md: Unify indentation.
14530 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14535 * config/rs6000/rs6000.md (u code attribute): Add FIX and
14537 (extendsi<mode>2): Add support for doing sign extension via
14538 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
14539 don't have ISA 3.0 instructions.
14540 (extendsi<mode>2 splitter): Likewise.
14541 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
14542 generate the normal insns since SImode can now go in vector
14543 registers. Disallow the special UNSPECs needed for previous
14544 machines to hide SImode being used. Add new insns
14545 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
14546 (fix_trunc<mode>si2_stfiwx): Likewise.
14547 (fix_trunc<mode>si2_internal): Likewise.
14548 (fixuns_trunc<mode>si2): Likewise.
14549 (fixuns_trunc<mode>si2_stfiwx): Likewise.
14550 (fctiw<u>z_<mode>_smallint): Likewise.
14551 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
14552 of floating point to 32-bit integer from doing a direct move to
14553 the GPR registers to do a store.
14554 (fctiwz_<mode>): Break long line.
14556 2017-05-05 Bin Cheng <bin.cheng@arm.com>
14558 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
14559 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
14560 (addr_list, addr_offset_valid_p): New.
14561 (split_address_groups): Check offset validity with above function.
14562 (gt-tree-ssa-loop-ivopts.h): Include header file.
14564 2017-05-05 Nathan Sidwell <nathan@acm.org>
14566 * config.gcc (arm*-*-*): Add missing 'fi'.
14568 2017-05-05 Steve Ellcey <sellcey@cavium.com>
14570 * doc/invoke.texi (-fopt-info): Explicitly say order of options
14571 included in -fopt-info does not matter.
14572 * doc/optinfo.texi (-fopt-info): Fix description of default
14573 behavour. Explicitly say order of options included in -fopt-info
14576 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
14578 * config.gcc: Allow combinations of aprofile and rmprofile values for
14579 --with-multilib-list.
14580 * config/arm/t-multilib: New file.
14581 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
14582 variables. Remove setting of ISA and floating-point ABI in
14583 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
14584 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
14585 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
14586 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
14588 * config/arm/t-rmprofile: Likewise except for the matches changes.
14589 * doc/install.texi (--with-multilib-list): Document the combination of
14590 aprofile and rmprofile values and warn about pitfalls in doing that.
14592 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
14594 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
14595 (movdi_aarch64): Likewise.
14597 2017-05-05 Jakub Jelinek <jakub@redhat.com>
14599 PR tree-optimization/80632
14600 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
14602 (build_arrays): Initialize it for virtual phis.
14603 (fix_phi_nodes): Use it for virtual phis.
14605 PR tree-optimization/80558
14606 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
14607 [x, y] op z into [x op, y op z] for op & or | if conditions
14610 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
14611 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14614 * config/arm/arm.md (use_literal_pool): Remove.
14615 (64-bit immediate split): No longer takes cost into consideration
14616 if arm_disable_literal_pool is enabled.
14617 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
14618 used when arm_disable_literal_pool is enabled.
14619 (arm_max_const_double_inline_cost): Remove use of
14620 arm_disable_literal_pool.
14621 (push_minipool_fix): Add assert.
14622 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
14623 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
14624 (no_literal_pool_sf_immediate): New.
14626 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14628 PR tree-optimization/80613
14629 * tree-ssa-dce.c (propagate_necessity): Remove cases for
14630 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
14632 2017-05-05 Richard Biener <rguenther@suse.de>
14634 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
14636 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
14638 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
14639 of this flag from insn conditions due to removal from r247495.
14641 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
14643 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
14645 (arm_early_store_addr_dep_ptr): Likewise.
14646 * config/arm/aarch-common-protos.h
14647 (arm_early_load_addr_dep_ptr): Add prototype.
14648 (arm_early_store_addr_dep_ptr): Likewise.
14649 * config/arm/cortex-a53.md: Add new bypasses.
14651 2017-05-05 Jakub Jelinek <jakub@redhat.com>
14653 * tree.c (next_type_uid): Change type to unsigned.
14654 (type_hash_canon): Decrement back next_type_uid if
14655 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
14656 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
14659 2017-05-04 Martin Sebor <msebor@redhat.com>
14661 * builtins.c: Fix a trivial typo in a comment.
14663 PR middle-end/79234
14664 * builtins.c (check_sizes): Adjust to handle reading past the end.
14665 Avoid printing excessive upper bound of ranges. Use %E to print
14666 tree nodes instead of converting them to %wu.
14667 (expand_builtin_memchr): New function.
14668 (compute_dest_size): Rename...
14669 (compute_objsize): ...to this.
14670 (expand_builtin_memcpy): Adjust.
14671 (expand_builtin_mempcpy): Adjust.
14672 (expand_builtin_strcat): Adjust.
14673 (expand_builtin_strcpy): Adjust.
14674 (check_strncat_sizes): Adjust.
14675 (expand_builtin_strncat): Adjust.
14676 (expand_builtin_strncpy): Adjust and simplify.
14677 (expand_builtin_memset): Adjust.
14678 (expand_builtin_bzero): Adjust.
14679 (expand_builtin_memcmp): Adjust.
14680 (expand_builtin): Handle memcmp.
14681 (maybe_emit_chk_warning): Check strncat just once.
14683 2017-05-04 Martin Sebor <msebor@redhat.com>
14685 PR preprocessor/79214
14686 PR middle-end/79222
14687 PR middle-end/79223
14688 * builtins.c (check_sizes): Add inlining context and issue
14689 warnings even when -Wno-system-headers is set.
14690 (check_strncat_sizes): Same.
14691 (expand_builtin_strncat): Same.
14692 (expand_builtin_memmove): New function.
14693 (expand_builtin_stpncpy): Same.
14694 (expand_builtin): Handle memmove and stpncpy.
14696 2017-05-04 Bin Cheng <bin.cheng@arm.com>
14698 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
14699 which is not used any more.
14701 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14703 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
14705 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14707 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
14708 (cortexa53_tunings): Likewise.
14709 (cortexa57_tunings): Likewise.
14710 (cortexa72_tunings): Likewise.
14711 (cortexa73_tunings): Likewise.
14713 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14715 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
14716 Set loop alignment to 8.
14718 2017-05-04 Martin Sebor <msebor@redhat.com>
14720 PR translation/80280
14721 * builtins.c (expand_builtin_object_size): Add missing quoting to
14722 %D and like directives.
14723 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
14724 (hsa_type_for_tree_type): Same.
14725 (verify_function_arguments): Same.
14726 * symtab.c (symbol_table::change_decl_assembler_name): Same.
14727 * varasm.c (get_section): Same.
14730 2017-05-04 Martin Sebor <msebor@redhat.com>
14732 PR translation/80280
14733 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
14735 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14737 * config/aarch64/aarch64.c (generic_addrcost_table):
14738 Change HI/TI mode setting.
14740 2017-05-04 Martin Jambor <mjambor@suse.cz>
14742 PR tree-optimization/80622
14743 * tree-sra.c (comes_initialized_p): New function.
14744 (build_accesses_from_assign): Only set write lazily when
14745 comes_initialized_p is false.
14746 (analyze_access_subtree): Use comes_initialized_p.
14747 (propagate_subaccesses_across_link): Assert !comes_initialized_p
14748 instead of testing for PARM_DECL.
14750 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14752 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
14753 constraint on operand 0 to allow more general addressing modes.
14754 Adjust output template.
14755 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
14757 * config/aarch64/aarch64-protos.h
14758 (aarch64_address_valid_for_prefetch_p): Declare prototype.
14759 * config/aarch64/constraints.md (Dp): New address constraint.
14760 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
14763 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
14765 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
14766 update use of estimate_ipcp_clone_size_and_time.
14767 (estimate_local_effects): Update use of
14768 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
14769 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
14770 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
14771 Return nonspecialized time.
14773 2017-05-04 Richard Biener <rguenther@suse.de>
14775 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
14776 for the last VUSE which def dominates the PHI. Directly call
14778 (get_continuation_for_phi_1): Remove.
14780 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
14782 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
14783 to explain the use of truncating division. Cap the number of
14784 iterations to the maximum given by nb_iterations_upper_bound,
14787 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
14789 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
14790 * configure: Regenerate.
14791 * config.in: Regenerate.
14792 * config/i386/driver-mingw32.c: new file.
14793 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
14794 * config.host: Link driver-mingw32.o on MinGW host.
14795 * doc/install.texi: Document new --enable-mingw-wildcard configure
14798 2017-05-04 Marek Polacek <polacek@redhat.com>
14800 PR tree-optimization/80612
14801 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
14803 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14804 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
14806 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
14807 (movt splitter): Likewise.
14808 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
14809 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
14810 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
14811 block for Thumb-1 with MOVT.
14812 (thumb2_legitimate_address_p): Move code block ...
14813 (can_avoid_literal_pool_for_label_p): ... into this new function.
14814 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
14816 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
14817 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
14818 "M-profile targets with the MOVT instruction".
14820 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14822 * config/arm/arm-builtins.c (arm_init_builtins): Rename
14823 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
14824 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
14826 2017-05-04 Martin Liska <mliska@suse.cz>
14828 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
14829 variable cond_code.
14831 2017-05-04 Richard Biener <rguenther@suse.de>
14833 * tree.c (array_at_struct_end_p): Handle arrays at struct
14834 end with flexarrays more conservatively. Refactor and treat
14835 arrays of arrays or aggregates more strict. Fix
14836 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
14837 * tree.c (array_at_struct_end_p): Adjust prototype.
14838 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
14839 * gimple-fold.c (get_range_strlen): Likewise.
14840 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
14842 2017-05-04 Richard Biener <rguenther@suse.de>
14844 PR tree-optimization/31130
14845 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
14847 (supports_overflow_infinity): Likewise.
14848 (is_negative_overflow_infinity): Likewise.
14849 (is_positive_overflow_infinity): Likewise.
14850 (is_overflow_infinity): Likewise.
14851 (stmt_overflow_infinity): Likewise.
14852 (overflow_infinity_range_p): Likewise.
14853 (usable_range_p): Remove as always returning true.
14854 (make_overflow_infinity): Remove.
14855 (negative_overflow_infinity): Likewise.
14856 (positive_overflow_infinity): Likewise.
14857 (avoid_overflow_infinity): Likewise.
14858 (set_value_range): Adjust accordingly.
14859 (set_value_range_to_nonnegative): Likewise, remove now unused
14860 overflow_infinity arg.
14861 (vrp_operand_equal_p): Adjust.
14862 (update_value_range): Likewise.
14863 (range_int_cst_singleton_p): Likewise.
14864 (operand_less_p): Likewise.
14865 (compare_values_warnv): Likewise.
14866 (extract_range_for_var_from_comparison_expr): Likewise.
14867 (vrp_int_const_binop): Likewise.
14868 (zero_nonzero_bits_from_vr): Likewise.
14869 (extract_range_from_multiplicative_op_1): Likewise.
14870 (extract_range_from_binary_expr_1): Likewise.
14871 (extract_range_from_unary_expr): Likewise.
14872 (extract_range_from_comparison): Likewise.
14873 (extract_range_basic): Likewise.
14874 (adjust_range_with_scev): Likewise.
14875 (compare_ranges): Likewise.
14876 (compare_range_with_value): Likewise.
14877 (dump_value_range): Likewise.
14878 (test_for_singularity): Likewise, remove strict_overflow_p parameter
14880 (simplify_cond_using_ranges): Adjust.
14882 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
14884 * brig-builtins.def: Added a builtin for class_f64.
14885 * builtin-types.def: Added a builtin type needed by class_f64.
14887 2017-05-03 Jason Merrill <jason@redhat.com>
14889 * timevar.def: Add TV_CONSTEXPR.
14891 2017-05-03 David Malcolm <dmalcolm@redhat.com>
14893 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
14895 2017-05-03 Martin Jambor <mjambor@suse.cz>
14897 * ipa-prop.c (ipa_update_after_lto_read): Removed.
14898 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
14899 * ipa-cp.c (ipcp_propagate_stage): Do not call
14900 ipa_update_after_lto_read.
14901 * ipa-inline.c (ipa_inline): Likewise.
14903 2017-05-03 Martin Jambor <mjambor@suse.cz>
14905 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
14906 tag. Added a default constructor and a destructor.
14907 (ipa_edge_args_sum_t): New class;
14908 (ipa_edge_args_sum): Declare.
14909 (ipa_edge_args_vector): Remove declaration.
14910 (IPA_EDGE_REF): Use ipa_edge_args_sum.
14911 (ipa_free_edge_args_substructures): Remove declaration.
14912 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
14913 (ipa_edge_args_info_available_for_edge_p): Likewise.
14914 * ipa-prop.c (ipa_edge_args_vector): Removed.
14915 (edge_removal_hook_holder): Likewise.
14916 (edge_duplication_hook_holder): Likewise.
14917 (ipa_edge_args_sum): New variable.
14918 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
14919 ipa_edge_args_vector.
14920 (ipa_free_edge_args_substructures): Likewise.
14921 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
14922 ipa_edge_args_vector.
14923 (ipa_edge_removal_hook): Turned into method
14924 ipa_edge_args_sum_t::remove.
14925 (ipa_edge_duplication_hook): Turned into method
14926 ipa_edge_args_sum_t::duplicate.
14927 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
14928 registering edge hooks.
14929 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
14930 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
14931 ipa_edge_args_sum instead of ipa_edge_args_vector.
14932 * ipa-profile.c (ipa_profile): Likewise.
14934 2017-05-03 Martin Jambor <mjambor@suse.cz>
14936 * symbol-summary.h (function_summary): New method exists.
14937 (function_summary::symtab_removal): Deallocate through release.
14938 (call_summary): New class.
14939 (gt_ggc_mx): New overload.
14940 (gt_pch_nx): Likewise.
14941 (gt_pch_nx): Likewise.
14943 2017-05-03 Jeff Law <law@redhat.com>
14945 PR tree-optimization/78496
14946 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
14947 from simplify_cond_using_ranges. Split off code to walk
14948 backwards through casts into ...
14949 (simplify_cond_using_ranges_2): New function.
14950 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
14951 (execute_vrp): After identifying jump threads, call
14952 simplify_cond_using_ranges_2.
14954 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
14957 * ipa-inline.h (inline_summary): Add ctor.
14958 (create_ggc): Do not use ggc_cleared_alloc.
14960 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
14961 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14963 * gcc.c (handle_braces): Support escaping in switch matching
14965 * doc/invoke.texi (Spec Files): Document it.
14966 Remove superfluous @code markup in items.
14968 2017-05-03 David Malcolm <dmalcolm@redhat.com>
14970 * diagnostic-show-locus.c (struct column_range): New struct.
14971 (get_affected_columns): New function.
14972 (get_printed_columns): New function.
14973 (struct correction): New struct.
14974 (correction::ensure_capacity): New function.
14975 (correction::ensure_terminated): New function.
14976 (struct line_corrections): New struct.
14977 (line_corrections::~line_corrections): New dtor.
14978 (line_corrections::add_hint): New function.
14979 (layout::print_trailing_fixits): Reimplement in terms of the new
14981 (selftest::test_overlapped_fixit_printing): New function.
14982 (selftest::diagnostic_show_locus_c_tests): Call it.
14984 2017-05-03 Nathan Sidwell <nathan@acm.org>
14986 Canonicalize canonical type hashing
14987 * tree.h (type_hash_canon_hash): Declare.
14988 * tree.c (type_hash_list, attribute_hash_list): Move into
14989 type_hash_canon_hash.
14990 (build_type_attribute_qual_variant): Break out hash code calc into
14991 type_hash_canon_hash.
14992 (type_hash_canon_hash): New. Generic type hash computation.
14993 (build_range_type_1, build_array_type_1, build_function_type,
14994 build_method_type_directly, build_offset_type, build_complex_type,
14995 make_vector_type): Call it.
14997 2017-05-03 Richard Biener <rguenther@suse.de>
14999 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
15000 When all DRs have unknown misaligned do not always peel
15001 when there is a store but apply the same costing model as if
15002 there were only loads.
15004 2017-05-03 Richard Biener <rguenther@suse.de>
15007 PR tree-optimization/80492
15008 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
15009 compare_base_decls returning dont-know properly.
15011 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
15013 * config/arm/iterators.md (CCSI): New mode iterator.
15014 (arch): New mode attribute.
15015 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
15016 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
15017 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
15018 code iterator for success result mode.
15019 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
15020 the corresponding new insn generators.
15022 2017-05-03 Bin Cheng <bin.cheng@arm.com>
15025 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15026 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
15028 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
15030 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
15031 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
15032 (DDR_A): Wrap DDR argument in brackets.
15033 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
15034 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
15035 (DDR_REVERSED_P): Likewise.
15037 2017-05-03 Jakub Jelinek <jakub@redhat.com>
15039 PR tree-optimization/79472
15040 * tree-switch-conversion.c (struct switch_conv_info): Add
15041 contiguous_range and default_case_nonstandard fields.
15042 (collect_switch_conv_info): Compute contiguous_range and
15043 default_case_nonstandard fields, don't clear final_bb if
15044 contiguous_range and only the default case doesn't have the required
15046 (check_all_empty_except_final): Set default_case_nonstandard instead
15047 of failing if contiguous_range and the default case doesn't have empty
15049 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
15050 and only the default case doesn't have the required constants. Skip
15052 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
15053 if default_case_nonstandard.
15054 (build_constructors): Build constant 1 just once. Assert that default
15055 values aren't inserted in between cases if contiguous_range. Skip
15057 (build_arrays): Skip virtual phis.
15058 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
15059 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
15060 Handle virtual phis.
15061 (gen_inbound_check): Handle default_case_nonstandard case.
15062 (process_switch): Adjust check_final_bb caller. Call
15063 gather_default_values with the first non-default case instead of
15064 default case if default_case_nonstandard.
15066 2017-05-02 Nathan Sidwell <nathan@acm.org>
15068 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
15069 check. Fix formatting.
15071 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
15073 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
15074 errors when comparing specialized and unspecialized times.
15076 2017-05-02 David Malcolm <dmalcolm@redhat.com>
15078 * diagnostic-show-locus.c
15079 (layout::should_print_annotation_line_p): Make private.
15080 (layout::print_annotation_line): Make private.
15081 (layout::annotation_line_showed_range_p): Make private.
15082 (layout::show_ruler): Make private.
15083 (layout::print_source_line): Make private. Pass in line and
15084 line_width, rather than calling location_get_source_line. Drop
15086 (layout::print_leading_fixits): New method.
15087 (layout::print_any_fixits): Rename to...
15088 (layout::print_trailing_fixits): ...this, and make private.
15089 Don't print newline fixits.
15090 (diagnostic_show_locus): Move logic for printing one row into...
15091 (layout::print_line): ...this new function. Move the
15092 location_get_source_line call and error-handling from
15093 print_source_line to here. Call print_leading_fixits, and rename
15094 print_any_fixits to print_trailing_fixits.
15095 (selftest::test_fixit_insert_containing_newline): Update now that
15096 newlines are partially supported.
15097 (selftest::test_fixit_insert_containing_newline_2): New test.
15098 (selftest::test_fixit_replace_containing_newline): Update comments.
15099 (selftest::diagnostic_show_locus_c_tests): Call the new test.
15100 * edit-context.c (class added_line): New class.
15101 (class edited_line): Describe newline handling in comment.
15102 (edited_line::actually_edited_p): New method.
15103 (edited_line::print_content): Delete redundant decl.
15104 (edited_line::m_predecessors): New field.
15105 (edited_file::print_content): Call edited_line::print_content.
15106 (edited_file::print_diff): Update to support newlines.
15107 (edited_file::print_diff_hunk): Likewise.
15108 (edited_file::print_run_of_changed_lines): New function.
15109 (edited_file::print_diff_line): Convert to...
15110 (print_diff_line): ...this.
15111 (edited_file::get_effective_line_count): New function.
15112 (edited_line::edited_line): Initialize new field m_predecessors.
15113 (edited_line::~edited_line): Clean up m_predecessors.
15114 (edited_line::apply_fixit): Handle newlines.
15115 (edited_line::get_effective_line_count): New function.
15116 (edited_line::print_content): New function.
15117 (edited_line::print_diff_lines): New function.
15118 (selftest::test_applying_fixits_insert_containing_newline): New
15120 (selftest::test_applying_fixits_replace_containing_newline): New
15122 (selftest::insert_line): New function.
15123 (selftest::test_applying_fixits_multiple_lines): Add example of
15125 (selftest::edit_context_c_tests): Call the new tests.
15127 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15129 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
15130 parameter cand. Update dump information.
15131 (get_computation_cost): Update uses.
15133 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15135 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
15136 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
15137 (get_computation_at, rewrite_use_address): Update use of
15138 get_computation_aff.
15140 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15142 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
15143 (get_computation): Delete.
15144 (get_computation_cost): Implement like get_computation_cost_at.
15145 Use get_computation_at.
15146 (get_computation_cost_at): Delete.
15147 (rewrite_use_nonlinear_expr): Use get_computation_at.
15148 (rewrite_use_compare, remove_unused_ivs): Ditto.
15150 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15152 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
15154 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15156 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
15157 (ivopts_global_cost_for_size): Rename parameter and update uses.
15158 (iv_ca_recount_cost): Update uses.
15159 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
15160 candidates seperately in n_invs and n_cands.
15161 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
15163 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15165 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
15166 (find_inv_vars_cb): New.
15167 (find_depends): Renamed to ...
15168 (find_inv_vars): ... this.
15169 (add_candidate_1, force_var_cost): Call find_inv_vars.
15170 (split_address_cost, determine_group_iv_cost_cond): Ditto.
15172 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15174 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
15175 inv_vars. Add inv_exprs.
15176 (struct iv_cand): Rename depends_on to inv_vars.
15177 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
15178 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
15179 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
15180 (dump_cand): Dump inv_vars.
15181 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
15182 (record_invariant, find_depends, add_candidate_1): Ditto.
15183 (set_group_iv_cost, force_var_cost): Ditto.
15184 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
15185 (get_computation_cost_at, get_computation_cost): Ditto.
15186 (determine_group_iv_cost_generic): Ditto.
15187 (determine_group_iv_cost_address): Ditto.
15188 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
15189 (determine_group_iv_costs): Ditto.
15190 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
15191 (iv_ca_set_remove_invariants): Renamed to ...
15192 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
15193 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
15194 (iv_ca_set_add_invariants): Renamed to ...
15195 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
15196 (iv_ca_set_cp): Use iv_ca_set_add_invs.
15197 (iv_ca_has_deps): Support inv_vars and inv_exprs.
15198 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
15199 (create_new_ivs): Remove useless dump.
15201 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15203 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
15205 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
15206 (iv_ca_set_no_cp, create_new_iv): Ditto.
15208 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15210 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
15212 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
15214 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
15215 function tree_check2.
15217 2017-05-02 Martin Liska <mliska@suse.cz>
15219 * doc/gcov.texi: Add missing preposition.
15220 * gcov.c (function_info::function_info): Properly fill up
15221 all member variables.
15223 2017-05-02 Tamar Christina <tamar.christina@arm.com>
15225 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
15227 2017-05-02 Tamar Christina <tamar.christina@arm.com>
15229 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
15231 2017-05-02 Martin Liska <mliska@suse.cz>
15234 * lto-streamer-in.c (lto_read_tree_1): Remove
15235 LTO_STREAMER_DEBUG.
15236 * lto-streamer.c (struct tree_hash_entry): Likewise.
15237 (struct tree_entry_hasher): Likewise.
15238 (tree_entry_hasher::hash): Likewise.
15239 (tree_entry_hasher::equal): Likewise.
15240 (lto_streamer_init): Likewise.
15241 (lto_orig_address_map): Likewise.
15242 (lto_orig_address_get): Likewise.
15243 (lto_orig_address_remove): Likewise.
15244 * lto-streamer.h: Likewise.
15245 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
15246 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
15248 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
15250 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
15251 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
15252 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
15253 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
15254 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
15255 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
15256 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
15257 (mm_maskz_sub_ss): New intrinsics.
15258 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15259 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
15260 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
15261 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
15262 (__builtin_ia32_subss_mask_round): New builtins.
15263 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15264 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15265 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
15267 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
15268 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
15270 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
15273 2017-05-02 Martin Jambor <mjambor@suse.cz>
15275 PR tree-optimization/78687
15276 * tree-sra.c (access): New field parent.
15277 (process_subtree_disqualification): New function.
15278 (disqualify_candidate): Call it.
15279 (build_accesses_from_assign): Reset write flag if creating an
15281 (build_access_subtree): Fill in parent field and also prpagate
15282 down grp_write flag.
15283 (create_artificial_child_access): New parameter set_grp_write, set
15284 grp_write to its value.
15285 (propagate_subaccesses_across_link): Also propagate grp_write flag
15287 (propagate_all_subaccesses): Push the closest parent back to work
15288 queue if add_access_to_work_queue returned true.
15290 2017-05-02 Richard Biener <rguenther@suse.de>
15292 * common.opt (fstrict-overflow): Alias negative to fwrapv.
15293 * doc/invoke.texi (fstrict-overflow): Remove all traces of
15294 -fstrict-overflow documentation.
15295 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
15296 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
15297 flag_strict_overflow.
15298 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
15299 * lto-opts.c (lto_write_options): Do not stream it.
15300 * lto-wrapper.c (merge_and_complain): Do not handle it.
15301 * opts.c (default_options_table): Do not set -fstrict-overflow.
15302 (finish_options): Likewise do not clear it when sanitizing.
15303 * simplify-rtx.c (simplify_const_relational_operation): Do not
15304 test flag_strict_overflow.
15306 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
15308 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
15309 using enabled attribute.
15310 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
15311 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
15312 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
15313 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
15314 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
15315 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
15316 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
15317 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
15318 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
15319 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
15321 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
15323 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
15325 2017-05-02 Richard Biener <rguenther@suse.de>
15327 PR tree-optimization/80591
15329 2017-04-10 Richard Biener <rguenther@suse.de>
15331 * tree-ssa-structalias.c (find_func_aliases): Properly handle
15334 2017-05-02 Richard Biener <rguenther@suse.de>
15336 PR tree-optimization/80549
15337 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
15338 (cleanup_tree_cfg_noloop): Create forwarders to known loop
15339 headers if they do not have a preheader.
15341 2017-05-02 Martin Liska <mliska@suse.cz>
15344 * common.opt: Fix typo.
15345 * doc/invoke.texi: Likewise.
15347 2017-05-01 Jan Beulich <jbeulich@suse.com>
15349 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
15350 swapping, add (x,x,m,x,n) alternative.
15352 2017-05-01 Nathan Sidwell <nathan@acm.org>
15354 * calls.c (combine_pending_stack_adjustment_and_call): Remove
15355 unnecessary unadjusted_alignment check.
15357 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
15360 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
15362 * gimplify.c (gimplify_cilk_detach): New function.
15363 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
15364 * tree-core.h: Document EXPR_CILK_SPAWN.
15365 * tree.h (EXPR_CILK_SPAWN): Define.
15367 2017-05-01 David Malcolm <dmalcolm@redhat.com>
15369 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
15370 to use new fixit_hint representation, using the "replace" logic.
15371 (get_line_span_for_fixit_hint): Likewise.
15372 (layout::print_any_fixits): Likewise.
15373 (selftest::test_one_liner_many_fixits): Rename to...
15374 (selftest::test_one_liner_many_fixits_1): ...this, and update
15375 comment and expected output to reflect that the multiple fix-it
15376 hints are now consolidated into one insertion.
15377 (selftest::test_one_liner_many_fixits_2): New test.
15378 (selftest::test_diagnostic_show_locus_one_liner): Update for
15380 (selftest::test_fixit_consolidation): Update for fix-it API
15382 * diagnostic.c (print_parseable_fixits): Likewise.
15383 * edit-context.c (edited_line::m_line_events): Convert from
15384 auto_vec <line_event *> to auto_vec <line_event>.
15385 (class line_event): Convert from abstract base class to a concrete
15386 class, taking over the role of replace_event.
15387 (class insert_event): Delete.
15388 (class replace_event): Rename to class line_event. Convert to
15390 (edit_context::add_fixits): Reimplement.
15391 (edit_context::apply_insert): Delete.
15392 (edit_context::apply_replace): Rename to...
15393 (edit_context::apply_fixit): ...this. Convert to half-open range.
15394 (edited_file::apply_insert): Delete.
15395 (edited_file::apply_replace): Rename to...
15396 (edited_file::apply_fixit): ...this.
15397 (edited_line::~edited_line): Drop deletion of events.
15398 (edited_line::apply_insert): Delete.
15399 (edited_line::apply_replace): Rename to...
15400 (edited_line::apply_fixit): ...this. Convert to half-open range.
15401 Update for change to type of m_line_events.
15402 * edit-context.h (edit_context::apply_insert): Delete.
15403 (edit_context::apply_replace): Rename to...
15404 (edit_context::apply_fixit): ...this.
15406 2017-05-01 Martin Sebor <msebor@redhat.com>
15408 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
15411 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
15414 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
15415 __get_cpuid_max returns 0.
15416 (__get_cpuid_count): Ditto.
15418 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
15420 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
15421 replacement expression is another instance of one of its arguments.
15423 2017-05-01 Jakub Jelinek <jakub@redhat.com>
15426 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
15427 check for stack push/pop autoinc.
15428 * config/i386/i386.c (ix86_agi_dependent): Return false
15429 if the only reason why modified_in_p returned true is that
15430 addr is SP based and set_insn is a push or pop.
15432 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
15434 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
15437 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
15440 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
15441 (account_size_time): Use two predicates - exec_pred and
15443 (evaluate_conditions_for_known_args): Compute both clause and
15445 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
15446 (inline_summary_t::duplicate): Update.
15447 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
15449 (compute_inline_parameters): Likewise.
15450 (estimate_edge_size_and_time): Update caluclation of time.
15451 (estimate_node_size_and_time): Compute both time and nonspecialized
15453 (estimate_ipcp_clone_size_and_time): Update.
15454 (inline_merge_summary): Update.
15455 (do_estimate_edge_time): Update.
15456 (do_estimate_edge_size): Update.
15457 (do_estimate_edge_hints): Update.
15458 (inline_read_section, inline_write_summary): Stream both new predicates.
15459 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
15461 (compute_inlined_call_time): Cleanup.
15462 (big_speedup_p): Update.
15463 (edge_badness): Update.
15464 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
15465 (size_time_entry): Replace predicate by exec_predicate and
15466 nonconst_predicate.
15467 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
15468 (estimate_edge_time): Return also nonspec_time.
15469 (reset_edge_growth_cache): Update.
15471 2017-04-29 Jakub Jelinek <jakub@redhat.com>
15473 PR rtl-optimization/80491
15474 * ifcvt.c (noce_process_if_block): When looking for x setter
15475 with missing else_bb, don't check only the insn right before
15476 cond_earliest, but look for the last insn that x is modified in
15477 within the same bb.
15479 PR rtl-optimization/80491
15480 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
15482 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
15484 PR tree-optimization/80487
15485 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
15487 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15489 PR tree-optimization/79697
15490 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
15491 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
15492 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
15494 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
15495 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
15497 2017-04-28 Martin Sebor <msebor@redhat.com>
15499 PR tree-optimization/80523
15500 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
15501 (init_target_to_host_charmap, target_to_host, target_strtol10): New
15503 (maybe_warn, format_directive, parse_directive): Use new functions.
15504 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
15506 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
15508 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
15510 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
15512 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
15513 target_header_dir): Set correctly.
15514 * configure: Regenerated.
15515 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
15516 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
15517 instead of SYSTEM_HEADER_DIR.
15519 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
15521 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
15522 (estimate_local_effects): Likewise.
15523 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
15524 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
15525 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
15526 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
15527 do_estimate_edge_time, estimate_edge_time): Likewise.
15528 * ipa-inline-analysis.c (estimate_node_size_and_time,
15529 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
15530 (estimate_time_after_inlining): Remove.
15532 2017-04-28 Martin Liska <mliska@suse.cz>
15534 * doc/gcov.texi: Enhance documentation of gcov.
15536 2017-04-28 Martin Liska <mliska@suse.cz>
15538 * doc/gcov.texi: Sort options in alphabetic order.
15539 * doc/gcov-dump.texi: Likewise.
15540 * doc/gcov-tool.texi: Likewise.
15541 * gcov.c (print_usage): Likewise.
15542 * gcov-dump.c (print_usage): Likewise.
15543 * gcov-tool.c (print_merge_usage_message): Likewise.
15544 (print_rewrite_usage_message): Likewise.
15545 (print_overlap_usage_message): Likewise.
15547 2017-04-28 Martin Liska <mliska@suse.cz>
15549 PR gcov-profile/53915
15550 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
15552 2017-04-28 Martin Liska <mliska@suse.cz>
15554 PR gcov-profile/79891
15555 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
15556 is marked by compiler as living on a line.
15557 (get_cycles_count): Remove usage of the union.
15558 (output_intermediate_file): Likewise.
15559 (find_source): Fix GNU coding style.
15560 (accumulate_line_counts): Remove old non-all block mode.
15561 (output_lines): Remove usage of the union.
15562 * profile.c (output_location): Include all BBs, even if
15563 belonging to a same line (and file) as a previous BB.
15565 2017-04-28 Martin Liska <mliska@suse.cz>
15567 * gcov.c (process_args): Handle new argument 'w'.
15568 (read_graph_file): Assign ID to BBs.
15569 (output_branch_count): Display BB # if verbose flag is set.
15570 (output_lines): Likewise for arcs.
15571 (print_usage): Add '--verbose' option help.
15572 * doc/gcov.texi: Document --verbose (-w) option.
15574 2017-04-28 Martin Liska <mliska@suse.cz>
15576 * gcov.c (struct block_location_info): New struct.
15577 (process_file): Fill up the new structure.
15578 (read_graph_file): Replace usage of encoding by the newly added
15580 (add_line_counts): Likewise.
15581 (accumulate_line_counts): Remove usage of the union.
15582 (function_info::function_info): New function.
15583 (function_info::~function_info): Likewise.
15584 (process_file): Call delete instead of release_function.
15585 (release_function): Release the function.
15586 (release_structures): Call delete instead of release_function.
15587 (solve_flow_graph): Replace usage of num_blocks.
15588 (find_exception_blocks): Likewise.
15589 (output_lines): Fix GNU coding style.
15591 2017-04-28 Martin Liska <mliska@suse.cz>
15594 * coverage.c (coverage_remove_note_file): New function.
15595 * coverage.h: Declare the function.
15596 * toplev.c (finalize): Clean if an error has been seen.
15598 2017-04-28 Martin Liska <mliska@suse.cz>
15600 PR gcov-profile/80031
15601 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
15602 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
15603 * gcov.c (read_graph_file): Read just number of blocks.
15604 * profile.c (branch_prob): Do not stream 0 flags per a basic
15607 2017-04-28 Martin Liska <mliska@suse.cz>
15609 * gcov-dump.c (tag_*): Add new argument to declarations.
15610 (dump_gcov_file): Likewise.
15611 (tag_blocks): Add and use new argument depth.
15612 (tag_arcs): Likewise.
15613 (tag_lines): Likewise.
15614 (tag_counters): Likewise.
15615 (tag_summary): Likewise.
15616 (dump_working_sets): Use depth to do a proper indentation.
15618 2017-04-28 Jakub Jelinek <jakub@redhat.com>
15621 * cgraph.h (symtab_node::debug_symtab): No longer inline.
15622 * symtab.c (symtab_node::debug_symtab): Move definition here.
15624 2017-04-28 Richard Biener <rguenther@suse.de>
15626 * lto-streamer.h (LTO_major_version): Bump to 7.
15628 2017-04-28 Richard Biener <rguenther@suse.de>
15630 * tree-vrp.c (assert_info): New struct.
15631 (add_assert_info): New helper.
15632 (register_edge_assert_for_2): Refactor to add asserts to a vector
15634 (register_edge_assert_for_1): Likewise.
15635 (register_edge_assert_for): Likewise.
15636 (finish_register_edge_assert_for): New helper actually registering
15637 asserts where live on edge.
15638 (find_conditional_asserts): Adjust.
15639 (find_switch_asserts): Likewise.
15640 (evrp_dom_walker::try_find_new_range): Generalize.
15641 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
15643 2017-04-27 Marek Polacek <polacek@redhat.com>
15646 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
15647 arg10 and arg11 to itype.
15649 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
15651 * doc/extend.texi (Object Size Checking): Improve grammar.
15653 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
15656 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
15657 that the logic for permitting reciprocal estimates matches that
15660 2017-04-27 Jakub Jelinek <jakub@redhat.com>
15663 * tree.c (type_cache_hasher::equal): Only compare
15664 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
15665 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
15666 non-aggregate element types.
15667 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
15668 about the flag on ARRAY_TYPEs in the comment, formatting fix.
15670 2017-04-27 Richard Biener <rguenther@suse.de>
15672 PR middle-end/80533
15673 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
15674 stripping ARRAY_REFs from MEM_EXPR make sure we're not
15675 keeping a reference to a trailing array.
15677 2017-04-27 Richard Biener <rguenther@suse.de>
15679 PR middle-end/80539
15680 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
15681 being in loop-closed SSA form conservatively.
15682 (chrec_fold_multiply_poly_poly): Likewise.
15684 2017-04-27 Tamar Christina <tamar.christina@arm.com>
15686 PR middle-end/79665
15687 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
15688 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
15690 2017-04-27 Jakub Jelinek <jakub@redhat.com>
15693 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
15694 (aarch64_function_arg_alignment): Return unsigned int again, but still
15695 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
15696 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
15697 Don't emit -Wpsabi note.
15698 (aarch64_function_arg_boundary): Likewise.
15699 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
15702 2017-04-26 Nathan Sidwell <nathan@acm.org>
15704 * tree.h (crc32_unsigned_n): Declare.
15705 (crc32_unsigned, crc32_unsigned): Make inline.
15706 * tree.c (crc32_unsigned_bits): Replace with ...
15707 (crc32_unsigned_n): ... this.
15708 (crc32_unsigned, crc32_byte): Remove.
15709 (crc32_string): Remove unnecessary braces.
15711 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
15713 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
15714 * ipa-inline-analysis.c (MAX_TIME): Remove.
15715 (account_size_time): Use sreal for time.
15716 (dump_inline_summary): Update.
15717 (estimate_function_body_sizes): Update.
15718 (estimate_edge_size_and_time): Update.
15719 (estimate_calls_size_and_time): Update.
15720 (estimate_node_size_and_time): Update.
15721 (inline_merge_summary): Update.
15722 (inline_update_overall_summary): Update.
15723 (estimate_time_after_inlining): Update.
15724 (inline_read_section): Update.
15725 (inline_write_summary): Update.
15726 * ipa-inline.c (compute_uninlined_call_time): Update.
15727 (compute_inlined_call_time): Update.
15728 (recursive_inlining): Update.
15729 (inline_small_functions): Update.
15730 (dump_overall_stats): Update.
15731 * ipa-inline.h: Include sreal.h.
15732 (size_time_entry): Turn time to sreal.
15733 (inline_summary): Turn self_time nad time to sreal.
15735 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
15737 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
15739 (sreal::stream_out, sreal::stream_in): New.
15740 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
15742 2017-04-25 Jakub Jelinek <jakub@redhat.com>
15744 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
15747 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
15750 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
15752 (dimode_scalar_chain::compute_convert_gain): Ditto.
15753 (dimode_scalar_chain::make_vector_copies): Ditto.
15754 (dimode_scalar_chain::convert_reg): Ditto.
15755 (dimode_scalar_chain::convert_insn): Ditto.
15756 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
15757 (VI248_AVX512BW_1): New mode iterator.
15758 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
15759 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
15762 2017-04-25 Martin Sebor <msebor@redhat.com>
15764 PR tree-optimization/80497
15765 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
15766 constants are representable in HOST_WIDE_INT.
15767 (parse_directive): Ditto.
15769 2017-04-25 Martin Sebor <msebor@redhat.com>
15772 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
15773 (new_zero_array): Adjust signature.
15774 (dom_info::dom_init): Used unsigned rather that size_t.
15775 (dom_info::dom_info): Same.
15777 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15778 Jakub Jelinek <jakub@redhat.com>
15781 * config/arm/arm.c: Include gimple.h.
15782 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
15783 returns negative, increment ncrn only if it returned positive.
15784 (arm_needs_doubleword_align): Return int instead of bool,
15785 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
15786 members, but if there is any such non-FIELD_DECL
15787 > PARM_BOUNDARY aligned decl, return -1 instead of false.
15788 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
15789 returns negative, increment nregs only if it returned positive.
15790 (arm_setup_incoming_varargs): Likewise.
15791 (arm_function_arg_boundary): Emit -Wpsabi note if
15792 arm_needs_doubleword_align returns negative, return
15793 DOUBLEWORD_ALIGNMENT only if it returned positive.
15795 2017-04-25 Marek Polacek <polacek@redhat.com>
15798 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
15799 first argument to type.
15801 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
15804 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
15805 type checks to test for compatibility instead of equality.
15807 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15808 Jakub Jelinek <jakub@redhat.com>
15811 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
15813 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
15814 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
15815 the alignment computation, but return their maximum in warn_alignment.
15816 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
15817 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
15819 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
15820 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
15823 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15825 * config/arc/simdext.md (dmpyh): Fix typo.
15827 2017-04-25 Richard Biener <rguenther@suse.de>
15829 PR tree-optimization/80492
15830 * alias.c (compare_base_decls): Handle registers with asm
15831 specification conservatively.
15832 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
15833 compare_base_decls returning dont-know properly.
15835 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15837 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
15838 (legitimate_offset_address_p): New function.
15839 (arc_legitimate_address_p): Use above function.
15841 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15843 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
15845 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15847 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
15848 ACCH registers whenever they are available.
15850 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15852 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
15853 double regs fix when not used.
15855 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15857 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
15859 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
15860 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
15862 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15864 * config/arc/arc.c (arc_output_addsi): Check for h-register class
15865 when emitting short ADD instructions.
15867 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15869 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
15871 (cmpsi_cc_c_insn): Likewise.
15872 (cbranchsi4_scratch): Compute proper instruction length using
15873 compact_hreg_operand.
15874 * config/arc/predicates.md (compact_hreg_operand): New predicate.
15876 2017-04-25 Richard Biener <rguenther@suse.de>
15878 PR middle-end/80509
15879 * passes.c (pass_manager::pass_manager): Initialize
15880 m_name_to_pass_map.
15882 2017-04-25 Richard Biener <rguenther@suse.de>
15884 PR tree-optimization/79201
15885 * tree-ssa-sink.c (statement_sink_location): Handle calls.
15887 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15890 * config/s390/vector.md: Split MEM->GPR vector moves for
15891 non-s_operand addresses.
15893 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15896 * config/s390/predicates.md (reload_const_wide_int_operand): New
15898 * config/s390/s390.md ("movti"): Remove d/P alternative.
15899 ("movti_bigconst"): New pattern definition.
15901 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15904 * s390-protos.h (s390_expand_cs_hqi): Removed.
15905 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
15906 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
15907 modes as well as CCZ1mode and CCZmode.
15908 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
15909 signature of s390_emit_compare_and_swap.
15910 (s390_expand_cs_hqi): Likewise, make static.
15911 (s390_expand_cs_tdsi): Generate an explicit compare before trying
15912 compare-and-swap, in some cases.
15913 (s390_expand_cs): Wrapper function.
15914 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
15916 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
15917 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
15918 patterns for small and large integers. Forbid symref memory operands.
15919 Move expander to s390.c. Require cc register.
15920 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
15921 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
15922 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
15923 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
15924 symref memory operands. Remove CC mode and call s390_match_ccmode
15926 ("atomic_exchange<mode>"): Allow and implement all integer modes.
15928 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15930 * config/s390/s390.md (define_peephole2): New peephole to help
15931 combining the load-and-test pattern with volatile memory.
15933 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15935 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
15936 with CCZmode for TARGET_Z196.
15938 2017-04-25 Jakub Jelinek <jakub@redhat.com>
15940 PR rtl-optimization/80501
15941 * combine.c (make_compound_operation_int): Set subreg_code to SET
15942 even for AND with mask of the sign bit of mode.
15944 PR rtl-optimization/80500
15945 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
15946 sum's initial value.
15948 2017-04-25 Julian Brown <julian@codesourcery.com>
15949 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15951 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
15953 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
15955 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
15957 2017-04-25 Julian Brown <julian@codesourcery.com>
15958 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15960 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
15961 (thunderx2t99_sha): New Reservation.
15963 2017-04-25 Julian Brown <julian@codesourcery.com>
15964 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15966 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
15967 type for 1-element load.
15969 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
15971 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
15973 2017-04-24 Martin Jambor <mjambor@suse.cz>
15975 PR tree-optimization/80293
15976 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
15977 char arrays not totally scalarizable if it is false.
15978 (analyze_all_variable_accesses): Pass correct value in the new
15979 parameter. Add a statistics counter.
15981 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
15983 PR middle-end/79931
15984 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
15986 2017-04-24 Richard Biener <rguenther@suse.de>
15988 PR tree-optimization/80494
15989 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
15990 out for complex types.
15992 2017-04-24 Richard Biener <rguenther@suse.de>
15994 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
15995 * tree-ssa-sccvn.c (print_scc): Print SCC size.
15996 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
15997 (DFS): Adjust and never fail.
15998 (sccvn_dom_walker::fail): Remove.
15999 (sccvn_dom_walker::before_dom_children): Adjust.
16000 (run_scc_vn): Likewise and never fail.
16001 * tree-ssa-pre.c (pass_pre::execute): Adjust.
16002 (pass_fre::execute): Likewise.
16004 2017-04-24 Richard Biener <rguenther@suse.de>
16006 PR tree-optimization/79725
16007 * tree-ssa-sink.c (statement_sink_location): Return whether
16008 failure reason was zero uses. Move that check later.
16009 (sink_code_in_bb): Deal with zero uses by removing the stmt
16012 2017-04-24 Richard Biener <rguenther@suse.de>
16015 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
16016 pointer-based references.
16018 2017-04-24 Richard Biener <rguenther@suse.de>
16021 * pass_manager.h (pass_manager::operator new): Remove.
16022 (pass_manager::operator delete): Likewise.
16023 * passes.c (pass_manager::operator new): Remove.
16024 (pass_manager::operator delete): Likewise.
16025 (pass_manager::pass_manager): Zero individual pass members.
16027 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
16030 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
16031 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
16032 Check "XEXP (src, 1)" operand here.
16033 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
16034 Check "XEXP (src, 1)" operand here.
16035 (dimode_scalar_chain::make_vector_copies): Detect count register
16036 of a shift instruction. Zero extend count register from QImode
16037 to DImode to satisfy vector shift pattern count operand predicate.
16038 Substitute vector shift count operand with a DImode copy.
16039 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
16042 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
16044 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
16045 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
16046 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
16047 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
16048 (UNSPEC_NOREX_MEM): Remove definition.
16050 2017-04-21 Richard Biener <rguenther@suse.de>
16052 PR tree-optimization/79547
16053 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
16054 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
16055 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
16056 without any constraints.
16058 2017-04-21 Richard Biener <rguenther@suse.de>
16060 PR tree-optimization/78847
16061 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
16063 2017-04-21 Richard Biener <rguenther@suse.de>
16065 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
16066 (build_distinct_type_copy): Likewise.
16067 (build_variant_type_copy): Likewise.
16068 * tree.c (build_qualified_type): Pass down mem-stat info.
16069 (build_distinct_type_copy): Likewise.
16070 (build_variant_type_copy): Likewise.
16072 2017-04-21 Richard Biener <rguenther@suse.de>
16074 PR tree-optimization/80237
16075 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
16077 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
16078 for a simplified result.
16080 2016-04-21 Richard Biener <rguenther@suse.de>
16082 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
16083 sth as strict as a simple_iv but a chrec without symbols and an
16084 operand defined in the loop we are peeling (and not some subloop).
16085 (propagate_constants_for_unrolling): Propagate all constants.
16087 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
16090 * config/i386/i386.c (print_reg): Remove assert for disalowed
16091 regno values, call output_operand_lossage instead.
16093 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
16096 * config/i386/constraints.md (Yc): New register constraint.
16097 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
16098 Use Yc constraint for alternative 2 of operand 0. Remove
16099 preferred_for_speed attribute.
16101 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
16103 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
16104 lastprivate clauses in SIMT case.
16106 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
16108 * doc/invoke.texi (-Wextra-semi): Document new warning option.
16110 2017-04-20 Richard Biener <rguenther@suse.de>
16112 PR tree-optimization/57796
16113 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
16114 as N scalar stores.
16115 (vect_model_load_cost): Cost gathers as N scalar loads.
16117 2017-04-20 Richard Biener <rguenther@suse.de>
16119 * ggc-page.c (ggc_allocated_p): Rename to ...
16120 (safe_lookup_page_table_entry): ... this and return the lookup
16122 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
16124 2017-04-20 Richard Biener <rguenther@suse.de>
16126 PR tree-optimization/80453
16127 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
16128 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
16129 from the conditions.
16130 (vn_phi_eq): Pass them down.
16131 (vn_phi_lookup): Record them.
16132 (vn_phi_insert): Likewise.
16134 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
16136 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
16137 uninitialized variable warning to avoid buffer overrun.
16139 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
16142 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
16143 is suppressed for '{ 0 }' in C.
16145 2017-04-20 Jakub Jelinek <jakub@redhat.com>
16147 * BASE-VER: Set to 8.0.0.
16149 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
16151 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
16152 priority .init_array and .fini_array section with SECTION_NOTYPE
16155 2017-04-20 Jakub Jelinek <jakub@redhat.com>
16157 PR middle-end/80423
16158 * tree.h (build_array_type): Add typeless_storage default argument.
16159 * tree.c (type_cache_hasher::equal): Also compare
16160 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
16161 (build_array_type): Add typeless_storage argument, set
16162 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
16164 (build_nonshared_array_type): Adjust build_array_type_1 caller.
16165 (build_array_type): Likewise. Add typeless_storage argument.
16167 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
16168 Jakub Jelinek <jakub@redhat.com>
16170 PR tree-optimization/80426
16171 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
16172 operation on symbolic operands, also compute the overflow for the
16173 invariant part when the operation degenerates into a negation.
16175 2017-04-19 Jakub Jelinek <jakub@redhat.com>
16178 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
16179 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
16182 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
16184 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
16187 * config/avr/avr.c (tree.h): Include it.
16188 (cgraph.h): Include it.
16189 (avr_encode_section_info): Don't warn for uninitialized progmem
16190 variable if it's just an alias.
16192 2017-04-19 Richard Biener <rguenther@suse.de>
16195 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
16196 when needed by AutoPGO.
16198 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
16201 * doc/lto.texi: Remove an extra 'that'.
16203 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
16205 PR rtl-optimization/80429
16206 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
16207 are only used in debug insns.
16209 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
16210 Vladimir Makarov <vmakarov@redhat.com>
16212 * config/sparc/predicates.md (input_operand): Add comment. Return
16213 true for any memory operand when LRA is in progress.
16214 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
16216 2017-04-18 Jeff Law <law@redhat.com>
16219 * mips.md ({return,simple_return}_internal): Do not overwrite
16222 2017-04-18 Jakub Jelinek <jakub@redhat.com>
16224 PR tree-optimization/80443
16225 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
16226 instead of adding 1, subtract -1 and similarly instead of subtracting
16229 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
16231 PR rtl-optimization/80357
16232 * haifa-sched.c (tmp_bitmap): New variable.
16233 (model_recompute): Handle duplicate use records.
16234 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
16235 (free_global_sched_pressure_data): Free it.
16237 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
16240 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
16241 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
16242 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
16243 instead of SYSTEM_HEADER_DIR.
16245 2017-04-18 Jeff Law <law@redhat.com>
16247 PR middle-end/80422
16248 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
16249 predecessors after walking up the insn chain.
16251 2017-04-18 Jakub Jelinek <jakub@redhat.com>
16254 * dwarf2out.c (modified_type_die): Try harder not to emit internal
16255 sizetype type into debug info.
16257 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
16260 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
16261 unneeded test for TARGET_UPPER_REGS_SF.
16262 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
16264 2017-04-18 Jakub Jelinek <jakub@redhat.com>
16267 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
16268 instead of gsi_after_labels.
16270 2017-04-18 Jeff Law <law@redhat.com>
16272 * regcprop.c (maybe_mode_change): Avoid creating copies of the
16276 2017-04-13 Jeff Law <law@redhat.com>
16277 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
16278 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
16280 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
16283 * config/avr/avr.c (intl.h): Include it.
16284 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
16286 2017-04-18 Martin Liska <mliska@suse.cz>
16288 PR gcov-profile/78783
16289 * gcov-tool.c (gcov_output_files): Validate that destination
16290 file is either removed by the tool or by a user.
16292 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
16293 Guy Benyei <guybe@mellanox.com>
16295 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
16296 block, and do not negate it, the stored id is already negative.
16298 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
16300 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
16302 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
16305 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
16306 masks of options that should be turned off if the VSX vector
16307 options are turned off.
16308 (OTHER_P8_VECTOR_MASKS): Likewise.
16309 (OTHER_VSX_VECTOR_MASKS): Likewise.
16310 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
16311 rs6000_disable_incompatible_switches to validate no type switches
16313 (rs6000_incompatible_switch): New function to disallow turning on
16314 other vector options if -mno-vsx, -mno-power8-vector, or
16315 -mno-power9-vector are specified.
16317 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
16319 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
16321 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
16323 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
16324 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
16325 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
16326 (ARG_POINTER_CFA_OFFSET): Likewise.
16328 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
16330 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
16331 conditions to take advantage of various optimizations.
16333 2017-04-13 Jeff Law <law@redhat.com>
16335 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
16336 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
16337 (zero_extendsidi2_dext): Likewise.
16339 2017-04-13 Jakub Jelinek <jakub@redhat.com>
16342 * fold-const.c (fold_ternary_loc): Revert
16343 use op0 instead of fold_convert_loc (loc, type, arg0) part of
16346 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
16348 PR rtl-optimization/80343
16349 * lra-remat.c (update_scratch_ops): Assign original hard reg to
16350 new scratch pseudo.
16352 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
16355 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
16356 to ubsan_encode_value.
16358 2017-04-13 Jeff Law <law@redhat.com>
16360 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
16361 appearing in DEBUG_INSNs.
16363 2017-04-13 Martin Liska <mliska@suse.cz>
16365 PR gcov-profile/80413
16366 * gcov-io.c (gcov_write_string): Copy to buffer just when
16367 allocated size is greater than zero.
16369 2017-04-13 Jakub Jelinek <jakub@redhat.com>
16372 * dwarf2out.c (decls_for_scope): Ignore declarations of
16373 current_function_decl in BLOCK_NONLOCALIZED_VARS.
16375 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
16378 * ipa-visibility.c (non_local_p): Fix typos.
16379 (localize_node): When localizing symbol in same comdat group,
16380 dissolve the group only when we know external symbols are going
16382 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
16384 2017-04-12 Jakub Jelinek <jakub@redhat.com>
16386 PR tree-optimization/79390
16387 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
16388 order does not result in usable sequence, retry with reversed operand
16394 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
16395 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
16396 op0 instead of fold_convert_loc (loc, type, arg0).
16398 2017-04-12 Jeff Law <law@redhat.com>
16400 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
16401 has a delay slot in the generated code.
16403 * config/cris/cris.md (cris_preferred_reload_class): Return
16404 GENNONACR_REGS rather than GENERAL_REGS.
16406 2017-04-12 Jakub Jelinek <jakub@redhat.com>
16409 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
16410 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
16411 signedness of the result type.
16413 2017-04-12 Richard Biener <rguenther@suse.de>
16414 Jeff Law <law@redhat.com>
16416 PR tree-optimization/80359
16417 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
16418 trim stores to TARGET_MEM_REFs.
16420 2017-04-12 Richard Biener <rguenther@suse.de>
16422 PR tree-optimization/79390
16423 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
16424 threading case even more.
16426 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
16429 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
16430 for quad_address_p for TImode, instead of just not indexed_address.
16432 2017-04-12 Richard Biener <rguenther@suse.de>
16433 Bernd Edlinger <bernd.edlinger@hotmail.de>
16435 PR middle-end/79671
16436 * alias.c (component_uses_parent_alias_set_from): Handle
16437 TYPE_TYPELESS_STORAGE.
16438 (get_alias_set): Likewise.
16439 * tree-core.h (tree_type_common): Add typeless_storage flag.
16440 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
16441 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
16442 for types containing members with TYPE_TYPELESS_STORAGE.
16443 (place_field): Likewise.
16444 (layout_type): Likewise for ARRAY_TYPE.
16445 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
16446 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
16447 TYPE_TYPELESS_STORAGE.
16448 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
16450 2017-04-12 Jakub Jelinek <jakub@redhat.com>
16453 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
16454 first argument to type.
16456 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16460 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
16461 CONST0_RTX (mode) rather than const0_rtx where appropriate.
16462 (rs6000_expand_binop_builtin): Likewise.
16463 (rs6000_expand_ternop_builtin): Likewise; also add missing
16464 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
16465 vshasigma built-ins.
16466 * doc/extend.texi: Document that vec_xxpermdi's third argument
16467 must be a constant.
16469 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
16471 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
16472 Use shift_const cost parameter when calculating gain of STV shifts.
16474 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
16476 PR rtl-optimization/70478
16477 * lra-constraints.c (process_alt_operands): Check memory for
16478 disfavoring memory insn operand.
16480 2017-04-11 Jakub Jelinek <jakub@redhat.com>
16482 PR middle-end/80100
16483 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
16484 left shift in unsigned HOST_WIDE_INT type.
16486 PR rtl-optimization/80385
16487 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
16488 (not (neg X)) into (plus X -1) for complex or non-integral modes.
16491 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
16492 if they have any depend clauses.
16494 2017-04-11 Martin Liska <mliska@suse.cz>
16497 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
16498 * ipa-split.c (split_function): Create a local comdat symbol
16499 if caller is in a comdat group.
16501 2017-04-11 Martin Liska <mliska@suse.cz>
16504 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
16507 2017-04-11 Martin Sebor <msebor@redhat.com>
16509 PR middle-end/80364
16510 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
16511 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
16513 (directive::set_width, directive::set_precision, format_character):
16515 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
16518 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
16521 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
16522 conflict, set target->arch_name instead of target->cpu_name.
16524 2017-04-11 Richard Biener <rguenther@suse.de>
16526 PR tree-optimization/80374
16527 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
16528 build_zero_cst, remove fold_convertible_p check again.
16530 2017-04-11 Martin Liska <mliska@suse.cz>
16533 * ubsan.c (instrument_object_size): Do not instrument register
16536 2017-04-11 Jakub Jelinek <jakub@redhat.com>
16539 * config/i386/i386-builtin-types.def
16540 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
16541 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
16542 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
16543 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
16544 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
16545 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
16546 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
16547 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
16548 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
16549 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
16550 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
16551 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
16552 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
16553 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
16554 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
16555 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
16556 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
16557 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
16558 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
16559 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
16560 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
16561 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
16562 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
16563 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
16564 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
16565 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
16566 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
16567 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
16568 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
16569 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
16570 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
16571 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
16572 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
16573 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
16574 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
16575 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
16576 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
16577 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
16578 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
16579 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
16580 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
16581 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
16582 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
16583 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
16584 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
16585 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
16587 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
16588 flag to second_arg_count, handle 4 argument function type _COUNT
16589 aliases, handle second_arg_count on second argument rather than last.
16591 2017-04-10 Jeff Law <law@redhat.com>
16593 PR tree-optimization/80374
16594 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
16595 record anything if we can not convert integer_zero_node to the
16598 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
16601 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16602 Enhance special handling given to the TARGET_P9_MINMAX option in
16603 relation to certain other options.
16605 2017-04-10 Bin Cheng <bin.cheng@arm.com>
16607 PR tree-optimization/80153
16608 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
16609 remove POINTER_PLUS_EXPR's base part directly, rather than through
16612 2017-04-10 Richard Biener <rguenther@suse.de>
16613 Bin Cheng <bin.cheng@arm.com>
16615 PR tree-optimization/80153
16616 * tree-affine.c (aff_combination_to_tree): Get base pointer from
16617 the first element of pointer type aff_tree. Build result expr in
16619 (add_elt_to_tree): Convert to type unconditionally. Remove other
16620 fold_convert calls.
16621 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
16622 (rewrite_use_nonlinear_expr): Check invariant using iv information.
16624 2017-04-10 Richard Biener <rguenther@suse.de>
16626 * tree-ssa-structalias.c (find_func_aliases): Properly handle
16629 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
16631 PR rtl-optimization/70478
16632 * lra-constraints.c (curr_small_class_check): New.
16633 (update_and_check_small_class_inputs): New.
16634 (process_alt_operands): Update curr_small_class_check. Disfavor
16635 alternative insn memory operands. Check available regs for small
16638 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
16641 * config/mips/mips.opt (-mvirt): Update description.
16642 * doc/invoke.texi (-mvirt): Likewise.
16644 2017-04-10 Richard Biener <rguenther@suse.de>
16646 PR middle-end/80362
16647 * fold-const.c (fold_binary_loc): Look at unstripped ops when
16648 looking for NEGATE_EXPR in -A / -B to A / B folding.
16650 2017-04-10 Martin Liska <mliska@suse.cz>
16652 PR gcov-profile/80224
16653 * gcov.c (print_usage): Fix usage string.
16654 (get_gcov_intermediate_filename): Remove.
16655 (output_gcov_file): Use both for normal and intermediate format.
16656 (generate_results): Do not initialize special file for
16657 intermediate format.
16659 2017-04-10 Richard Biener <rguenther@suse.de>
16661 PR tree-optimization/80304
16662 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
16665 2017-04-10 Nathan Sidwell <nathan@acm.org>
16668 * config/rs6000/rs6000.c (rs6000_vector_type): New.
16669 (rs6000_init_builtins): Use it.
16671 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16673 * config/arm/arm.md (<mrc>): Add mode to SET source.
16674 (<mrrc>): Likewise.
16676 2017-04-10 Richard Biener <rguenther@suse.de>
16678 PR middle-end/80344
16679 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
16681 2017-04-10 Jakub Jelinek <jakub@redhat.com>
16684 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
16685 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
16686 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
16687 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
16688 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
16689 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
16690 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
16691 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
16692 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
16693 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
16694 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
16695 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
16696 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
16697 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
16698 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
16699 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
16700 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
16701 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
16702 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
16703 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
16704 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
16705 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
16706 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
16708 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
16710 PR rtl-optimization/70478
16711 * lra-constraints.c: Reverse the last patch.
16713 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
16715 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
16716 Add comment for WCHAR_T.
16718 2017-04-08 Martin Liska <mliska@suse.cz>
16721 2017-04-07 Martin Liska <mliska@suse.cz>
16724 * ipa-split.c (split_function): Add function part to a same comdat
16727 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16730 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
16732 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
16734 * rs6000/rs6000.c (vec_load_pendulum): Rename...
16735 (vec_pairing): ...to this.
16736 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
16737 (rs6000_sched_init): Adjust for name change.
16738 (struct rs6000_sched_context): Likewise.
16739 (rs6000_init_sched_context): Likewise.
16740 (rs6000_set_sched_context): Likewise.
16742 2017-04-07 Jakub Jelinek <jakub@redhat.com>
16748 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
16750 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
16751 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
16752 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
16754 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
16756 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
16758 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
16760 PR rtl-optimization/70703
16761 * ira-color.c (update_conflict_hard_regno_costs): Use
16762 int64_t instead of HOST_WIDE_INT.
16764 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
16766 PR rtl-optimization/70478
16767 * lra-constraints.c (process_alt_operands): Disfavor alternative
16768 insn memory operands.
16770 2017-04-07 Jeff Law <law@redhat.com>
16772 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
16773 CALL and NOTE_INSN_CALL_ARG_LOCATION.
16775 2017-04-07 Martin Liska <mliska@suse.cz>
16778 * config/aarch64/aarch64.c (aarch64_process_target_attr):
16779 Show error message instead of an ICE.
16781 2017-04-07 Martin Liska <mliska@suse.cz>
16784 * ipa-split.c (split_function): Add function part to a same comdat
16787 2017-04-07 Richard Biener <rguenther@suse.de>
16789 PR middle-end/80341
16790 * tree.c (get_unwidened): Also handle ! for_type case for
16792 * convert.c (do_narrow): Split out from ...
16793 (convert_to_integer_1): ... here. Do not pass final truncation
16794 type to get_unwidened for TRUNC_DIV_EXPR.
16796 2017-04-07 Richard Biener <rguenther@suse.de>
16798 * tree-affine.c (wide_int_ext_for_comb): Take type rather
16800 (aff_combination_const): Adjust.
16801 (aff_combination_scale): Likewise.
16802 (aff_combination_add_elt): Likewise.
16803 (aff_combination_add_cst): Likewise.
16804 (aff_combination_convert): Likewise.
16805 (add_elt_to_tree): Likewise. Remove unused argument.
16806 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
16808 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
16810 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
16812 * config/arm/arm.c (arm_default_short_enums): Use
16813 ARM_DEFAULT_SHORT_ENUMS.
16814 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
16816 2017-04-06 Jakub Jelinek <jakub@redhat.com>
16819 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
16820 members with redundant out-of-class redeclaration.
16822 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
16825 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
16826 * config/i386/i386.md (*zero_extendsidi2):
16827 Add (?*x,*x) and (?*v,*v) alternatives.
16829 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
16832 * config/i386/i386.c (ix86_expand_builtin)
16833 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
16834 mode from insn data. Convert operands to insn operand mode.
16835 Copy operands that don't satisfy insn predicate to a register.
16837 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
16839 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
16842 2017-04-06 Richard Biener <rguenther@suse.de>
16844 PR tree-optimization/80334
16845 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
16846 preserve alignment of accesses.
16848 2017-04-06 Richard Biener <rguenther@suse.de>
16850 PR tree-optimization/80262
16851 * tree-sra.c (build_ref_for_offset): Preserve address-space
16853 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
16854 Drop useless address-space information on MEM_REF offsets.
16856 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
16858 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
16860 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
16862 PR rtl-optimization/70703
16863 * ira-color.c (update_conflict_hard_regno_costs): Use
16864 HOST_WIDE_INT instead of long.
16866 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
16869 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
16870 not defined for x86_64 target. Add -mmmx target option when __SSE2__
16872 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
16873 for x86_64 target. Handle -m3dnowa option.
16875 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
16877 PR rtl-optimization/70703
16878 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
16879 (update_conflict_hard_regno_costs): Use long instead of unsigned
16880 arithmetic for cost calculation.
16882 2017-04-05 Jakub Jelinek <jakub@redhat.com>
16883 Bernd Edlinger <bernd.edlinger@hotmail.de>
16886 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
16889 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
16892 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
16893 ptr_mode with Pmode throughout.
16894 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
16895 into probe_stack_range and use DImode.
16897 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
16900 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
16901 call_eh_return is true.
16903 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16905 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
16906 Initialize last_match_fntype_index.
16908 2017-04-05 Jakub Jelinek <jakub@redhat.com>
16911 * tree-nvr.c: Include internal-fn.h.
16912 (pass_return_slot::execute): Ignore internal calls without
16915 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16916 Richard Biener <rguenther@suse.de>
16919 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
16920 captures used multiple times, except for the last use.
16921 * generic-match-head.c: Include gimplify.h.
16923 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16925 PR tree-optimization/79390
16926 * target.h (struct noce_if_info): Declare.
16927 * targhooks.h (default_noce_conversion_profitable_p): Declare.
16928 * target.def (noce_conversion_profitable_p): New target hook.
16929 * ifcvt.h (struct noce_if_info): New type, moved from ...
16930 * ifcvt.c (struct noce_if_info): ... here.
16931 (noce_conversion_profitable_p): Renamed to ...
16932 (default_noce_conversion_profitable_p): ... this. No longer
16934 (noce_try_store_flag_constants, noce_try_addcc,
16935 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
16936 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
16937 instead of noce_conversion_profitable_p.
16938 * config/i386/i386.c: Include ifcvt.h.
16939 (ix86_option_override_internal): Don't override
16940 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
16941 (ix86_noce_conversion_profitable_p): New function.
16942 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
16943 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
16944 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
16945 * doc/tm.texi: Regenerated.
16947 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16949 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
16952 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16955 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
16956 instructions for small multiply cores.
16958 2017-04-04 Jeff Law <law@redhat.com>
16960 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
16962 (mips_expand_vec_perm_const): Initialize elements in orig_perm
16963 that are not set by the loop over the elements.
16965 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16968 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
16969 int mode, convert_modes it to mode as unsigned, otherwise use
16970 lowpart_subreg to mode rather than SImode.
16971 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
16972 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
16973 Use DImode instead of SImode for the shift count operand.
16974 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
16977 2017-04-04 Richard Biener <rguenther@suse.de>
16979 PR middle-end/80281
16980 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
16981 arithmetic done for the negate or the plus. Simplify.
16982 (A - (-B) -> A + B): Likewise.
16983 * fold-const.c (split_tree): Make sure to not negate pointers.
16985 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
16987 PR rtl-optimization/60818
16988 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
16989 a compare of comparisons with the thing compared if this results
16990 in a different machine mode.
16992 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
16994 * alias.c (base_alias_check): Fix typo in comment.
16995 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
16996 * cgraphunit.c (symbol_table::compile): Likewise.
16997 * collect2.c (maybe_run_lto_and_relink): Likewise.
16998 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
16999 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
17000 * config/avr/avr.c (avr_map_op_t): Likewise.
17001 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
17002 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
17003 * config/epiphany/epiphany.md (movcc): Likewise.
17004 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
17005 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
17007 * config/mips/mips.c (mips_save_restore_reg): Likewise.
17008 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
17009 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
17010 * config/sh/sh.c (sh_rtx_costs): Likewise.
17011 * fold-const.c (fold_truth_andor): Likewise.
17012 * genautomata.c (collapse_flag): Likewise.
17013 * gengtype.h (struct type::u::s): Likewise.
17014 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
17015 * input.c (FORMAT_AMOUNT): Likewise.
17016 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
17017 (known_aggs_to_agg_replacement_list): Likewise.
17018 * ipa-inline-analysis.c: Likewise.
17019 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
17020 * ipa-polymorphic-call.c
17021 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
17022 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
17023 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
17025 * modulo-sched.c (apply_reg_moves): Likewise.
17026 * omp-expand.c (build_omp_regions_1): Likewise.
17027 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
17028 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
17029 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
17030 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
17031 * value-prof.c: Likewise.
17032 * var-tracking.c (val_reset): Likewise.
17034 2017-04-03 Richard Biener <rguenther@suse.de>
17036 PR tree-optimization/80275
17037 * fold-const.c (split_address_to_core_and_offset): Handle
17040 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
17042 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
17043 descriptors is at least equal to that of functions.
17045 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
17047 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
17049 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
17052 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
17053 (mov<IMOD4:mode>): New expander.
17054 (*mov<IMOD4:mode>_internal): New insn and split pattern.
17056 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
17058 PR rtl-optimization/79405
17059 * fwprop.c (propagations_left): New variable.
17060 (forward_propagate_into): Decrement it.
17061 (fwprop_init): Initialize it.
17062 (fw_prop): If the variable has reached zero, stop propagating.
17063 (fwprop_addr): Ditto.
17065 2017-03-31 Jakub Jelinek <jakub@redhat.com>
17068 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
17069 a FUNCTION_DECL, pass it as decl instead of origin to
17072 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
17074 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
17077 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
17080 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
17081 TARGET_VSX_SMALL_INTEGER.
17083 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17085 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17086 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
17088 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
17090 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
17091 extraction from odd-numbered MSA register.
17093 2017-03-31 Jakub Jelinek <jakub@redhat.com>
17095 PR middle-end/80173
17096 * expmed.c (store_bit_field_1): Don't attempt to create
17097 a word subreg out of hard registers wider than word if they
17098 have HARD_REGNO_NREGS of 1 for their mode.
17100 PR middle-end/80163
17101 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
17102 conversions to integer types wider than word and pointer.
17105 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
17106 (rtx_equal_for_cselib_p): Pass 0 to it.
17107 * cselib.c (cselib_hasher::equal): Likewise.
17108 (rtx_equal_for_cselib_1): Add depth argument. If depth
17109 is 128, don't look up VALUE locs and punt. Increment
17110 depth in recursive calls when walking VALUE locs.
17112 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
17114 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
17115 (make_gcov_file_name): Use the canonical path name for generating
17117 (read_line): Fix handling of files with ascii null bytes.
17119 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
17121 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
17122 to initialise a vector register instead
17123 of using a const_int.
17125 2017-03-30 Jakub Jelinek <jakub@redhat.com>
17127 PR translation/80189
17128 * gimplify.c (omp_default_clause): Use %qs instead of %s in
17129 diagnostic messages.
17131 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
17134 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
17135 (dfp_diex_<mode>): Update mode of operand 1.
17136 * doc/extend.texi (dxex, dxexq): Document change to return type.
17137 (diex, diexq): Document change to argument type.
17139 2017-03-30 Martin Jambor <mjambor@suse.cz>
17142 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
17143 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
17144 it reflects the signature changes performed at the callee side.
17145 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
17146 to cgraph_build_function_type_skip_args.
17147 (build_function_decl_skip_args): Adjust call to the above function.
17149 2017-03-30 Jakub Jelinek <jakub@redhat.com>
17152 * config/i386/sse.md
17153 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
17154 register as dest whenever it is a MEM not rtx_equal_p to the
17155 corresponding dup operand, and when forcing into reg move the
17156 reg into the memory afterwards.
17157 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
17158 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
17159 for the force_reg mode.
17160 (avx512vl_vextractf128<mode>): Use register as dest either
17161 always when a MEM, or when it is a MEM not rtx_equal_p to the
17162 corresponding dup operand, or even not when it is a CONST_VECTOR
17163 depending on the mode and lo vs. hi.
17164 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
17166 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
17167 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
17168 Likewise. Require that operands[2] is even.
17169 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
17170 Remove extraneous parens. Require that operands[2] is a multiple
17172 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
17173 operands[0] is a MEM if <mask_applied>, the predicates/constraints
17174 disallow memory then.
17176 2017-03-30 Richard Biener <rguenther@suse.de>
17178 PR tree-optimization/77498
17179 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
17180 to non-constants over backedges.
17182 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
17184 PR rtl-optimization/80233
17185 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
17186 as last_combined_insn. Do not test for BARRIER_P separately.
17188 2017-03-29 Andreas Schwab <schwab@suse.de>
17191 * calls.c (prepare_call_address): Convert funexp to Pmode before
17192 copying to temp reg.
17194 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17196 PR tree-optimization/80158
17197 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
17198 Handle possible future case of more than one alternate
17200 (replace_rhs_if_not_dup): Likewise.
17201 (replace_one_candidate): Likewise.
17203 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
17205 PR rtl-optimization/80193
17206 * ira.c (ira): Do not check allocation for LRA.
17208 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
17210 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
17211 (nvptx_output_simt_exit): Declare.
17212 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
17213 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
17214 (init_softstack_frame): Move initialization of crtl->is_leaf to...
17215 (nvptx_declare_function_name): ...here. Emit declaration of local
17216 memory space buffer for omp_simt_enter insn.
17217 (nvptx_output_unisimt_switch): New.
17218 (nvptx_output_softstack_switch): New.
17219 (nvptx_output_simt_enter): New.
17220 (nvptx_output_simt_exit): New.
17221 * config/nvptx/nvptx.h (struct machine_function): New fields
17222 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
17223 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
17224 (UNSPECV_SIMT_EXIT): Ditto.
17225 (omp_simt_enter_insn): New insn.
17226 (omp_simt_enter): New expansion.
17227 (omp_simt_exit): New insn.
17228 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
17230 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
17231 (expand_GOMP_SIMT_ENTER_ALLOC): New.
17232 (expand_GOMP_SIMT_EXIT): New.
17233 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
17234 (GOMP_SIMT_ENTER_ALLOC): Ditto.
17235 (GOMP_SIMT_EXIT): Ditto.
17236 * target-insns.def (omp_simt_enter): New insn.
17237 (omp_simt_exit): Ditto.
17238 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
17240 (lower_rec_simd_input_clauses): Implement SIMT privatization.
17241 (lower_rec_input_clauses): Likewise.
17242 (lower_lastprivate_clauses): Handle SIMT privatization.
17244 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
17245 (ompdevlow_adjust_simt_enter): New.
17246 (find_simtpriv_var_op): New.
17247 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
17248 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
17250 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
17251 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
17252 (copy_decl_for_dup_finish): Ditto.
17254 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
17256 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
17259 * config/i386/i386.c (ix86_option_override_internal): Always
17260 allow -mpreferred-stack-boundary=3 for 64-bit targets.
17262 2017-03-28 Bin Cheng <bin.cheng@arm.com>
17264 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
17266 2017-03-28 Bin Cheng <bin.cheng@arm.com>
17268 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
17269 mark new edge's irreducible flag accordign to it.
17270 (vect_do_peeling): Check loop preheader edge's irreducible flag
17271 and pass it to function slpeel_add_loop_guard.
17273 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
17275 PR tree-optimization/80218
17276 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
17277 Update block frequencies and counts.
17279 2017-03-28 Richard Biener <rguenther@suse.de>
17281 PR tree-optimization/78644
17282 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
17283 of a simplification result we may not use it at all.
17285 2017-03-28 Richard Biener <rguenther@suse.de>
17288 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
17289 without arguments, generate default definition of a SSA name.
17291 2017-03-28 Richard Biener <rguenther@suse.de>
17293 PR middle-end/80222
17294 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
17295 TYPE_REF_CAN_ALIAS_ALL references.
17296 * fold-const.c (fold_indirect_ref_1): Likewise.
17298 2017-03-28 Martin Liska <mliska@suse.cz>
17301 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
17302 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
17304 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
17305 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
17307 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
17308 (EXTRA_SPECS): Define.
17309 (SUBTARGET_EXTRA_SPECS): Likewise.
17310 (SUBTARGET_CPP_SPEC): Likewise.
17311 * config/arc/elf.h (EXTRA_SPECS): Renamed to
17312 SUBTARGET_EXTRA_SPECS.
17313 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
17315 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
17317 * config/arc/simdext.md (vst64_insn): Update pattern.
17318 (vld32wh_insn): Likewise.
17319 (vld32wl_insn): Likewise.
17320 (vld64_insn): Likewise.
17321 (vld32_insn): Likewise.
17323 2017-03-28 Marek Polacek <polacek@redhat.com>
17326 * fold-const.c (fold_comparison): Use protected_set_expr_location
17327 instead of SET_EXPR_LOCATION.
17329 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
17331 * tree.c (add_expr): Avoid name lookup warning.
17333 2017-03-27 Jeff Law <law@redhat.com>
17335 PR tree-optimization/80216
17336 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
17337 function name. Limit recursion depth.
17338 (record_temporary_equivalences): Corresponding changes.
17340 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
17342 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
17345 2017-03-27 Jakub Jelinek <jakub@redhat.com>
17348 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
17350 * cfgcleanup.c (reg_note_cfa_p): New array.
17351 (insns_have_identical_cfa_notes): New function.
17352 (old_insns_match_p): Don't cross-jump in between /f
17353 and non-/f instructions. If both i1 and i2 are frame related,
17354 verify all CFA notes, their order and content.
17356 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
17359 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
17360 HImode and SImode with zero extend to DImode to one insn.
17361 (bswap<mode>2_extenddi): Likewise.
17362 (bswapsi2_extenddi): Likewise.
17363 (bswaphi2_extendsi): Likewise.
17364 (bswaphi2): Combine bswap HImode and SImode into one insn.
17365 Separate memory insns from swapping register.
17366 (bswapsi2): Likewise.
17367 (bswap<mode>2): Likewise.
17368 (bswaphi2_internal): Delete, no longer used.
17369 (bswapsi2_internal): Likewise.
17370 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
17371 store, and gpr<-gpr swap insns.
17372 (bswap<mode>2_store): Likewise.
17373 (bswaphi2_reg): Register only splitter, combine with the splitter.
17374 (bswaphi2 splitter): Likewise.
17375 (bswapsi2_reg): Likewise.
17376 (bswapsi2 splitter): Likewise.
17377 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
17378 the insns into load, store, and register/register insns.
17379 (bswapdi2_ldbrx): Likewise.
17380 (bswapdi2_load): Likewise.
17381 (bswapdi2_store): Likewise.
17382 (bswapdi2_reg): Likewise.
17384 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
17386 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
17387 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
17389 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
17392 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
17394 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
17395 special handling for target option conflicts between dform
17396 options (-mpower9-dform, -mpower9-dform-vector,
17397 -mpower9-dform-scalar) and -mno-direct-move.
17399 2017-03-27 Richard Biener <rguenther@suse.de>
17401 PR tree-optimization/80181
17402 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
17404 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
17406 * config/arc/predicates.md (move_double_src_operand): Replace the
17407 call to move_double_src_operand with a call to address_operand.
17409 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
17411 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
17412 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
17413 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
17415 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
17417 * config/arc/predicates.md (long_immediate_loadstore_operand):
17418 Consider scaled addresses cases.
17420 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
17422 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
17423 restored when in interrupt.
17424 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
17425 doesn't have delay slot.
17427 2017-03-27 Richard Biener <rguenther@suse.de>
17430 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
17431 inlined thunk clones.
17433 2017-03-27 Jakub Jelinek <jakub@redhat.com>
17436 * asan.c (instrument_derefs): Copy over last operand from
17437 original COMPONENT_REF to the new COMPONENT_REF with
17438 DECL_BIT_FIELD_REPRESENTATIVE.
17439 * ubsan.c (instrument_object_size): Likewise.
17441 2017-03-27 Richard Biener <rguenther@suse.de>
17443 PR tree-optimization/80170
17444 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
17445 sure DR/SCEV didnt fold in constants we do not see when looking
17446 at the reference base alignment.
17448 2017-03-27 Richard Biener <rguenther@suse.de>
17450 PR middle-end/80171
17451 * gimple-fold.c (fold_ctor_reference): Properly guard against
17452 NULL return value from canonicalize_constructor_val.
17454 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
17457 * config/i386/i386.c (ix86_expand_builtin)
17458 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
17459 flags reg setting and flags reg using instructions.
17460 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
17461 clobbering instructions to zero extend op2.
17463 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
17465 * doc/install.texi (Configuration) <--with-aix-soname>:
17466 Update link to AIX ld.
17468 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
17470 PR rtl-optimization/80160
17471 PR rtl-optimization/80159
17472 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
17473 reg_alternate_class into account.
17475 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
17478 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
17479 to consider in curr_insn_transform.
17481 2017-03-24 Jakub Jelinek <jakub@redhat.com>
17483 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
17484 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
17485 and emit_mode_inner.
17487 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17489 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
17490 argument to the overloaded builtin variants. Use the new flag to
17491 deprecate certain builtin variants.
17492 * config/s390/s390-builtin-types.def: Add new builtin types.
17493 * config/s390/s390-builtins.h: Support new flags field for
17494 overloaded builtins.
17495 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
17496 (s390_macro_to_expand): Enable vector float data type.
17497 (s390_cpu_cpp_builtins_internal): Indicate support of the new
17498 builtins by incrementing the __VEC__ version number.
17499 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
17501 (s390_resolve_overloaded_builtin): Emit error messages depending
17502 on the builtin flags.
17503 * config/s390/s390.c (s390_expand_builtin): Support additional
17504 flags argument. Change error message to match the messages
17505 emitted in s390-c.c.
17506 * config/s390/s390.md: New UNSPEC_* constants.
17507 (op_type): Add new instruction types.
17508 * config/s390/vecintrin.h: Add new builtins and test data class
17510 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
17511 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
17512 (VEC_INEXACT, VEC_NOINEXACT): New constants.
17513 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
17514 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
17515 ("vec_mergel<mode>"): V_HW -> VEC_HW.
17517 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
17518 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
17519 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
17520 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
17522 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
17523 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
17524 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
17525 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
17527 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
17528 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
17529 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
17530 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
17531 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
17532 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
17533 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
17535 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
17536 ("vec_scatter_element<V_HW_4:mode>_DI")
17537 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
17538 ("vec_fpint<mode>", "vflls")
17539 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
17540 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
17541 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
17542 ("*vec_cmphe<mode>_cc"): ... these.
17544 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
17545 mode constant instead of magic value.
17547 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17549 * config/s390/s390.c (s390_expand_vec_compare): Support other
17550 vector floating point modes than just V2DF.
17551 (s390_expand_vcond): Likewise.
17552 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
17553 (s390_cannot_change_mode_class): Prevent mode changes between TF
17554 and V1TF in vector registers.
17555 * config/s390/s390.md (DF, SF): New mode attributes.
17556 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
17557 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
17558 SFmode support for VRs.
17559 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
17561 (VFT, VF_HW): New mode iterators.
17562 (vw, sdx): New mode attributes.
17563 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
17564 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
17565 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
17566 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
17567 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
17568 also the new vector floating point modes. Renaming to ...
17570 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
17571 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
17572 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
17573 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
17574 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
17575 ("vec_unordered<mode>"): ... these.
17577 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
17578 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
17579 ("*vec_extendv2df"): New insn definitions.
17581 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17583 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
17584 ("mulditi3_2", "*muldi3_sign"): New patterns.
17585 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
17586 rename the pattern definition.
17588 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17590 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
17592 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
17594 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17596 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
17597 instruction if possible.
17598 * config/s390/vector.md (vec_halfnumelts): New mode
17600 ("*vec_vllezlf<mode>"): New pattern.
17602 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17604 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
17605 ("popcountv4si2", "popcountv2di2"): Rename to ...
17606 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
17607 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
17609 ("popcount<mode>2_vxe"): New pattern.
17611 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17613 * common/config/s390/s390-common.c (processor_flags_table): Add
17615 * config.gcc: Add arch12.
17616 * config/s390/driver-native.c (s390_host_detect_local_cpu):
17617 Default to arch12 for unknown CPU model numbers.
17618 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
17619 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
17620 PROCESSOR_max sanity check.
17621 * config/s390/s390-opts.h (enum processor_type): Add
17623 * config/s390/s390.c (processor_table): Add arch12.
17624 (s390_expand_builtin): Add check for B_VXE flag.
17625 (s390_issue_rate): Add PROCESSOR_ARCH12.
17626 (s390_get_sched_attrmask): Likewise.
17627 (s390_get_unit_mask): Likewise.
17628 (s390_sched_score): Enable z13 scheduling for arch12.
17629 (s390_sched_reorder): Likewise.
17630 (s390_sched_variable_issue): Likewise.
17631 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
17633 (s390_tune_attr): Use z13 scheduling also for arch12.
17634 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
17635 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
17636 (TARGET_VXE_P): New macros.
17637 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
17638 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
17639 * config/s390/s390.opt: Add arch12 as processor_type.
17641 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17643 * config/s390/s390.md
17644 ("fixuns_truncdddi2", "fixuns_trunctddi2")
17645 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
17646 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
17648 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
17649 Rename expanders to ...
17651 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
17652 ("fixuns_truncdddi2_emu"): ... these.
17654 ("fixuns_trunc<mode>si2_emu"): New expander.
17656 ("*fixuns_truncdfdi2_z13"): Rename to ...
17657 ("*fixuns_truncdfdi2_vx"): ... this.
17659 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17661 * config/s390/2964.md: Remove the single element vector compare
17662 instructions which are no longer used.
17663 * config/s390/s390.c (s390_select_ccmode): Remove handling of
17665 (s390_canonicalize_comparison): Remove handling of DFmode
17667 (s390_expand_vec_compare_scalar): Remove function.
17668 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
17669 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
17671 ("*cmp<mode>_ccs"): Add wfcdb instruction.
17673 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17675 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
17677 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
17678 will anyway by matched by mov<mode>_64dfp.
17680 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17682 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
17683 vlef/vstef. Add missing operand to vleif.
17685 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17687 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
17688 pair for all vector types with 64 bit elements.
17689 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
17690 * config/s390/vector.md (V_HW_64): ... here.
17691 (V_128_NOSINGLE): New mode iterator.
17692 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
17693 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
17694 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
17695 ("*vec_load_pairv2di"): Change to ...
17696 ("*vec_load_pair<mode>"): ... this one.
17698 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17700 * config/s390/constraints.md: Add comments.
17701 (jKK): Reject element sizes > 8 bytes.
17702 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
17704 * config/s390/s390.md: Add the s_operand checks formerly in
17705 s390_split_ok_p to various splitters where they are still
17707 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
17708 for 128 bit vectors. Plus two splitters.
17710 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17712 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
17715 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17718 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
17719 error if the boundary argument is not constant.
17721 2017-03-24 Jakub Jelinek <jakub@redhat.com>
17723 PR rtl-optimization/80112
17724 * loop-doloop.c (doloop_condition_get): Don't check condition
17725 if cmp isn't SET with IF_THEN_ELSE src.
17727 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17729 PR tree-optimization/80158
17730 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
17731 replacing a candidate statement, also replace it for the
17732 candidate's alternate interpretation.
17733 (replace_rhs_if_not_dup): Likewise.
17734 (replace_one_candidate): Likewise.
17736 2017-03-24 Richard Biener <rguenther@suse.de>
17738 PR tree-optimization/80167
17739 * graphite-isl-ast-to-gimple.c
17740 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
17742 (translate_isl_ast_to_gimple::get_rename): Likewise.
17744 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
17746 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
17747 handling of certain combinations of target options, including the
17748 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
17749 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
17751 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17754 * config/arm/arm.md (*load_multiple): Add reload_completed to
17755 matching condition.
17757 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17758 Richard Biener <rguenth@suse.de>
17760 PR tree-optimization/79908
17761 PR tree-optimization/80136
17762 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
17763 been cast away, gimplify_and_add suffices.
17765 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
17767 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
17769 2017-03-23 Richard Biener <rguenther@suse.de>
17771 PR tree-optimization/80032
17772 * gimplify.c (gimple_push_cleanup): Forced unconditional
17773 cleanups still have to go to the conditional_cleanups
17776 2017-03-22 Jakub Jelinek <jakub@redhat.com>
17778 PR tree-optimization/80072
17779 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
17781 (next_operand_entry_id): Change type to unsigned int.
17782 (sort_by_operand_rank): Make sure to return the right return value
17783 even if unsigned fields are bigger than INT_MAX.
17784 (struct oecount): Change cnt and id type to unsigned int.
17785 (oecount_hasher::equal): Formatting fix.
17786 (oecount_cmp): Make sure to return the right return value
17787 even if unsigned fields are bigger than INT_MAX.
17788 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
17791 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
17792 TREE_READONLY on result if writing it more than once.
17795 * doc/invoke.texi (-fsanitize=thread): Document that with
17796 -fnon-call-exceptions atomics are not able to throw
17800 * tsan.c: Include tree-eh.h.
17801 (instrument_builtin_call): Call maybe_clean_eh_stmt or
17802 maybe_clean_or_replace_eh_stmt where needed.
17803 (instrument_memory_accesses): Add cfg_changed argument.
17804 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
17805 if it returned true.
17806 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
17808 PR rtl-optimization/63191
17809 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
17810 wrapper function, moved the whole old content into ...
17811 (ix86_delegitimize_address_1): ... this. New inline function.
17812 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
17813 true as last argument instead of ix86_delegitimize_address.
17815 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
17817 * config/aarch64/aarch64.c (generic_branch_cost): Copy
17818 cortexa57_branch_cost.
17820 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
17822 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
17824 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17827 * doc/md.texi (Constraints): Document wA constraint.
17828 * config/rs6000/constraints.md (wA): New.
17829 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
17830 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
17831 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
17832 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
17834 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
17837 * gimplify.c (is_oacc_declared): New function.
17838 (oacc_default_clause): Use it to set default flags for acc declared
17839 variables inside parallel regions.
17840 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
17841 declared variables.
17842 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
17843 declare attribute to any decl as necessary.
17845 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
17848 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
17849 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
17850 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
17851 (arm_arch_lpae): This.
17852 * config/arm/arm.c (arm_arch7ve): Rename into ...
17853 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
17854 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
17857 2017-03-22 Martin Liska <mliska@suse.cz>
17860 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
17861 error message instead of an ICE.
17863 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17865 * doc/extend.texi (6.11 Additional Floating Types): Revise.
17867 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
17869 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
17871 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
17874 2017-03-21 Martin Sebor <msebor@redhat.com>
17876 * doc/extend.texi: Use "cannot" instead of "can't."
17877 * doc/hostconfig.texi: Same.
17878 * doc/install.texi: Same.
17879 * doc/invoke.texi: Same.
17880 * doc/loop.texi: Same.
17881 * doc/md.texi: Same.
17882 * doc/objc.texi: Same.
17883 * doc/rtl.texi: Same.
17884 * doc/tm.texi: Same.
17885 * doc/tm.texi.in: Same.
17886 * doc/trouble.texi: Same.
17888 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
17891 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
17892 (collect_checksum_attributes): Set it.
17893 (die_checksum_ordered): Use it.
17895 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17897 PR tree-optimization/79908
17898 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
17899 change: For a VA_ARG whose LHS has been cast away, use
17900 force_gimple_operand to construct the side effects.
17902 2017-03-21 David Malcolm <dmalcolm@redhat.com>
17904 PR translation/80001
17905 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
17906 more amenable to translation.
17907 (oacc_loop_auto_partitions): Likewise.
17909 2017-03-21 Marek Polacek <polacek@redhat.com>
17910 Martin Sebor <msebor@redhat.com>
17912 PR tree-optimization/80109
17913 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
17914 on INTEGRAL_TYPE_P.
17916 2017-03-21 Jakub Jelinek <jakub@redhat.com>
17917 Segher Boessenkool <segher@kernel.crashing.org>
17920 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
17921 check reg_used_between_p between insn and one of succ or succ2
17922 depending on if succ is artificial insn not inserted into insn
17925 2017-03-21 Martin Liska <mliska@suse.cz>
17927 PR gcov-profile/80081
17928 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
17929 * doc/gcc.texi: Include gcov-dump stuff.
17930 * doc/gcov-dump.texi: New file.
17932 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
17934 PR rtl-optimization/79150
17935 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
17936 conditional jump, if the jump is the last insn of the loop.
17938 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17939 Richard Biener <rguenth@suse.de>
17941 PR tree-optimization/79908
17942 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
17943 been cast away, use force_gimple_operand to construct the side
17946 2017-03-21 Martin Liska <mliska@suse.cz>
17948 PR libfortran/79956
17949 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
17952 2017-03-21 Brad Spengler <spender@grsecurity.net>
17955 * plugin.c (htab_hash_plugin): New function.
17956 (add_new_plugin): Use it and adjust.
17957 (parse_plugin_arg_opt): Adjust.
17958 (init_one_plugin): Likewise.
17960 2017-03-21 Richard Biener <rguenther@suse.de>
17962 PR tree-optimization/80032
17963 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
17964 if set force the cleanup to happen unconditionally.
17965 (gimplify_target_expr): Push inserted clobbers with force_uncond
17966 to avoid them being removed by control-dependent DCE.
17968 2017-03-21 Richard Biener <rguenther@suse.de>
17970 PR tree-optimization/80122
17971 * tree-inline.c (copy_bb): Do not expans va-arg packs or
17972 va_arg_pack_len when the inlined call stmt requires pack
17974 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
17976 2017-03-21 Jakub Jelinek <jakub@redhat.com>
17979 * tsan.c (instrument_builtin_call): If the memory model argument
17980 is not a constant, assume it is valid.
17983 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
17986 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
17988 PR rtl-optimization/79910
17989 * combine.c (can_combine_p): Do not allow combining an I0 or I1
17990 if its dest is used by an insn before I2 (other than the combined
17991 insns themselves, which are properly handled already).
17993 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
17996 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
17998 * combine.c (record_used_regs): New static function.
17999 (try_combine): Handle situations where there is an additional
18000 instruction between I2 and I3 which needs to have a LOG_LINK
18004 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
18006 * combine.c (try_combine): Delete redundant i1 test. Call
18007 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
18009 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18012 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
18013 alternatives 13/14.
18015 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18017 PR tree-optimization/80054
18018 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
18019 the optimization if a PHI or any of its arguments is not dominated
18020 by the candidate's basis. Use gphi* rather than gimple* as
18022 (replace_profitable_candidates): Clean up a gimple* variable that
18023 should be a gphi* variable.
18025 2017-03-20 Martin Sebor <msebor@redhat.com>
18028 * doc/extend.texi (attribute constructor): Document present limitation.
18030 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
18033 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
18034 __POWER9_VECTOR__ #ifdef control, change template definition to
18035 use Power9-specific built-in function.
18036 (vec_any_eq): Likewise.
18037 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
18038 to control outcomes from this test.
18039 (vector_ae_<mode>p): For VEC_F modes, likewise.
18041 2017-03-20 Ian Lance Taylor <iant@google.com>
18043 * config/i386/i386.c (ix86_function_regparm): Save an extra
18044 register for -fsplit-stack with DECL_STATIC_CHAIN.
18046 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
18049 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
18050 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
18052 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
18054 * config/riscv/riscv.c (riscv_print_operand): Use "fence
18056 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
18059 2017-03-20 Marek Polacek <polacek@redhat.com>
18062 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
18064 2017-03-20 Richard Biener <rguenther@suse.de>
18066 PR tree-optimization/80113
18067 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
18068 allocate extra SSA name for PHI def.
18069 (add_close_phis_to_outer_loops): Likewise.
18070 (add_close_phis_to_merge_points): Likewise.
18071 (copy_loop_close_phi_args): Likewise.
18072 (copy_cond_phi_nodes): Likewise.
18074 2017-03-20 Martin Liska <mliska@suse.cz>
18076 PR middle-end/79753
18077 * tree-chkp.c (chkp_build_returned_bound): Do not build
18078 returned bounds for a LHS that's not a BOUNDED_P type.
18080 2017-03-20 Martin Liska <mliska@suse.cz>
18084 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
18085 COMPLEX_CST and VECTOR_CST.
18087 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18090 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
18091 target operand. A new splitter adds the clobber statement in case
18092 the target operand is dead anyway.
18094 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
18096 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
18097 to age-old versions of binutils and glibc.
18099 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
18101 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
18103 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
18105 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
18107 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
18109 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
18110 requirement for binutils 2.13.
18112 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
18114 * combine.c (try_combine): Delete redundant i1 test. Call
18115 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
18117 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
18119 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
18120 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
18122 <riscv64-*-elf>: Re-arrange section
18123 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
18124 <riscv32-*-linux>: Likewise.
18125 <riscv64-*-elf>: Likewise
18126 <riscv64-*-linux>: Likewise.
18128 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
18131 * aarch64.opt(verbose-cost-dump): Fix typo.
18133 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
18136 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
18137 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
18139 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
18141 * reload.c (find_reloads): When reloading a nonoffsettable address,
18142 use RELOAD_OTHER for it and its address reloads.
18144 PR rtl-optimization/79910
18145 * combine.c (record_used_regs): New static function.
18146 (try_combine): Handle situations where there is an additional
18147 instruction between I2 and I3 which needs to have a LOG_LINK
18150 2017-03-17 Jeff Law <law@redhat.com>
18152 PR tree-optimization/71437
18153 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
18154 conditional in the hash table first.
18155 (vrp_dom_walker::before_dom_children): Extract condition from
18156 ASSERT_EXPR. Record condition, its inverion and any implied
18157 conditions as well.
18159 2017-03-17 Marek Polacek <polacek@redhat.com>
18160 Markus Trippelsdorf <markus@trippelsdorf.de>
18162 PR tree-optimization/80079
18163 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
18166 2017-03-17 Richard Biener <rguenther@suse.de>
18168 PR middle-end/80075
18169 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
18170 Properly verify the LHS before the RHS possibly claims to be
18172 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
18175 2017-03-17 Martin Jambor <mjambor@suse.cz>
18177 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
18178 (List of -O2 options): Likewise.
18179 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
18182 2017-03-17 Tom de Vries <tom@codesourcery.com>
18184 * gcov-dump.c (print_usage): Print bug_report_url.
18186 2017-03-17 Richard Biener <rguenther@suse.de>
18188 PR middle-end/80050
18189 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
18190 (parser::peek): Likewise.
18192 2017-03-17 Richard Biener <rguenther@suse.de>
18194 PR tree-optimization/80048
18195 * sese.c (free_sese_info): Properly release rename_map and
18196 copied_bb_map elements.
18198 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
18200 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
18201 Add linked-list forward and backlinks. Insert on
18202 construction, remove on destruction.
18203 (class pass_store_merging): Add m_stores_head field.
18204 (pass_store_merging::terminate_and_process_all_chains):
18205 Iterate over m_stores_head list.
18206 (pass_store_merging::terminate_all_aliasing_chains):
18208 (pass_store_merging::execute): Check for debug stmts first.
18209 Push new chains onto the m_stores_head stack.
18211 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
18214 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
18215 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
18216 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
18218 2017-03-16 Jeff Law <law@redhat.com>
18220 PR tree-optimization/71437
18221 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
18222 member function. Implementation moved into after_dom_children
18223 member function and into the threader's thread_outgoing_edges
18225 (dom_opt_dom_walker::after_dom_children): Simplify by moving
18226 some code into new thread_outgoing_edges.
18227 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
18228 definition. Simplify marker handling (do it here). Assume we always
18229 have the available expression and the const/copies tables.
18230 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
18232 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
18233 * tree-vrp.c (equiv_stack): No longer file scoped.
18234 (vrp_dom_walker): New class.
18235 (vrp_dom_walker::before_dom_children): New member function.
18236 (vrp_dom_walker::after_dom_children): Likewise.
18237 (identify_jump_threads): Setup domwalker. Use it rather than
18238 walking edges in a random order by hand. Simplify setup/finalization.
18239 (finalize_jump_threads): Remove.
18240 (vrp_finalize): Do not call identify_jump_threads here.
18241 (execute_vrp): Do it here instead and call thread_through_all_blocks
18244 PR tree-optimization/71437
18245 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
18247 (simplify_stmt_for_jump_threading): Add basic_block argument. All
18249 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
18250 (dom_opt_dom_walker::thread_across_edge): Remove
18251 handle_dominating_asserts argument. All callers changed.
18252 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
18253 changes. Remove calls to lhs_of_dominating_assert. Other
18254 uses of handle_dominating_asserts turn into unconditional code
18255 (simplify_control_stmt_condition_1): Likewise.
18256 (simplify_control_stmt_condition): Likewise.
18257 (thread_through_normal_block, thread_across_edge): Likewise.
18258 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
18259 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
18260 object if it is not an SSA_NAME.
18261 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
18262 before calling into the VRP specific simplifiers.
18263 (identify_jump_threads): Remove handle_dominating_asserts
18266 2017-03-16 Jakub Jelinek <jakub@redhat.com>
18269 * tree-diagnostic.c (default_tree_printer): No longer static.
18270 * tree-diagnostic.h (default_tree_printer): New prototype.
18272 2017-03-16 Tamar Christina <tamar.christina@arm.com>
18274 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
18275 Change ins into fmov.
18277 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18279 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
18280 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
18281 Use h_con constraint for operand 1.
18282 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
18283 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
18285 2017-03-15 Jeff Law <law@redhat.com>
18287 PR tree-optimization/71437
18288 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
18289 (record_temporary_equivalences): Use it.
18291 PR tree-optimization/71437
18292 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
18293 tree-ssa-scopedtables.
18294 (lookup_avail_expr, build_and_record_new_cond): Likewise.
18295 (record_conditions, record_cond, vuse_eq): Likewise.
18296 (record_edge_info): Adjust to API tweak of record_conditions.
18297 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
18298 (record_temporary_equivalences, optimize_stmt): Likewise.
18299 (eliminate_redundant_computations): Likewise.
18300 (record_equivalences_from_stmt): Likewise.
18301 * tree-ssa-scopedtables.c: Include options.h and params.h.
18302 (vuse_eq): New function, moved from tree-ssa-dom.c
18303 (build_and_record_new_cond): Likewise.
18304 (record_conditions): Likewise. Accept vector of conditions rather
18305 than edge_equivalence structure for first argument.
18306 for the first argument.
18307 (avail_exprs_stack::lookup_avail_expr): New member function, moved
18308 from tree-ssa-dom.c.
18309 (avail_exprs_stack::record_cond): Likewise.
18310 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
18311 from tree-ssa-dom.c.
18312 (avail_exprs_stack): Add new member functions lookup_avail_expr
18314 (record_conditions): Declare.
18316 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
18319 * lra-constraints.c (process_alt_operands): Increase reject for
18320 reloading an input/output operand.
18322 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
18325 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
18326 insns to convert from signed/unsigned char/short to IEEE 128-bit
18328 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
18330 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
18333 * config/i386/i386.c (ix86_vector_duplicate_value): Create
18334 subreg of inner mode for values already in registers.
18336 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
18338 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
18339 iteration reg is used after the loop.
18341 2017-03-14 Martin Sebor <msebor@redhat.com>
18343 PR tree-optimization/79800
18344 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
18345 precision in negative-positive range.
18346 (format_floating): Call non-const overload with adjusted precision.
18348 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
18351 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
18354 2017-03-14 Martin Sebor <msebor@redhat.com>
18356 PR middle-end/80020
18357 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
18358 * builtins.def (aligned_alloc): Use it.
18361 * Makefile.in (GTFILES): Add calls.c.
18362 * calls.c: Include "gt-calls.h".
18364 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
18366 PR rtl-optimization/79728
18367 * regs.h (struct target_regs): New field
18368 x_contains_allocatable_regs_of_mode.
18369 (contains_allocatable_regs_of_mode): New macro.
18370 * reginfo.c (init_reg_sets_1): Initialize it, and change
18371 contains_reg_of_mode so it includes global regs as well.
18372 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
18373 rather than contains_regs_of_mode.
18375 2017-03-14 Martin Liska <mliska@suse.cz>
18377 * doc/invoke.texi: Document options that can't be combined with
18378 -fcheck-pointer-bounds.
18380 2017-03-14 Martin Liska <mliska@suse.cz>
18382 PR middle-end/79831
18383 * doc/invoke.texi (-Wchkp): Document the option.
18385 2017-03-14 Martin Liska <mliska@suse.cz>
18387 * Makefile.in: Install gcov-dump.
18389 2017-03-14 Martin Liska <mliska@suse.cz>
18391 * multiple_target.c (expand_target_clones): Bail out for
18392 an invalid attribute.
18394 2017-03-14 Richard Biener <rguenther@suse.de>
18396 * alias.c (struct alias_set_entry): Pack properly.
18397 * cfgloop.h (struct loop): Likewise.
18398 * cse.c (struct set): Likewise.
18399 * ipa-utils.c (struct searchc_env): Likewise.
18400 * loop-invariant.c (struct invariant): Likewise.
18401 * lra-remat.c (struct cand): Likewise.
18402 * recog.c (struct change_t): Likewise.
18403 * rtl.h (struct address_info): Likewise.
18404 * symbol-summary.h (function_summary): Likewise.
18405 * tree-loop-distribution.c (struct partition): Likewise.
18406 * tree-object-size.c (struct object_size_info): Likewise.
18407 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
18408 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
18409 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
18410 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
18411 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
18412 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
18413 (struct _stmt_vec_info): Likewise.
18415 2017-03-14 Martin Liska <mliska@suse.cz>
18418 * multiple_target.c (create_dispatcher_calls): Check that
18419 a target can create a function dispatcher.
18421 2017-03-14 Martin Liska <mliska@suse.cz>
18424 * multiple_target.c (expand_target_clones): Drop local.local
18425 flag for default implementation.
18427 2017-03-14 Richard Biener <rguenther@suse.de>
18429 PR tree-optimization/80030
18430 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
18432 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
18434 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
18435 gcc_fallthrough() instead of __attribute__((fallthrough));
18437 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
18439 * doc/gcc.texi: Remove "up" link to (DIR).
18440 * doc/gccint.texi: Ditto.
18442 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
18444 * doc/install.texi (Specific) <avr>: Remove reference to
18447 2017-03-13 Jeff Law <law@redhat.com>
18449 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
18450 attribute rather than comments.
18452 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
18453 match_scratch operand is highest.
18455 2017-03-13 Martin Liska <mliska@suse.cz>
18457 PR middle-end/78339
18458 * ipa-pure-const.c (warn_function_noreturn): If the declarations
18459 is a CHKP clone, use original declaration.
18461 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18463 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
18464 (arc_conditional_register_usage): Use a different allocation order
18465 when optimizing for size.
18466 * common/config/arc/arc-common.c (arc_option_optimization_table):
18467 Section anchors default on when optimizing for size.
18469 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18471 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
18473 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18475 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
18476 * config/arc/arc.md (cpu_facility): Add cd variant.
18477 (*movqi_insn): Add code density variant.
18478 (*movhi_insn): Likewise.
18479 (*movqi_insn): Likewise.
18480 (*addsi3_mixed): Likewise.
18481 (subsi3_insn): Likewise.
18483 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18485 * config/arc/arc.md (movsi_cond_exec): Update constraint.
18487 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18489 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
18490 expressions with MINUS and UNARY ops.
18492 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18495 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
18497 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
18498 between vec_select and vector argument.
18499 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
18500 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
18501 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
18502 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
18503 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
18504 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
18506 2017-03-13 Richard Biener <rguenther@suse.de>
18509 * params.def (vect-max-peeling-for-alignment): Fix typo.
18511 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
18513 * doc/install.texi (Specific) <mips-*-*>: Remove description of
18514 issue that only occurred with binutils below 2.18.
18516 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
18518 * doc/install.texi (Specific) <cris-axis-elf>: No longer
18519 refer to binutils 2.11/2.12 minimum.
18521 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
18523 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
18524 ftp.kernel.org and simplify binutils requirement.
18526 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
18528 * doc/invoke.texi (Warning Options): Fix spelling of link-time
18530 (Optimize Options): Ditto. Also remove redundancy.
18532 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18534 PR translation/79848
18535 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
18537 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
18538 to G_ to avoid double translation.
18540 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18542 PR translation/79923
18543 * auto-profile.c (get_combined_location): Convert leading
18544 character of diagnostics to lower case and remove trailing period.
18545 (read_profile): Likewise for various diagnostics.
18546 * config/arm/arm.c (arm_option_override): Remove trailing period
18547 from various diagnostics.
18548 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
18549 (msp430_expand_delay_cycles): Likewise.
18551 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18554 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
18555 full command-line argument, rather than just "str".
18556 (aarch64_validate_march): Likewise.
18557 (aarch64_validate_mtune): Likewise.
18559 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
18561 PR rtl-optimization/78911
18562 * lra-assigns.c (must_not_spill_p): New function.
18563 (spill_for): Use it.
18565 2017-03-10 Jakub Jelinek <jakub@redhat.com>
18567 PR tree-optimization/79981
18568 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
18569 ATOMIC_COMPARE_EXCHANGE ifn result.
18570 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
18571 IFN_ATOMIC_COMPARE_EXCHANGE.
18573 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18576 * opts.c (parse_sanitizer_options): Add missing question mark to
18577 "did you mean" message.
18579 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18581 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
18583 (VMULEUH_UNS): Likewise.
18584 (VMULOUB_UNS): Likewise.
18585 (VMULOUH_UNS): Likewise.
18586 * config/rs6000/rs6000.c (builtin_function_type): Remove
18587 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
18589 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18592 * read-rtl-function.c (function_reader::read_rtx_operand): Update
18593 x with result of extra_parsing_for_operand_code_0.
18594 (function_reader::extra_parsing_for_operand_code_0): Convert
18595 return type from void to rtx, returning x. When reading
18596 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
18597 larger size containing struct block_symbol.
18599 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
18601 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
18602 -mfloat128-hardware without -m64.
18604 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
18607 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
18608 entries to the case statement that marks unsigned arguments to
18609 overloaded functions.
18611 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
18613 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
18614 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
18616 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
18619 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
18620 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
18622 2017-03-10 Martin Liska <mliska@suse.cz>
18626 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
18627 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
18630 2017-03-10 Olivier Hainque <hainque@adacore.com>
18632 * tree-switch-conversion (array_value_type): Start by resetting
18633 candidate type to it's main variant.
18635 2017-03-10 Jakub Jelinek <jakub@redhat.com>
18637 PR rtl-optimization/79909
18638 * combine.c (try_combine): Use simplify_replace_rtx on individual
18639 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
18640 of the whole CALL_INSN_FUNCTION_USAGE.
18642 PR tree-optimization/79972
18643 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
18644 get_range_info on SSA_NAMEs. Formatting fixes.
18646 2017-03-10 Richard Biener <rguenther@suse.de>
18647 Jakub Jelinek <jakub@redhat.com>
18649 PR tree-optimization/77975
18650 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
18651 edge to be constant.
18652 (get_val_for): For constant x return it. Formatting fix.
18653 (loop_niter_by_eval): Avoid pointless looping if the next iteration
18654 would use the same bases as the current one.
18656 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18658 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
18659 instead of vec_select for V1TImode.
18660 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
18662 (VSX_LE_128): Add V1TI to this mode iterator.
18663 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
18664 (*vsx_le_perm_store_<mode>): Likewise.
18665 (pre-reload splitter for VSX stores): Likewise.
18666 (post-reload splitter for VSX stores): Likewise.
18667 (*vsx_xxpermdi2_le_<mode>): Likewise.
18668 (*vsx_lxvd2x2_le_<mode>): Likewise.
18669 (*vsx_stxvd2x2_le_<mode>): Likewise.
18671 2017-03-09 Michael Eager <eager@eagercon.com>
18673 Correct failures with --enable-checking=yes,rtl.
18675 * config/microblaze/microblaze.c (microblaze_expand_shift):
18676 Replace GET_CODE test with CONST_INT_P and INTVAL test with
18677 test for const0_rtx.
18678 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
18679 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
18681 2017-03-09 Richard Biener <rguenther@suse.de>
18683 PR tree-optimization/79977
18684 * graphite-scop-detection.c (scop_detection::merge_sese):
18685 Handle the case of extra exits to blocks dominating the entry.
18687 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
18689 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
18692 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
18694 PR rtl-optimization/79949
18695 * lra-constraints.c (process_alt_operands): Check memory when
18696 trying to predict a cycle. Print about the overall increase.
18698 2017-03-09 Richard Biener <rguenther@suse.de>
18700 PR middle-end/79971
18701 * gimple-expr.c (useless_type_conversion_p): Preserve
18702 TYPE_SATURATING for fixed-point types.
18704 2017-03-09 Richard Biener <rguenther@suse.de>
18707 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
18708 alignment of BLKmode params.
18710 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18713 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
18714 (VALL_NO_V2Q): Likewise.
18716 * config/aarch64/aarch64-simd.md
18717 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
18719 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
18720 VALL_NO_V2Q mode iterator.
18721 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
18723 2017-03-09 Martin Liska <mliska@suse.cz>
18725 PR tree-optimization/79631
18726 * tree-chkp-opt.c (chkp_is_constant_addr): Call
18727 tree_int_cst_sign_bit just for INTEGER constants.
18729 2017-03-09 Martin Liska <mliska@suse.cz>
18733 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
18736 2017-03-09 Marek Polacek <polacek@redhat.com>
18739 * tree.c (inchash::add_expr): Handle TREE_VEC.
18741 2017-03-09 Martin Liska <mliska@suse.cz>
18744 (chkp_narrow_size_and_offset): New function.
18745 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
18746 (void chkp_parse_bit_field_ref): New function.
18747 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
18748 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
18750 2017-03-09 Martin Liska <mliska@suse.cz>
18753 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
18754 (chkp_find_bounds_1): Remove gcc_unreachable.
18756 2017-03-09 Jakub Jelinek <jakub@redhat.com>
18759 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
18760 BUILT_IN_SYNC*, determine the access type from the size suffix and
18761 always build a MEM_REF with that type. Handle forgotten
18762 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
18765 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
18766 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
18767 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
18768 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
18769 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
18770 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
18771 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
18772 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
18773 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
18774 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
18775 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
18776 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
18777 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
18778 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
18779 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
18780 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
18781 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
18782 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
18783 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
18784 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
18785 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
18786 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
18787 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
18788 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
18789 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
18790 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
18791 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
18792 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
18793 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
18794 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
18795 definitions outside of __OPTIMIZE__ guarded section.
18798 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
18799 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
18800 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
18801 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
18804 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18806 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
18807 ("vfenez<mode>"): Add missing constraints.
18809 2017-03-08 Martin Sebor <msebor@redhat.com>
18812 * config/nds32/nds32.c (nds32_option_override):
18813 Fix misspelled diagnostic.
18815 2017-03-08 Jakub Jelinek <jakub@redhat.com>
18818 * gimplify.c (gimplify_omp_for): Replace index var in outer
18819 taskloop statement with an artificial variable and add
18820 OMP_CLAUSE_PRIVATE clause for it.
18822 2017-03-08 Richard Biener <rguenther@suse.de>
18824 PR tree-optimization/79955
18825 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
18826 for accesses that are completely outside of the variable.
18828 2017-03-08 Andrew Haley <aph@redhat.com>
18830 PR tree-optimization/79943
18831 * tree-ssa-loop-split.c (compute_new_first_bound): When
18832 calculating the new upper bound, (END-BEG) should be added, not
18835 2017-03-08 Jakub Jelinek <jakub@redhat.com>
18837 * config/avr/avr.md (setmemhi): Make sure match_dup
18838 operand number comes before match_scratch.
18840 2017-03-08 Richard Biener <rguenther@suse.de>
18842 PR tree-optimization/79920
18843 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
18844 with ncopies == 1 to ...
18845 (vect_transform_slp_perm_load): ... here. Properly compute
18846 all element loads by iterating VF times over the group. Do
18847 not handle ncopies (computed in a broken way) in
18848 vect_create_mask_and_perm.
18850 2017-03-08 Jakub Jelinek <jakub@redhat.com>
18853 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
18854 is a uniform vector, use uniform_vector_p return value instead of
18855 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
18857 2017-03-07 Marek Polacek <polacek@redhat.com>
18859 PR middle-end/79809
18860 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
18861 (alloca_call_type): Likewise.
18863 2017-03-07 Martin Liska <mliska@suse.cz>
18865 * gcov.c (process_args): Put comment to correct location.
18867 2017-03-07 Martin Liska <mliska@suse.cz>
18869 PR middle-end/68270
18870 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
18871 Use array_at_struct_end_p instead of DECL_CHAIN (field).
18872 (chkp_narrow_bounds_for_field): Likewise.
18873 (chkp_parse_array_and_component_ref): Pass one more argument to
18876 2017-03-07 Richard Biener <rguenther@suse.de>
18878 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
18881 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
18883 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
18884 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
18886 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18889 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
18890 to end of description.
18891 (PARAM_MAX_STORES_TO_MERGE): Likewise.
18893 2017-03-07 Jakub Jelinek <jakub@redhat.com>
18895 PR rtl-optimization/79901
18896 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
18898 (*avx512f_<code><mode>3<mask_name>): ... this.
18899 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
18900 iterator instead of VI8_AVX2_AVX512BW.
18902 PR rtl-optimization/79901
18903 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
18904 min/max expander, expand it using expand_vec_cond_expr.
18907 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
18910 2017-03-06 Jakub Jelinek <jakub@redhat.com>
18913 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
18914 to void * for PCH reasons.
18915 * dwarf2out.c (output_loc_operands, output_die): Cast
18916 v.val_vec.array to unsigned char *.
18918 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
18921 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
18924 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
18926 PR rtl-optimization/79571
18927 * lra-constraints.c (process_alt_operands): Calculate static
18928 reject and subtract it from overall when only addresses will be
18931 2017-03-06 Julia Koval <julia.koval@intel.com>
18934 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
18935 incoming stack boundary to 128 for 64-bit targets.
18937 2017-03-06 Richard Biener <rguenther@suse.de>
18939 PR tree-optimization/79894
18940 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
18941 to NULL after folding it.
18943 2017-03-06 Richard Biener <rguenther@suse.de>
18945 PR tree-optimization/79824
18946 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
18947 check disabling peeling for gaps.
18949 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
18951 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
18952 attributes): Document gettimeofday.
18954 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
18956 * config/s390/s390.c (s390_option_override_internal): Set
18957 PARAM_MIN_VECT_LOOP_BOUND
18959 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
18961 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
18962 * config/s390/s390.md: Likewise.
18964 2017-03-06 Jakub Jelinek <jakub@redhat.com>
18967 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
18968 (<avx2_avx512>_perm<mode>): Rename to ...
18969 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
18971 (<avx512>_perm<mode>_mask): Rename to ...
18972 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
18974 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
18975 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
18976 instead of VI8F_256_512.
18977 (avx512f_perm<mode>): New define_expand.
18978 (avx512f_perm<mode>_mask): Likewise.
18979 (avx512f_perm<mode>_1<mask_name>): New define_insn.
18980 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
18982 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18984 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
18985 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
18987 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
18989 2017-03-06 Martin Liska <mliska@suse.cz>
18992 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
18993 when having a SSA NAME w/o VAR_DECL assigned to it.
18995 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18997 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
18998 msa_dpsub_<su>_d): Fix MODE for vec_select.
19000 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
19002 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
19004 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
19006 2017-03-06 Richard Biener <rguenther@suse.de>
19008 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
19009 * plugin.c (register_plugin_info): Likewise.
19010 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
19012 2017-03-05 Jakub Jelinek <jakub@redhat.com>
19014 * config/i386/sse.md (sse_storehps, sse_storelps,
19015 avx_<castmode><avxsizesuffix>_<castmode>,
19016 avx512f_<castmode><avxsizesuffix>_<castmode>,
19017 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
19018 in condition that at least one operand is not a MEM.
19020 2017-03-03 Jakub Jelinek <jakub@redhat.com>
19022 PR middle-end/79805
19023 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
19024 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
19026 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
19027 gimple_call_nothrow_p flag based on whether original builtin can throw.
19028 If it can, emit following stmts on the fallthrough edge.
19029 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
19030 don't create new bb if inserting just debug stmts on the edge, try to
19031 insert them on the fallthru bb or just reset debug stmts.
19033 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
19036 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
19037 restore recog_data (including the operand rtxes inside it) around
19038 the call to get_insn_template.
19040 2017-03-03 Martin Sebor <msebor@redhat.com>
19042 PR tree-optimization/79699
19043 * context.c (context::~context): Free MPFR caches to avoid
19044 a memory leak on program exit.
19046 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19048 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
19049 Use wide_int::ulow () instead of .elt (0).
19051 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
19053 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
19054 (*pushxf): Limit oF constraint to 32bit targets and add oC
19055 constraint for 64bit targets.
19056 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
19057 (*pushdf): Change rmF constraint to rmC.
19059 2017-03-03 Martin Liska <mliska@suse.cz>
19061 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
19062 Remove unused variable.
19064 2017-03-03 Jakub Jelinek <jakub@redhat.com>
19067 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
19068 is a memory operand, increase num_memory.
19069 (ix86_expand_args_builtin): Likewise.
19071 2017-03-03 Jan Hubicka <jh@suse.cz>
19074 * ipa-devirt.c (maybe_record_node): Properly handle
19075 __cxa_pure_virtual visibility.
19077 2017-03-03 Martin Liska <mliska@suse.cz>
19079 PR tree-optimization/79803
19080 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
19082 (pass_loop_prefetch::execute): Disabled optimization if an
19083 assumption about L1 cache size is not met.
19085 2017-03-03 Martin Liska <mliska@suse.cz>
19087 PR rtl-optimization/79574
19088 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
19089 (hash_scan_set): Likewise.
19090 (dump_hash_table): Likewise.
19091 (hoist_code): Likewise.
19093 2017-03-03 Richard Biener <rguenther@suse.de>
19095 * fixed-value.c (fixed_from_string): Restore use of elt (1)
19096 in place of uhigh ().
19097 (fixed_convert_from_real): Likewise.
19099 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
19102 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
19104 2017-03-03 Richard Biener <rguenther@suse.de>
19106 PR middle-end/79818
19107 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
19108 TYPE_OVERFLOW_UNDEFINED check.
19110 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19112 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
19114 (vector_ae_<mode>_p): Likewise.
19115 (vector_nez_<mode>_p): Likewise.
19116 (vector_ne_v2di_p): Likewise.
19117 (vector_ae_v2di_p): Likewise.
19118 (vector_ne_<mode>_p): Likewise.
19119 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
19121 (vsx_tsqrt<mode>2_fe): Likewise.
19123 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
19126 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
19128 2017-03-02 Jakub Jelinek <jakub@redhat.com>
19130 PR rtl-optimization/79780
19131 * cprop.c (one_cprop_pass): When second and further conditional trap
19132 in a single basic block is turned into an unconditional trap, turn it
19133 into a deleted note to avoid RTL verification failures.
19135 2017-03-02 Richard Biener <rguenther@suse.de>
19137 * fold-const.c (const_binop): Use ulow () instead of elt (0).
19139 2017-03-02 Richard Biener <rguenther@suse.de>
19141 PR tree-optimization/79345
19143 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
19144 param and abort the walk, returning -1 if it is hit.
19145 (walk_aliased_vdefs): Take a limit param and pass it on.
19146 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
19147 defaulting to 0 and return a signed int.
19148 * tree-ssa-uninit.c (struct check_defs_data): New struct.
19149 (check_defs): New helper.
19150 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
19151 about uninitialized memory.
19152 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
19153 bogus uninitialized warning.
19154 (fixed_convert_from_real): Likewise.
19156 2017-03-02 Bin Cheng <bin.cheng@arm.com>
19158 PR tree-optimization/66768
19159 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
19160 iv_use if base object can't be determined.
19162 2017-03-02 Jakub Jelinek <jakub@redhat.com>
19164 PR tree-optimization/79345
19165 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
19166 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
19167 (get_pattern_stats): Initialize it.
19168 * genemit.c (gen_expand): Verify match_scratch numbers come after
19169 match_operand/match_dup numbers.
19170 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
19171 match_scratch numbers.
19172 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
19174 * config/s390/s390.md (trunctdsd2): Likewise.
19176 2017-03-02 Richard Biener <rguenther@suse.de>
19178 * wide-int.h (wide_int_storage::operator=): Implement in terms
19181 2017-03-02 Richard Biener <rguenther@suse.de>
19183 PR tree-optimization/79777
19184 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
19185 the to insert expression to sth existing.
19187 2017-03-01 Martin Sebor <msebor@redhat.com>
19189 PR middle-end/79692
19190 * gimple-ssa-sprintf.c
19191 (directive::known_width_and_precision): New function.
19192 (format_integer): Use it.
19193 (get_mpfr_format_length): Consider the full range of precision
19194 when computing %g output with the # flag. Set the likely byte
19195 count to 3 rather than 1 when precision is indeterminate.
19196 (format_floating): Correct the lower bound of precision.
19198 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19200 * doc/invoke.texi: Document default code model for 64-bit Linux.
19202 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19205 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
19206 udiv rather than div since input pattern is unsigned.
19208 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
19210 * config/i386/i386.c (print_reg): Warn for values of
19211 unsupported size in integer register.
19213 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
19216 * config/rs6000/predicates.md (current_file_function_operand): Do
19217 not allow self calls to be local if the function is replaceable.
19219 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
19222 * config/rs6000/altivec.h (vec_ctz and others): Change the
19223 preprocessor macro that controls conditional compilation from
19224 _ARCH_PWR9 to __POWER9_VECTOR__.
19225 (vec_all_ne): Change parameterization of __altivec_scalar_pred
19226 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
19227 control (instead of _ARCH_PWR9 control) so that template
19228 definition uses power9-specific function.
19229 (vec_any_eq): Likewise.
19230 (vec_all_ne): Change macro definition to use a power9-specific
19231 expansion under #ifdef __POWER9_VECTOR__ control (instead of
19232 _ARCH_PWR9 control).
19233 (vec_any_eq) Likewise.
19234 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
19235 expansion for CMPNEF to remove support for xvcmpnesp instruction.
19236 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
19237 support for xvcmpnedp instruction.
19238 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
19239 macro expansion so that Power9 implementation of vec_all_ne does
19240 not use the AltiVec predicate framework.
19241 (VCMPNEH_P): Likewise.
19242 (VCMPNEW_P): Likewise.
19243 (VCMPNED_P): Likewise.
19244 (VCMPNEFP_P): Likewise.
19245 (VCMPNEDP_P): Likewise.
19246 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
19247 implementation of vec_any_eq to not use AltiVec predicate
19249 (VCMPAEH_P): Likewise.
19250 (VCMPAEW_P): Likewise.
19251 (VCMPAED_P): Likewise.
19252 (VCMPAEFP_P): Likewise.
19253 (VCMPAEDP_P): Likewise.
19254 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
19255 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
19256 not use the AltiVec predicate framework.
19257 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
19258 of vec_any_eq to not use AltiVec predicate framework.
19259 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
19260 support for predefined __POWER9_VECTOR__ macro to indicate that
19261 Power9 instruction selection is enabled.
19262 (altivec_overloaded_builtins): Remove extraneous
19263 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
19264 function argument types RS6000_BTI_bool_V16QI and
19265 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
19266 entry for overloaded function argument types RS6000_BTI_bool_V4SI
19267 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
19268 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
19269 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
19270 Power9 for implementations of vec_cmpne. Change the signature for
19271 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
19272 (representing vec_all_ne) to remove the previously described first
19273 argument of type RS6000_BTI_INTSI, as this was an artifact of
19274 reliance on the AltiVec predicate framework, which is no longer
19275 used in the implementation of these functions. Add
19276 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
19277 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
19278 since, unlike the AltiVec predicate framework implementation, we
19279 do not share function descriptors between vec_alle and vec_anyeq.
19280 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
19281 set of modes that receive special treatment even when
19282 TARGET_P9_VECTOR is true. The special treatment emits code that
19283 does not depend on Power9 instructions.
19284 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
19285 define_expand to not rely on AltiVec predicate framework.
19286 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
19288 (vector_ne_v2di_p): Change this define_expand to not rely on
19289 AltiVec predicate framework.
19290 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
19292 (vector_ne_<mode>_p): Change this define_expand to not rely on
19293 AltiVec predicate framework.
19294 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
19296 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
19297 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
19298 define_insn pattern.
19299 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
19300 define_insn pattern because the xvcmpne<VSs>. instruction is not
19302 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
19303 instruction is not supported.
19305 2017-03-01 Jakub Jelinek <jakub@redhat.com>
19307 * config/nvptx/nvptx.c: Include intl.h.
19309 2017-03-01 Martin Jambor <mjambor@suse.cz>
19312 * ipa-prop.h (ipa_bits): Removed field known.
19313 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
19314 to pointers. Adjusted their comments to warn about their sharing.
19315 (ipcp_transformation_summary): Change bits to a vector of pointers.
19316 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
19317 (ipa_get_ipa_bits_for_value): Declare.
19318 * tree-vrp.h (value_range): Mark as GTY((for_user)).
19319 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
19320 (ipa_bits_hash_table): Likewise.
19321 (ipa_vr_ggc_hash_traits): Likewise.
19322 (ipa_vr_hash_table): Likewise.
19323 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
19324 being pointers and vr_known being removed.
19325 (ipa_set_jf_unknown): Likewise.
19326 (ipa_get_ipa_bits_for_value): New function.
19327 (ipa_set_jfunc_bits): Likewise.
19328 (ipa_get_value_range): New overloaded functions.
19329 (ipa_set_jfunc_vr): Likewise.
19330 (ipa_compute_jump_functions_for_edge): Use the above functions to
19331 construct bits and vr parts of jump functions.
19332 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
19333 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
19335 (ipcp_grow_transformations_if_necessary): Also allocate
19336 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
19338 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
19339 them. Fix too long lines.
19340 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
19341 vr_known being removed.
19342 (ipa_read_jump_function): Use new setter functions to construct bits
19343 and vr parts of jump functions or set them to NULL.
19344 (write_ipcp_transformation_info): Adjust for bits being pointers.
19345 (read_ipcp_transformation_info): Likewise.
19346 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
19348 Include gt-ipa-prop.h.
19349 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
19351 (ipcp_store_bits_results): Likewise.
19352 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
19353 Do not write to existing jump functions but use a temporary instead.
19355 2017-03-01 Jakub Jelinek <jakub@redhat.com>
19358 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
19359 attempt to use its first operand as BIT_FIELD_REF base.
19361 2017-03-01 Richard Biener <rguenther@suse.de>
19363 PR middle-end/79721
19364 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
19365 interpolating formula in wrapping arithmetic.
19366 (chrec_apply): Convert chrec_evaluate return value to wanted type.
19368 2017-03-01 Jakub Jelinek <jakub@redhat.com>
19370 PR tree-optimization/79734
19371 * tree-vect-generic.c (expand_vector_condition): Optimize
19372 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
19373 Handle VEC_COND_EXPR where comparison has different inner width from
19374 type's inner width.
19376 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
19378 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
19379 markup, and similar issues. Remove @opindex entries for things
19380 that aren't options. Add missing -mmpy-option entries.
19382 2017-02-28 Jakub Jelinek <jakub@redhat.com>
19384 PR tree-optimization/79737
19385 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
19386 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
19387 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
19388 instead of byte_size. Formatting fix.
19389 (shift_bytes_in_array_right): Formatting fix.
19391 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
19394 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
19395 condition on optimize for the leaf function test.
19397 2017-02-28 Martin Liska <mliska@suse.cz>
19400 * read-rtl-function.c (function_reader::handle_unknown_directive):
19401 Bail out when one uses -flto.
19403 2017-02-28 Martin Liska <mliska@suse.cz>
19405 * common.opt: Replace space with tabular for options of <number>
19407 * config/i386/i386.opt: Show <number> value for
19408 -mlarge-data-threshold.
19409 * opts.c (print_filtered_help): Do not display number in hexadecimal
19412 2017-02-28 Martin Liska <mliska@suse.cz>
19414 * common.opt: Fix --help=option -Q for options which are of
19417 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
19419 * config/i386/i386.c (print_reg): Error out for values
19420 of 8-bit size in invalid integer register.
19422 2017-02-28 Martin Sebor <msebor@redhat.com>
19424 PR tree-optimization/79691
19425 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
19427 2017-02-28 Jakub Jelinek <jakub@redhat.com>
19430 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
19431 gcc_unreachable with output_operand_lossage.
19433 2017-02-28 Richard Biener <rguenther@suse.de>
19435 PR tree-optimization/79740
19436 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
19438 (visit_nary_op): Insert the nary into the hashtable if we
19439 pattern-matched sth.
19440 * tree-ssa-pre.c (eliminate_insert): Robustify.
19442 2017-02-28 Richard Biener <rguenther@suse.de>
19444 PR middle-end/79731
19445 * fold-const.c (decode_field_reference): Reject out-of-bound
19448 2017-02-28 Jakub Jelinek <jakub@redhat.com>
19450 * config/i386/i386.c: Include intl.h.
19451 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
19452 instead of just cond ? "..." : "...".
19453 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
19454 * coverage.c (read_counts_file): Likewise.
19455 * omp-offload.c: Include intl.h.
19456 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
19457 of just cond ? "..." : "...".
19458 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
19459 of just cond ? "..." : "...".
19461 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
19464 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
19466 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
19467 'tune for' CPU name.
19468 * config/arm/arm-cpu-data.h: Regenerated.
19470 2017-02-28 Richard Biener <rguenther@suse.de>
19472 PR tree-optimization/79732
19473 * tree-inline.c (expand_call_inline): Do not shadow var.
19475 2017-02-28 Richard Biener <rguenther@suse.de>
19477 PR tree-optimization/79723
19478 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
19479 address-space properly.
19481 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
19483 * doc/optinfo.texi (Optimization groups): Fix option used for
19485 * doc/invoke.texi (-fopt-info): Document "omp".
19486 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
19487 (OPTGROUP_ALL): Add OPTGROUP_OMP.
19488 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
19489 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
19490 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
19492 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
19494 * dumpfile.c (optgroup_options): Instead of "openmp", associate
19495 OPTGROUP_OMP with "omp".
19497 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
19500 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
19501 for arithmetic shift of unsigned V2DI.
19503 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
19505 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
19506 arc/linux.h headers.
19507 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
19508 (LINK_SPEC): Likewise.
19509 (ARC_TLS_EXTRA_START_SPEC): Likewise.
19510 (EXTRA_SPECS): Likewise.
19511 (STARTFILE_SPEC): Likewise.
19512 (ENDFILE_SPEC): Likewise.
19513 (LIB_SPEC): Likewise.
19514 (TARGET_SDATA_DEFAULT): Likewise.
19515 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
19516 (MULTILIB_DEFAULTS): Likewise.
19517 (DWARF2_UNWIND_INFO): Likewise.
19518 * config/arc/big.h: New file.
19519 * config/arc/elf.h: Likewise.
19520 * config/arc/linux.h: Likewise.
19521 * config/arc/t-uClibc: Remove.
19523 2017-02-27 Bin Cheng <bin.cheng@arm.com>
19525 PR tree-optimization/77536
19526 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
19527 (tree_transform_and_unroll_loop): Use above function to compute the
19528 estimated niter of unrolled loop and use it when scaling profile.
19529 Also use count info rather than frequency if it's non-zero.
19530 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
19531 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
19532 (vect_transform_loop): Call above function.
19534 2017-02-27 Richard Biener <rguenther@suse.de>
19536 PR tree-optimization/45397
19537 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
19538 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
19539 (visit_nary_op): Add pattern matching for CSEing sign-changed
19540 or truncated operations with wider ones.
19542 2017-02-27 Richard Biener <rguenther@suse.de>
19544 PR tree-optimization/79690
19545 * tree-vect-stmts.c (vectorizable_store): Use vector type
19546 built from the DR with address-space.
19548 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
19550 * doc/invoke.texi (Optimize Options): Refine the description
19551 of asan-use-after-return.
19553 2017-02-25 Alan Modra <amodra@gmail.com>
19555 PR rtl-optimization/79584
19556 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
19557 base, not ad->base_term, the reg within base. Remove assertion
19558 that ad->base == ad->base_term. Replace gen_int_mode using
19559 bogus mode with const0_rtx.
19561 2017-02-25 Jakub Jelinek <jakub@redhat.com>
19563 PR middle-end/79396
19564 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
19565 FMA_EXPR like tcc_binary or tcc_unary.
19567 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
19570 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
19572 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
19573 (output_loc_operands): Handle DW_OP_call_ref and
19574 DW_OP_GNU_variable_value.
19575 (struct variable_value_struct): New type.
19576 (struct variable_value_hasher): Likewise.
19577 (variable_value_hash): New variable.
19578 (string_types): Remove.
19579 (copy_loc_descr): New function.
19580 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
19581 (prepend_loc_descr_to_each): New function.
19582 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
19583 instead of add_loc_descr_to_each if the first argument is single
19584 location list and the second has multiple.
19585 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
19586 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
19587 when looking for variable value which doesn't have other location info.
19588 (loc_list_from_tree): Formatting fix.
19589 (gen_array_type_die): Simplify DW_AT_string_length handling.
19590 (adjust_string_types): Remove.
19591 (gen_subprogram_die): Don't call adjust_string_types nor test/set
19592 string_types. Call resolve_variable_values.
19593 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
19594 (resolve_addr_in_expr): Likewise. Add A argument.
19595 (copy_deref_exprloc): Remove deref argument. Adjust for the
19596 original expression being DW_OP_GNU_variable_value with optionally
19597 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
19598 optionally after it.
19599 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
19600 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
19601 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
19602 (variable_value_hasher::hash, variable_value_hasher::equal): New
19604 (resolve_variable_value_in_expr, resolve_variable_value,
19605 resolve_variable_values, note_variable_value_in_expr,
19606 note_variable_value): New functions.
19607 (dwarf2out_early_finish): Call note_variable_value on all toplevel
19610 2017-02-24 Jakub Jelinek <jakub@redhat.com>
19613 * opts.h (handle_generated_option): Add GENERATED_P argument.
19614 * opts-common.c (handle_option): Adjust function comment.
19615 (handle_generated_option): Add GENERATED_P argument, pass it to
19617 (control_warning_option): Pass false to handle_generated_option
19619 * opts.c (maybe_default_option): Pass true to handle_generated_option
19621 * optc-gen.awk: Likewise.
19623 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19625 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
19626 a REG, look at the REG it is a SUBREG of.
19627 (splitter for cmpeqsi_t): Ditto.
19629 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19631 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
19632 the special USEs with the pattern of the insn, not the insn itself.
19634 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
19637 * doc/invoke.texi: Document -mload-store-pairs.
19639 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19640 Sandra Loosemore <sandra@codesourcery.com>
19642 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
19643 argument isn't a CONST_INT.
19644 (nios2_alternate_compare_const): Assert op is a CONST_INT.
19645 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
19646 (nios2_validate_compare): Bypass alternate compare logic if *op2
19647 is not a CONST_INT.
19648 (ldstwm_operation_p): Return false if first_base is not a REG or
19649 if first_offset is not a CONST_INT.
19651 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19653 * config/cris/cris.md: Use correct operand in a define_peephole2.
19655 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19657 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
19659 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19661 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
19662 this_insn if it is an INSN or JUMP_INSN.
19663 (force_offsettable): Look at base, not at addr.
19664 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
19665 on things that aren't necessarily CONST_INTs.
19667 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
19669 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
19670 -mfpmath=sse is the default also for x86-32 targets with SSE2
19671 instruction set when @option{-ffast-math} is enabled
19673 2017-02-24 Jeff Law <law@redhat.com>
19675 PR rtl-optimizatoin/79286
19676 * ira.c (update_equiv_regs): Drop may_trap_p exception to
19679 2017-02-24 Richard Biener <rguenther@suse.de>
19681 PR tree-optimization/79389
19682 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
19685 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
19687 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
19688 function comment to reflect reality.
19689 (loop_exits_before_overflow): Fix typo in function description.
19691 2017-02-24 Richard Biener <rguenther@suse.de>
19693 PR tree-optimization/79389
19694 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
19695 properly that a threading opportunity exists. Detect conditional
19696 copy/constant propagation opportunities.
19698 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
19700 * config/visium/visium.md (type): Add trap.
19701 (b): New mode attribute.
19702 (*btst): Rename into...
19703 (*btst<mode>): ...this and adjust.
19704 (*cbranchsi4_btst_insn): Rename into...
19705 (*cbranch<mode>4_btst_insn): ...this and adjust.
19706 (trap): New define_insn.
19708 2017-02-23 Jakub Jelinek <jakub@redhat.com>
19710 PR tree-optimization/79389
19711 * ifcvt.c (struct noce_if_info): Add rev_cond field.
19712 (noce_reversed_cond_code): New function.
19713 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
19714 reversed_comparison_code. Formatting fix.
19715 (noce_try_store_flag): Test rev_cond != NULL in addition to
19716 reversed_comparison_code.
19717 (noce_try_store_flag_constants): Likewise.
19718 (noce_try_store_flag_mask): Likewise.
19719 (noce_try_addcc): Use rev_cond if non-NULL instead of
19720 reversed_comparison_code.
19721 (noce_try_cmove_arith): Likewise. Formatting fixes.
19722 (noce_try_minmax, noce_try_abs): Clear rev_cond.
19723 (noce_find_if_block): Initialize rev_cond.
19724 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
19725 instead of false as last argument never attempt to reverse it
19728 2017-02-23 Bin Cheng <bin.cheng@arm.com>
19730 PR tree-optimization/79663
19731 * tree-predcom.c (combine_chains): Process refs in reverse order
19732 only for ZERO length chains, and add explaining comment.
19734 2017-02-23 Jeff Law <law@redhat.com>
19736 PR tree-optimization/79578
19737 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
19738 in call to operand_equal_p.
19740 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
19743 * config/i386/cpuid.h: Fix another undefined behavior.
19745 2017-02-23 Richard Biener <rguenther@suse.de>
19747 PR tree-optimization/79683
19748 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
19749 vector types for data-refs.
19751 2017-02-23 Martin Liska <mliska@suse.cz>
19753 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
19755 2017-02-23 Jakub Jelinek <jakub@redhat.com>
19757 PR middle-end/79665
19758 * internal-fn.c (get_range_pos_neg): Moved to ...
19759 * tree.c (get_range_pos_neg): ... here. No longer static.
19760 * tree.h (get_range_pos_neg): New prototype.
19761 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
19762 are known to be in between 0 and signed maximum inclusive, try to
19763 expand both unsigned and signed divmod and use the cheaper one from
19766 2017-02-22 Jeff Law <law@redhat.com>
19768 PR tree-optimization/79578
19769 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
19770 to compare base operands.
19772 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
19775 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
19776 gpc_reg_operand instead of fpr_reg_operand.
19778 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
19780 * config/mips/mips.c (mips_return_in_memory): Force FP
19781 vector types to be returned in memory for o32 ABI.
19783 2017-02-22 Jakub Jelinek <jakub@redhat.com>
19785 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
19786 instead of DW_TAG_member for static data member declarations and don't
19787 set no_linkage_name for static inline data members.
19788 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
19791 2017-02-22 Martin Liska <mliska@suse.cz>
19793 * doc/invoke.texi: Replace inequality signs with square brackets
19796 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19798 PR tree-optimization/68644
19799 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
19801 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
19804 * lra-constraints.c (simplify_operand_subreg): Handle
19805 WORD_REGISTER_OPERATIONS targets.
19807 2017-02-22 Jakub Jelinek <jakub@redhat.com>
19810 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
19811 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
19812 elimination by swapping fld*.
19814 2017-02-22 Richard Biener <rguenther@suse.de>
19816 PR tree-optimization/79673
19817 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
19818 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
19819 irrelevant address-space qualifiers and avoiding a
19820 ADDR_SPACE_CONVERT_EXPR from fold_convert.
19822 2017-02-22 Richard Biener <rguenther@suse.de>
19824 PR tree-optimization/79666
19825 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
19826 to not symbolically negate if that may introduce undefined
19829 2017-02-22 Martin Liska <mliska@suse.cz>
19832 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
19833 * data-streamer-out.c (streamer_write_gcov_count_stream):
19835 * value-prof.c (stream_out_histogram_value): Make assert more
19836 precise based on type of counter.
19838 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
19841 * config/i386/i386.md (standard_x87sse_constant_load splitter):
19842 Use nonimmediate_operand instead of memory_operand for operand 1.
19843 (float-extend standard_x87sse_constant_load splitter): Ditto.
19845 2017-02-21 Jeff Law <law@redhat.com>
19847 PR tree-optimization/79621
19848 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
19849 blocks with edges to themselves.
19851 2017-02-21 Jakub Jelinek <jakub@redhat.com>
19854 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
19855 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
19856 Use gimple_call_builtin_p.
19859 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
19860 on temporarily removed DEBUG_INSNs.
19862 PR tree-optimization/79649
19863 * tree-loop-distribution.c (classify_partition): Give up on
19864 non-generic address space loads/stores.
19866 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
19868 * doc/loop.texi (Loop manipulation): Remove nonexistent
19869 tree_ssa_loop_version from the documentation.
19870 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
19872 2017-02-21 Jakub Jelinek <jakub@redhat.com>
19875 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
19876 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
19877 * config/rs6000/rs6000.c: Include except.h.
19878 (rs6000_expand_split_stack_prologue): Call
19879 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
19881 2017-02-21 Martin Jambor <mjambor@suse.cz>
19884 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
19885 have been analyzed.
19887 2017-02-21 Martin Jambor <mjambor@suse.cz>
19889 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
19890 for backward compatibility only.
19891 * doc/invoke.texi (Option Summary): Remove all references to
19892 -fipa-cp-alignment.
19894 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
19898 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19900 * lra-constraints.c (curr_insn_transform): Handle
19901 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
19903 2017-02-21 Martin Liska <mliska@suse.cz>
19905 * config/i386/i386.opt: Replace -masm-dialect with -masm.
19907 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
19909 PR translation/79638
19910 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
19912 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
19915 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
19916 (arm_function_ok_for_sibcall): Return false for an indirect call by
19917 descriptor if all the argument registers are used.
19918 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
19919 alignment of the function.
19921 2017-02-21 Jakub Jelinek <jakub@redhat.com>
19923 PR tree-optimization/61441
19924 * simplify-rtx.c (simplify_const_unary_operation): For
19925 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
19926 the sNaN unmodified.
19928 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
19930 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
19931 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
19932 instead of SYSTEM_HEADER_DIR.
19934 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
19935 Martin Liška <mliska@suse.cz>
19937 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
19938 Fix typos and grammar, use active voice, and clarify.
19940 2017-02-20 Marek Polacek <polacek@redhat.com>
19942 PR middle-end/79537
19943 * gimplify.c (gimplify_expr): Handle unused *&&L;.
19946 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
19948 2017-02-20 Jakub Jelinek <jakub@redhat.com>
19951 * config/i386/i386.c (ix86_expand_builtin): Handle
19952 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
19953 ix86_builtins_isa[fcode].isa as a requirement of those
19954 flags and any other flag in the bitmask.
19955 (ix86_init_mmx_sse_builtins): Use 0 instead of
19956 ~OPTION_MASK_ISA_64BIT as mask.
19957 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
19958 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
19959 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
19960 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
19962 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19965 * lra-constraints.c (split_reg): Check requested split mode
19966 is supported by the register.
19968 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19970 * lra-constraints.c (simplify_operand_subreg): Remove early
19973 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19976 * lra-constraints.c (curr_insn_transform): Tighten condition
19977 for converting SUBREG reloads from OP_OUT to OP_INOUT.
19979 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19982 * lra-constraints.c (curr_insn_transform): Handle
19983 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
19985 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
19988 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
19990 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
19992 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
19995 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
19998 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20000 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
20003 2017-02-18 Jakub Jelinek <jakub@redhat.com>
20005 * final.c (last_columnnum, override_columnnum): New variables.
20006 (final_start_function): Set last_columnnum, pass it to begin_prologue
20007 hook and pass 0 to dwarf2out_begin_prologue.
20008 (final_scan_insn): Update override_columnnum. Pass last_columnnum
20009 to source_line debug hook.
20010 (notice_source_line): Compute last_columnnum and for debug_column_info
20011 return true on column changes.
20012 * debug.h (struct gcc_debug_hooks): Add column argument to
20013 source_line and begin_prologue hooks.
20014 (debug_nothing_int_charstar_int_bool): Remove prototype.
20015 (debug_nothing_int_int_charstar,
20016 debug_nothing_int_int_charstar_int_bool): New prototypes.
20017 (dwarf2out_begin_prologue): Add column argument.
20018 * debug.c (do_nothing_debug_hooks): Adjust source_line and
20019 begin_prologue hooks.
20020 (debug_nothing_int_charstar_int_bool): Remove.
20021 (debug_nothing_int_int_charstar,
20022 debug_nothing_int_int_charstar_int_bool): New functions.
20023 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
20024 through to dwarf2out_source_line.
20025 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
20026 (dwarf2out_source_line): Add column argument, emit it if requested.
20027 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
20029 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
20030 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
20031 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
20032 through to dwarf2out_begin_prologue.
20033 (vmsdbgout_source_line): Add column argument, pass it through to
20034 dwarf2out_source_line.
20035 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
20036 dbxout_source_line caller.
20037 (dbxout_source_line): Add column argument.
20039 * common.opt (gno-column-info, gcolumn-info): New options.
20040 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
20041 (check_die): Also test for multiple DW_AT_decl_column attributes.
20042 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
20043 DW_AT_decl_column if requested.
20044 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
20046 (gen_variable_die): Likewise.
20047 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
20048 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
20051 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
20052 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
20053 (ix86_handle_option): Handle OPT_m3dnowa.
20054 * doc/invoke.texi (-m3dnowa): Document.
20055 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
20056 -m3dnowa instead of -m3dnow -march=athlon.
20059 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
20060 instead of gcc_assert for K, r and R code checks. Formatting fixes.
20062 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20065 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
20066 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
20067 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
20068 generator for vsx_xxpermdi_<mode>_be.
20069 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
20070 force big-endian semantics.
20071 (vsx_xxpermdi_<mode>_be): New define_expand with same
20072 implementation as previous version of vsx_xxpermdi_<mode>.
20074 2017-02-17 Jakub Jelinek <jakub@redhat.com>
20076 PR tree-optimization/79327
20077 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
20078 variable, its initialization and use.
20080 2017-02-17 Julia Koval <julia.koval@intel.com>
20082 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
20083 (OPTION_MASK_ISA_PKU_UNSET): New.
20084 (ix86_handle_option): Handle -mrdpid.
20085 * config/i386/cpuid.h (bit_RDPID): New.
20086 * config/i386/driver-i386.c (host_detect_local_cpu):
20087 Detect RDPID feature.
20088 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
20089 * config/i386/i386-c.c (ix86_target_macros_internal):
20091 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
20092 (ix86_valid_target_attribute_inner_p): Add "rdpid".
20093 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
20094 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
20095 * config/i386/i386.md (define_insn "rdpid"): New.
20096 * config/i386/i386.opt Add -mrdpid.
20097 * config/i386/immintrin.h (_rdpid_u32): New.
20099 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
20101 PR rtl-optimization/79541
20102 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
20103 instead of transforming it into USE.
20105 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
20107 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
20108 If HONOR_SNANS (SFmode) force the input to a register.
20109 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
20110 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
20111 an frsp or similar insn.
20113 2017-02-17 Martin Liska <mliska@suse.cz>
20115 PR rtl-optimization/79577
20116 * params.def (selsched-max-sched-times): Increase minimum to 1.
20118 2017-02-17 Martin Liska <mliska@suse.cz>
20120 PR rtl-optimization/79574
20121 * gcse.c (want_to_gcse_p): Prevent integer overflow.
20123 2017-02-17 Martin Liska <mliska@suse.cz>
20125 PR tree-optimization/79529
20126 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
20127 ssa_defined_default_def_p to handle cases which are implicitly
20129 * tree-ssa.c (ssa_defined_default_def_p): New function.
20130 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
20131 which are implicitly defined.
20132 * tree-ssa.h (ssa_defined_default_def_p): Declare.
20134 2017-02-17 Richard Biener <rguenther@suse.de>
20136 PR middle-end/79576
20137 * params.def (max-ssa-name-query-depth): Limit to 10.
20139 2017-02-17 Richard Biener <rguenther@suse.de>
20141 PR tree-optimization/79552
20142 * tree-ssa-structalias.c (visit_loadstore): Properly verify
20145 2017-02-17 Richard Biener <rguenther@suse.de>
20148 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
20150 2017-02-17 Marek Polacek <polacek@redhat.com>
20152 PR middle-end/79536
20153 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
20154 (fold_negate_expr): New wrapper.
20156 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
20158 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
20159 Correct terminology and de-emphasize pre-standard behavior.
20161 2017-02-16 Alan Modra <amodra@gmail.com>
20163 PR rtl-optimization/79286
20164 * ira.c (def_dominates_uses): New function.
20165 (update_equiv_regs): Don't create an equivalence for insns that
20166 may trap where the register def does not dominate the use.
20168 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
20170 PR rtl-optimization/78127
20171 * lra.c (lra): Call lra_eliminate before finish the loop after
20174 2017-02-16 Richard Biener <rguenther@suse.de>
20176 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
20178 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
20179 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
20180 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
20181 (isl_val_int_from_wi): New function.
20182 (extract_affine_gmp): Rename to ...
20183 (extract_affine_wi): ... this, take a widest_int.
20184 (extract_affine_int): Just wrap extract_affine_wi.
20185 (add_param_constraints): Use isl_val_int_from_wi.
20186 (add_loop_constraints): Likewise, and extract_affine_wi.
20188 2017-02-15 Jeff Law <law@redhat.com>
20190 PR middle-end/79521
20191 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
20192 ira_init_register_move_cost_if_necessary.
20194 2017-02-15 Martin Sebor <msebor@redhat.com>
20196 PR middle-end/32003
20197 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
20198 removed in a prior commit.
20200 2017-02-15 Bin Cheng <bin.cheng@arm.com>
20202 PR tree-optimization/79347
20203 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
20204 counters during peeling.
20206 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
20208 * Makefile.in (site.exp): Remove "set ISLVER".
20210 2017-02-15 Jakub Jelinek <jakub@redhat.com>
20213 * real.c (real_from_integer): Call real_convert even for decimal.
20215 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
20218 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
20220 2017-02-14 Andrew Pinski <apinski@cavium.com>
20222 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
20223 cores and change the partno/implementer to be correct.
20224 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
20225 the 'B" as the implementer.
20226 * config/aarch64/aarch64-tune.md: Regenerate.
20228 2017-02-14 Carl Love <cel@us.ibm.com>
20230 * config/rs6000/rs6000.c: Add case statement entry to make the
20231 xvcvuxdsp built-in argument unsigned.
20232 * config/rs6000/vsx.md: Fix the source and return operand types so they
20233 match the instruction definitions from the ISA document. Fix typo
20234 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
20237 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
20240 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
20241 member early_clobber_alts.
20242 * lra-lives.c (reg_early_clobber_p): New.
20243 (process_bb_lives): Use it.
20244 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
20245 (debug_operand_data): Initialize early_clobber_alts.
20246 (setup_operand_alternative): Set up early_clobber_alts.
20247 (collect_non_operand_hard_regs): Ditto. Pass early clobber
20248 alternatives to new_insn_reg.
20249 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
20251 (lra_update_insn_regno_info): Pass the new arg.
20253 2017-02-14 Jakub Jelinek <jakub@redhat.com>
20255 PR middle-end/79505
20256 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
20257 (new_oacc_loop_raw): Don't clear already cleared fields.
20260 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
20261 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
20262 _mm512_prefetch_i64gather_ps): New inline functions and macros.
20264 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
20267 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
20269 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
20272 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
20273 the extra instruction to the right place to store 128-bit constant
20276 2017-02-14 Martin Sebor <msebor@redhat.com>
20278 PR middle-end/79448
20279 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
20280 warning for strings of unknown length.
20282 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
20284 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
20286 2017-02-14 Jeff Law <law@redhat.com>
20289 * ira-costs.c (scan_one_insn): Initialize register move costs
20290 for pseudos seen in USE/CLOBBER insns.
20292 PR tree-optimization/79095
20293 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
20294 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
20295 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
20296 if the operands are known to be not equal, then the resulting range
20298 (intersect_ranges): If the new range is ~[0,0] and the old range is
20299 wide, then prefer ~[0,0].
20300 * tree-vrp.c (overflow_comparison_p_1): New function.
20301 (overflow_comparison_p): New function.
20302 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
20303 if NAME is used in an overflow test.
20304 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
20305 overflow check that can be expressed as an equality test, then adjust
20306 ops to be that equality test.
20308 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20310 * config/s390/s390-builtin-types.def: Remove flags argument.
20311 * config/s390/s390.c (s390_init_builtins): Likewise.
20313 2017-02-14 Martin Liska <mliska@suse.cz>
20315 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
20316 vector. Fix trailing white spaces.
20318 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
20320 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
20323 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20325 PR rtl-optimization/68664
20326 * config/arm/arm.c (arm_sched_can_speculate_insn):
20327 New function. Declare prototype.
20328 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
20330 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20332 PR rtl-optimization/68664
20333 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
20335 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
20337 2017-02-14 Amit Pawar <amit.pawar@amd.com>
20339 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
20340 max skip bytes for function, loop and jump.
20342 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20344 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
20345 ABS_EXPR for gimple dump.
20347 2017-02-14 Jakub Jelinek <jakub@redhat.com>
20350 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
20352 PR tree-optimization/79408
20353 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
20354 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
20355 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
20356 also if rhs1 is INTEGER_CST.
20358 2017-02-14 Richard Biener <rguenther@suse.de>
20360 PR middle-end/79432
20361 * tree-into-ssa.c (insert_phi_nodes): When the function can
20362 have abnormal edges rewrite SSA names with broken use-def
20363 dominance out of SSA and register them for PHI insertion.
20365 2017-02-13 Martin Sebor <msebor@redhat.com>
20367 PR middle-end/79496
20368 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
20369 clearing info.nowrite flag when snprintf size argument is a range.
20371 2017-02-13 Jakub Jelinek <jakub@redhat.com>
20373 * cprop.c (cprop_jump): Add missing space in string literal.
20374 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
20375 (get_constraint_for_component_ref): Likewise.
20376 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
20377 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
20378 * lra-constraints.c (process_alt_operands): Likewise.
20379 * ipa-inline.c (inline_small_functions): Likewise.
20380 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
20381 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
20382 * trans-mem.c (diagnose_tm_1_op): Likewise.
20383 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
20384 (grid_parallel_clauses_gridifiable): Likewise.
20386 * config/nvptx/mkoffload.c (process): Add space in between
20389 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
20390 "MOD4_SSE_REGS" and "ALL_REGS".
20392 * spellcheck.c (test_data): Add , in between "foo" and "food".
20394 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20397 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
20398 boundary crossing check and subsequent code generation agree.
20400 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20402 * config/aarch64/aarch64.c (has_memory_op): Delete.
20403 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
20406 2017-02-13 Jakub Jelinek <jakub@redhat.com>
20408 PR rtl-optimization/79388
20409 PR rtl-optimization/79450
20410 * combine.c (distribute_notes): When removing TEM_INSN for which
20411 corresponding dest has last value recorded, invalidate that last
20414 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20416 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
20417 of explicit '@'. Add missing assembly comment marker on branch costs
20420 2017-02-13 Nathan Sidwell <nathan@acm.org>
20422 * gengtype-lex.l (<in_struct>): Add '/'.
20424 2017-02-13 Martin Liska <mliska@suse.cz>
20427 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
20429 2017-02-13 Richard Biener <rguenther@suse.de>
20431 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
20433 * configure: Re-generate.
20434 * config.in: Likewise.
20435 * graphite-dependences.c: Simplify as if
20436 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
20437 * graphite-isl-ast-to-gimple.c: Likewise.
20438 * graphite-optimize-isl.c: Likewise.
20439 * graphite-poly.c: Likewise.
20440 * graphite-sese-to-poly.c: Likewise.
20441 * graphite.h: Likewise.
20442 * toplev.c: Include isl/version.h and use isl_version () for
20443 printing the ISL version.
20444 * doc/install.texi: Update ISL requirement.
20446 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
20448 * doc/standards.texi (Standards): Update reference to
20451 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
20453 * doc/extend.texi (Named Address Spaces): sourceware.org now
20455 * doc/install.texi (Binaries): Ditto.
20458 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
20460 * doc/cpp.texi: Replace "stringify"/"stringification" with C
20461 standard terminology "stringize"/"stringizing" throughout.
20462 * doc/cppinternals.texi: Likewise.
20464 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
20466 * doc/extend.texi: Fix some spelling mistakes and typos.
20467 * doc/invoke.texi: Likewise.
20469 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
20472 * params.def (inline-min-speedup) Change from 10 to 8.
20474 2017-02-11 Jakub Jelinek <jakub@redhat.com>
20476 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
20479 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
20482 * ipa-inline-analysis.c (get_minimal_bb): New function.
20483 (record_modified): Use it.
20484 (remap_edge_change_prob): Handle also ancestor functions.
20486 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
20488 * doc/contrib.texi (Contributors): Remove broken link into
20489 the Mauve CVS repository.
20491 2017-02-11 Jakub Jelinek <jakub@redhat.com>
20493 PR middle-end/79454
20494 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
20495 result computation whenever lhs doesn't have vector mode, not
20496 just when it has BLKmode.
20498 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
20500 * doc/makefile.texi (profiledbootstrap): Refer to the
20501 installation instructions only in textual form.
20503 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20506 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
20508 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
20510 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
20511 (Specific): Update mingw-w64 reference.
20513 (Specific): Remove broken link to Renesas RX processor.
20515 2017-02-10 Richard Biener <rguenther@suse.de>
20517 * toplev.c (process_options): Do not mention obsolete graphite
20518 options when printing sorry message about missing graphite support.
20519 Mention -floop-nest-optimize.
20521 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
20523 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
20524 (vtst_p16): Likewise.
20525 (vtstq_p8): Likewise.
20526 (vtstq_p16): Likewise.
20528 (vtstq_p64): Likewise.
20529 * config/arm/arm_neon.h (vgetq_lane_p64): New.
20530 (vset_lane_p64): New.
20531 (vsetq_lane_p64): New.
20533 2017-02-10 Jakub Jelinek <jakub@redhat.com>
20535 PR tree-optimization/79411
20536 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
20537 stmt operands are SSA_NAMEs used in abnormal phis.
20538 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
20541 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
20544 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
20547 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
20549 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
20551 2017-02-09 Jakub Jelinek <jakub@redhat.com>
20553 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
20557 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
20558 not arbitrary TREE_CONSTANT.
20561 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
20562 "omp declare target link" attribute unless is_global_var.
20563 * omp-offload.c (find_link_var_op): Likewise.
20565 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
20566 Chung-Lin Tang <cltang@codesourcery.com>
20568 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
20570 (gimplify_adjust_omp_clauses): Don't delete TILE.
20571 (gimplify_omp_for): Deal with TILE.
20572 * internal-fn.c (expand_GOACC_TILE): New function.
20573 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
20575 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
20576 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
20578 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
20579 avoid DIV for outermost collapse var.
20580 (expand_oacc_for): Insert tile element loop as needed. Adjust.
20581 Remove out of date comments, fix whitespace.
20582 * omp-general.c (omp_extract_for_data): Deal with tiling.
20583 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
20584 adjust OLF_DIM_BASE value.
20585 (struct omp_for_data): Add tiling field.
20586 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
20587 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
20588 for auto loops. Remove default auto determining, moved to
20589 oacc_loop_fixed_partitions.
20590 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
20591 stmts, add e_mask field.
20592 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
20593 (oacc_thread_numbers): Use oacc_dim_call.
20594 (oacc_xform_tile): New.
20595 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
20596 (finish_oacc_loop): Adjust for ifns vector.
20597 (oacc_loop_discover_walk): Append loop abstraction sites to list,
20598 add case for GOACC_TILE fns.
20599 (oacc_loop_xform_loop): Delete.
20600 (oacc_loop_process): Iterate over call list directly, and add
20601 handling for GOACC_TILE fns.
20602 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
20604 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
20605 vector partitioning to outer loops. Assign 2 partitions to loops
20606 when available. Add TILE handling.
20607 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
20608 (execite_oacc_device_lower): Process GOACC_TILE fns,
20609 ignore unknown specs.
20610 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
20611 * tree.c (omp_clause_num_ops): Adjust TILE ops.
20612 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
20614 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
20616 * configure.ac (ACX_BUGURL): Update.
20617 * configure: Regenerate.
20619 2017-02-09 Richard Biener <rguenther@suse.de>
20621 PR tree-optimization/69823
20622 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
20623 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
20625 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
20627 * config/arc/arc-c.def: Add __NPS400__ definition.
20628 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
20629 (TARGET_NPS400): Define.
20631 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
20633 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
20635 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
20636 pointer, arch_info.
20637 (arc_cpu_types): Fill the arch_info field with a pointer into the
20638 arc_arch_types table.
20639 (arc_selected_cpu): Declare.
20640 * config/arc/arc.c (arc_selected_cpu): Make global.
20641 (arc_selected_arch): Delete.
20642 (arc_base_cpu): Delete.
20643 (arc_override_options): Remove references to deleted variables,
20644 update access to arch information.
20645 (ARC_OPT): Update access to arch information.
20646 (ARC_OPTX): Likewise.
20647 * config/arc/arc.h (arc_base_cpu): Remove declaration.
20648 (TARGET_ARC600): Update access to arch information.
20649 (TARGET_ARC601): Likewise.
20650 (TARGET_ARC700): Likewise.
20651 (TARGET_EM): Likewise.
20652 (TARGET_HS): Likewise.
20653 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
20656 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
20659 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
20660 condition/operands for integer GE/LE/GEU/LEU operations.
20662 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
20664 PR translation/79397
20665 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
20668 2017-02-08 Martin Jambor <mjambor@suse.cz>
20671 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
20672 whether allocation happened.
20673 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
20674 nothing was allocated.
20676 2017-02-08 Jakub Jelinek <jakub@redhat.com>
20678 PR tree-optimization/79408
20679 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
20680 constant, but SSA_NAME with a known integer range, use the minimum
20681 of that range instead of op1 to determine if modulo can be replaced
20682 with its first operand.
20684 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20686 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
20688 2017-02-08 Richard Biener <rguenther@suse.de>
20690 PR tree-optimization/71824
20691 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
20692 Check all loops contained in the merged region.
20694 2017-02-07 Andrew Pinski <apinski@cavium.com>
20696 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
20698 2017-02-07 Andrew Pinski <apinski@cavium.com>
20700 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
20701 (thunderxt88): Likewise.
20702 (thunderxt81): Disable LSE and change v8.1 to v8.
20703 (thunderxt83): Likewise.
20705 2017-02-07 Jakub Jelinek <jakub@redhat.com>
20706 Richard Biener <rguenther@suse.de>
20708 PR middle-end/79399
20709 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
20710 type from int to size_t.
20711 * ira-costs.c (struct_costs_size): Change type from int to size_t.
20713 2017-02-07 Jakub Jelinek <jakub@redhat.com>
20715 PR rtl-optimization/79386
20716 * cprop.c (bypass_conditional_jumps): Initialize
20717 bypass_last_basic_block already before splitting bbs after
20718 unconditional traps...
20719 (bypass_conditional_jumps): ... rather than here.
20722 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
20723 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
20724 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
20725 fix -masm=intel patterns.
20727 2017-02-07 Richard Biener <rguenther@suse.de>
20729 PR tree-optimization/79256
20730 PR middle-end/79278
20731 * builtins.c (get_object_alignment_2): Use min_align_of_type
20732 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
20733 and ADJUST_FIELD_ALIGN.
20735 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
20737 * doc/tm.texi: Regenerate.
20738 * stor-layout.c (layout_decl): Adjust.
20739 (update_alignment_for_field): Likewise.
20740 (place_field): Likewise.
20741 (min_align_of_type): Likewise.
20742 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
20743 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
20744 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
20745 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
20746 * config/frv/frv.c (frv_adjust_field_align): Likewise.
20747 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
20748 * config/i386/i386.c (x86_field_alignment): Likewise.
20749 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
20750 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
20751 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
20752 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
20753 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
20754 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
20758 2017-01-30 Richard Biener <rguenther@suse.de>
20760 PR tree-optimization/79256
20761 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
20762 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
20765 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
20767 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
20768 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
20769 builtins to SImode and emit a zero-extend, if necessary.
20771 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
20773 * docs/invoke.texi (RISC-V Options): Alphabetize.
20775 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
20777 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
20780 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
20782 * config/riscv/riscv.c: New file.
20783 * common/config/riscv/riscv-common.c: Likewise.
20784 * config.gcc: Likewise.
20785 * config/riscv/constraints.md: Likewise.
20786 * config/riscv/elf.h: Likewise.
20787 * config/riscv/generic.md: Likewise.
20788 * config/riscv/linux.h: Likewise.
20789 * config/riscv/multilib-generator: Likewise.
20790 * config/riscv/peephole.md: Likewise.
20791 * config/riscv/pic.md: Likewise.
20792 * config/riscv/predicates.md: Likewise.
20793 * config/riscv/riscv-builtins.c: Likewise.
20794 * config/riscv/riscv-c.c: Likewise.
20795 * config/riscv/riscv-ftypes.def: Likewise.
20796 * config/riscv/riscv-modes.def: Likewise.
20797 * config/riscv/riscv-opts.h: Likewise.
20798 * config/riscv/riscv-protos.h: Likewise.
20799 * config/riscv/riscv.h: Likewise.
20800 * config/riscv/riscv.md: Likewise.
20801 * config/riscv/riscv.opt: Likewise.
20802 * config/riscv/sync.md: Likewise.
20803 * config/riscv/t-elf-multilib: Likewise.
20804 * config/riscv/t-linux: Likewise.
20805 * config/riscv/t-linux-multilib: Likewise.
20806 * config/riscv/t-riscv: Likewise.
20807 * configure.ac: Likewise.
20808 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
20809 Waterman as RISC-V maintainers.
20810 * doc/install.texi: Add RISC-V entries.
20811 * doc/invoke.texi: Add RISC-V options section.
20812 * doc/md.texi: Add RISC-V constraints section.
20813 * configure: Regenerated.
20815 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
20818 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
20819 false values to be constant vectors with all 0 or all 1 bits set.
20820 (vcondu<mode><mode>): Likewise.
20821 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
20823 (fpmask_comparison_operator): Update comment.
20824 (vecint_comparison_operator): New predicate.
20825 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
20826 vector conditionals when the true and false values are constant
20827 vectors with all 0 bits or all 1 bits set.
20829 2017-02-06 Martin Sebor <msebor@redhat.com>
20831 PR tree-optimization/79376
20832 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
20834 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
20836 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
20837 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
20838 to simplify split condition.
20840 2017-02-06 Jakub Jelinek <jakub@redhat.com>
20842 * omp-expand.c (oxpand_omp_atomic_fetch_op,
20843 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
20846 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
20848 PR rtl-optimization/68664
20849 * target.def (can_speculate_insn): New hook.
20850 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
20851 * doc/tm.texi: Regenerate.
20852 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
20853 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
20854 (rs6000_sched_can_speculate_insn): New function.
20856 2017-02-06 Jakub Jelinek <jakub@redhat.com>
20858 PR tree-optimization/79284
20859 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
20860 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
20861 vectorizable_mask_load_store, vectorizable_operation,
20862 vect_is_simple_cond, get_same_sized_vectype): Use it instead
20863 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
20864 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
20865 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
20866 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
20867 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
20868 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
20869 is_gimple_assign (stmt). Replace another such test with
20870 is_gimple_assign (stmt).
20872 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
20875 * config/avr/avr.c (rtl-iter.h): Include it.
20876 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
20877 (avr_legitimate_combined_insn): ...and implementation.
20879 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
20881 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
20882 * config/s390/s390.c (s390_const_operand_ok)
20883 (s390_canonicalize_comparison, s390_extract_part)
20884 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
20885 (s390_contiguous_bitmask_p, s390_rtx_costs)
20886 (legitimize_pic_address): Likewise.
20887 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
20888 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
20889 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
20890 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
20891 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
20893 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
20895 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
20896 REGNO($0) == REGNO($1).
20898 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20900 * config/s390/linux.h(SIZE_TYPE): Add comment.
20902 2017-02-06 Julian Brown <julian@codesourcery.com>
20903 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20904 Virendra Pathak <virendra.pathak@broadcom.com>
20906 * config/aarch64/aarch64-cores.def: Change the scheduler
20908 * config/aarch64/aarch64.md: Include thunderx2t99.md.
20909 * config/aarch64/thunderx2t99.md: New file.
20911 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20913 * doc/standards.texi (Go Language): Update link to language
20916 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
20918 * tree-eh.c (lower_resx): Sanitize profile.
20919 (cleanup_empty_eh_move_lp): Likewise.
20921 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
20924 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
20926 * cfgloopmanip.h (loop_version): Update prototype.
20927 * modulo-sched.c (sms_schedule): Update call of loop_version.
20928 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
20929 * tree-parloops.c (gen_parallel_loop): Likewise.
20930 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
20931 * tree-ssa-loop-split.c (split_loop): Likewise.
20932 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
20933 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
20935 2017-02-05 Martin Liska <mliska@suse.cz>
20938 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
20940 (print_operand_address): Initialize a struct to zero.
20942 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20944 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
20945 garbage collector only in textual form.
20947 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20949 * doc/extend.texi (x86 specific memory model extensions for
20950 transactional memory): Simplify a phrase.
20952 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
20955 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
20956 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
20957 (atomic_storedi_1): Likewise.
20959 2017-02-04 Jakub Jelinek <jakub@redhat.com>
20961 PR tree-optimization/79338
20962 * tree-parloops.c (gather_scalar_reductions): Don't call
20963 vect_analyze_loop_form for loop->inner before destroying loop's
20966 2017-02-03 Martin Sebor <msebor@redhat.com>
20968 PR tree-optimization/79327
20969 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
20970 when precision has resulted in leading zeros.
20971 (format_integer): Adjust the likely counter to assume an unknown
20972 argument that may be zero is non-zero.
20974 2017-02-03 Jason Merrill <jason@redhat.com>
20977 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
20978 avoid copying non-taken branch.
20980 2017-02-03 Jakub Jelinek <jakub@redhat.com>
20982 PR tree-optimization/79340
20983 * tree-vect-loop.c (vectorizable_reduction): Release
20984 vec_defs elements after safe_splicing them into other vectors.
20987 PR tree-optimization/79327
20988 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
20989 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
20991 (format_integer): Use wide_int_to_tree instead of build_int_cst
20992 + to_?hwi. If argmin is NULL, just set argmin and argmax to
20993 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
20994 of shortest and longest sequence.
20996 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
20998 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
20999 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
21001 2017-02-03 Walter Lee <walt@tilera.com>
21004 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
21005 after initial stackframe link reg save.
21006 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
21008 2017-02-03 Jakub Jelinek <jakub@redhat.com>
21011 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
21012 wu for stxssp alternative.
21014 2017-02-03 Martin Sebor <msebor@redhat.com>
21016 PR tree-optimization/79352
21017 * gimple-fold.c (get_range_strlen): Add argument.
21018 (get_range_strlen): Change return type to bool.
21019 (get_maxval_strlen): Pass in a dummy argument.
21020 * gimple-fold.h (get_range_strlen): Change return type to bool.
21021 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
21022 * tree.h (array_at_struct_end_p): Add argument.
21023 * tree.c (array_at_struct_end_p): Handle it.
21025 2017-02-03 Martin Liska <mliska@suse.cz>
21028 * multiple_target.c (create_dispatcher_calls): Redirect edge
21029 from a caller of a dispatcher.
21030 (expand_target_clones): Make the clones local.
21031 (ipa_target_clone): Do both target clones and resolvers.
21032 (ipa_dispatcher_calls): Remove the pass.
21033 (pass_dispatcher_calls::gate): Likewise.
21034 (make_pass_dispatcher_calls): Likewise.
21035 * passes.def (pass_target_clone): Put as very first IPA early
21038 2017-02-03 Martin Liska <mliska@suse.cz>
21040 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
21041 in case of a function with ifunc attribute.
21043 2017-02-03 Martin Liska <mliska@suse.cz>
21045 * cgraph.c (cgraph_node::dump): Dump function version info.
21046 * symtab.c (symtab_node::dump_base): Add missing new line.
21048 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
21050 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
21051 (ifcombine_ifandif): Use it.
21053 2017-02-03 Martin Liska <mliska@suse.cz>
21055 * doc/invoke.texi: Document default value for
21056 use-after-scope-direct-emission-threshold.
21058 2017-02-03 Martin Liska <mliska@suse.cz>
21060 PR tree-optimization/79339
21061 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
21062 (format_floating): Likewise.
21064 2017-02-03 Martin Liska <mliska@suse.cz>
21067 * ipa-prop.c (ipa_node_params_t::insert): Remove current
21069 (ipa_node_params_t::remove): Likewise.
21070 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
21071 initialization from removed ipa_node_params_t::insert.
21072 (ipa_node_params::~ipa_node_params): Move from removed
21073 ipa_node_params_t::release.
21074 * symbol-summary.h (symbol_summary::m_released): New member.
21075 Do not release a summary twice. Do not allow to call finalizer
21076 for types of a summary that live in GGC memory.
21078 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21080 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
21083 2017-02-02 Martin Sebor <msebor@redhat.com>
21085 PR middle-end/79275
21086 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
21087 (format_string): Tighten up the range of output for non-constant
21088 strings and correct the expected range for wide non-constant strings.
21090 2017-02-02 Martin Sebor <msebor@redhat.com>
21092 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
21094 PR middle-end/32003
21095 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
21097 (-fdump-tree-@var): Add to index and document how to come up
21098 with pass-specific option and dump file names.
21099 (-fdump-passes): Clarify where to look for output.
21101 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
21103 PR middle-end/77445
21104 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
21105 statistics of the analyzed path; allow threading for speed when
21106 any of BBs along the path are optimized for speed.
21108 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
21110 PR middle-end/78468
21111 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
21112 settings of the virtual registers.
21115 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
21117 * explow.c (get_dynamic_stack_size): Take known alignment of stack
21118 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
21121 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21123 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
21124 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
21126 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21128 * config/s390/s390.md: Add missing comments with the expanded
21130 * config/s390/vector.md: Likewise.
21131 * config/s390/vx-builtins.md: Likewise.
21133 2017-02-02 Jakub Jelinek <jakub@redhat.com>
21136 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
21137 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
21138 conditions on a single line.
21140 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21142 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
21143 __S390_VX__ to __VX__.
21145 2017-02-01 Andrew Pinski <apinski@cavium.com>
21147 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
21148 stmt_info to record_stmt_cost.
21149 (vect_get_known_peeling_cost): Pass stmt_info if known to
21151 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
21152 cpu_vector_cost field into
21153 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
21154 field into vec_int_stmt_cost and vec_fp_stmt_cost.
21155 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
21156 splitting of scalar_stmt_cost and vec_stmt_cost.
21157 (thunderx_vector_cost): Likewise.
21158 (cortexa57_vector_cost): LIkewise.
21159 (exynosm1_vector_cost): Likewise.
21160 (xgene1_vector_cost): Likewise.
21161 (thunderx2t99_vector_cost): Improve after the splitting of the two
21163 (aarch64_builtin_vectorization_cost): Update for the splitting of
21164 scalar_stmt_cost and vec_stmt_cost.
21166 2017-02-01 Torvald Riegel <triegel@redhat.com>
21167 Richard Henderson <rth@redhat.com>
21169 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
21170 conditional on existance of a fast atomic load.
21171 * optabs-query.c (can_atomic_load_p): New function.
21172 * optabs-query.h (can_atomic_load_p): Declare it.
21173 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
21174 no fast atomic load is available for the particular size of access.
21175 (expand_atomic_compare_and_swap): Likewise.
21176 (expand_atomic_load): Likewise.
21177 (expand_atomic_store): Likewise.
21178 (expand_atomic_fetch_op): Likewise.
21179 * testsuite/lib/target-supports.exp
21180 (check_effective_target_sync_int_128): Remove x86 because it provides
21181 no fast atomic load.
21182 (check_effective_target_sync_int_128_runtime): Likewise.
21184 2017-02-01 Richard Biener <rguenther@suse.de>
21186 * graphite.c: Include tree-vectorizer.h for find_loop_location.
21187 (graphite_transform_loops): Provide opt-info for optimized nests.
21188 * tree-parloop.c (parallelize_loops): Provide opt-info for
21189 parallelized loops.
21191 2017-02-01 Richard Biener <rguenther@suse.de>
21193 PR middle-end/79315
21194 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
21195 was not set before.
21197 2017-02-01 Richard Biener <rguenther@suse.de>
21199 PR tree-optimization/71824
21200 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
21201 Verify the loops are valid in the merged SESE region.
21202 (scop_detection::can_represent_loop_1): Check analyzing the
21203 evolution of the number of iterations in the region succeeds.
21205 2017-01-31 Ian Lance Taylor <iant@golang.org>
21207 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
21208 REG_ARGS_SIZE note to 32-bit push insns and call insn.
21210 2017-01-31 David Malcolm <dmalcolm@redhat.com>
21212 PR preprocessor/79210
21213 * input.c (get_substring_ranges_for_loc): Replace line_width
21214 assertion with error-handling.
21216 2017-01-31 Richard Biener <rguenther@suse.de>
21218 PR tree-optimization/77318
21219 * graphite-sese-to-poly.c (extract_affine): Fix assert.
21220 (create_pw_aff_from_tree): Take loop parameter.
21221 (add_condition_to_pbb): Pass loop of the condition to
21222 create_pw_aff_from_tree.
21224 2017-01-31 Jakub Jelinek <jakub@redhat.com>
21226 * config/s390/s390.c (s390_asan_shadow_offset): New function.
21227 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
21229 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
21233 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
21235 (convert_int_to_float128): Likewise.
21236 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
21237 (convert_int_to_float128): Likewise.
21238 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
21239 (UNSPEC_IEEE128_CONVERT): Likewise.
21240 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
21241 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
21242 Use local variables for IBM extended format.
21243 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
21244 (fix_trunc<mode>si2_fprs): Likewise.
21245 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
21246 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
21247 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
21248 to know that we can now have integers of all sizes in vector
21250 (fix<uns>_<mode>di2_hw): Likewise.
21251 (float<uns>_<mode>si2_hw): Likewise.
21252 (fix_<mode>si2_hw): Likewise.
21253 (fixuns_<mode>si2_hw): Likewise.
21254 (float<uns>_<mode>di2_hw): Likewise.
21255 (float_<mode>di2_hw): Likewise.
21256 (float_<mode>si2_hw): Likewise.
21257 (floatuns_<mode>di2_hw): Likewise.
21258 (floatuns_<mode>si2_hw): Likewise.
21259 (xscvqp<su>wz_<mode>): Delete, no longer used.
21260 (xscvqp<su>dz_<mode>): Likewise.
21261 (xscv<su>dqp_<mode>): Likewise.
21262 (ieee128_mfvsrd_64bit): Likewise.
21263 (ieee128_mfvsrd_32bit): Likewise.
21264 (ieee128_mfvsrwz): Likewise.
21265 (ieee128_mtvsrw): Likewise.
21266 (ieee128_mtvsrd_64bit): Likewise.
21267 (ieee128_mtvsrd_32bit): Likewise.
21269 2017-01-31 Martin Liska <mliska@suse.cz>
21272 * ipa-prop.c (ipa_free_all_node_params): Call release method
21273 instead of ~sumbol_summary to not to trigger double times
21276 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
21278 PR tree-optimization/71691
21279 * bitmap.h (class auto_bitmap): New.
21280 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
21281 is_maybe_undefined instead of ssa_undefined_value_p.
21283 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21285 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
21286 __S390_ARCH_LEVEL__ to __ARCH__.
21288 2017-01-31 Jakub Jelinek <jakub@redhat.com>
21290 PR tree-optimization/79267
21291 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
21292 if should_remove_lhs_p is true.
21294 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
21297 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
21298 (add_alignment_attribute): New.
21299 (base_type_die): Add alignment attribute.
21300 (subrange_type_die): Likewise.
21301 (modified_type_die): Likewise.
21302 (gen_array_type_die): Likewise.
21303 (gen_descr_array_type_die: Likewise.
21304 (gen_enumeration_type_die): Likewise.
21305 (gen_subprogram_die): Likewise.
21306 (gen_variable_die): Likewise.
21307 (gen_field_die): Likewise.
21308 (gen_ptr_to_mbr_type_die): Likewise.
21309 (gen_struct_or_union_type_die): Likewise.
21310 (gen_subroutine_type_die): Likewise.
21311 (gen_typedef_die): Likewise.
21312 (base_type_cmp): Compare alignment attribute.
21314 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21317 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
21318 (setb_unsigned) New pattern for setb with CCUNS.
21319 * config/rs6000/rs6000.c (expand_block_compare): Use a different
21320 subfc./subfe sequence to avoid overflow problems. Generate a
21321 shorter sequence with cmpld/setb for power9.
21322 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
21323 for generating subfc. instruction.
21324 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
21325 now uses this instruction.
21327 2017-01-30 Ian Lance Taylor <iant@google.com>
21330 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
21331 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
21333 2017-01-30 Martin Sebor <msebor@redhat.com>
21335 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
21336 Move constant to the right of a relational operator.
21337 (get_mpfr_format_length, format_character, format_string): Ditto.
21338 (should_warn_p, maybe_warn): Same.
21340 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
21342 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
21345 * asan.c (get_translation_unit_decl): Remove function.
21346 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
21348 2017-01-30 Martin Liska <mliska@suse.cz>
21350 PR gcov-profile/79259
21351 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
21352 -fprofile-generate.
21354 2017-01-30 Martin Liska <mliska@suse.cz>
21357 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
21358 Initialize variables with NULL value.
21360 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
21363 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
21365 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
21367 2017-01-30 Richard Biener <rguenther@suse.de>
21369 PR tree-optimization/79276
21370 * tree-vrp.c (process_assert_insertions): Properly adjust common
21371 when removing a duplicate.
21373 * gcc.dg/torture/pr79276.c: New testcase.
21375 2017-01-30 Richard Biener <rguenther@suse.de>
21377 PR tree-optimization/79256
21378 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
21379 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
21381 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
21383 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
21386 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
21387 ("*r<noxa>sbg_<mode>_sll_bitmask")
21388 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
21389 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
21390 Use contiguous_bitmask_nowrap_operand.
21392 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21395 * config/rs6000/altivec.h (vec_xl): Revise #define.
21396 (vec_xst): Likewise.
21398 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
21400 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
21402 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
21404 PR rtl-optimization/79194
21405 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
21406 traps before call to bypass_conditional_jumps.
21408 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
21410 PR tree-optimization/71374
21411 * lra-constraints.c (check_conflict_input_operands): New.
21412 (match_reload): Use it.
21414 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
21417 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
21418 account to calculate conflict_set.
21420 2017-01-27 Bin Cheng <bin.cheng@arm.com>
21422 PR rtl-optimization/78559
21423 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
21424 other_insn in combine.
21426 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
21428 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
21429 uint16_type_node for BT_UINT16.
21431 2017-01-27 David Malcolm <dmalcolm@redhat.com>
21433 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
21434 "RTL Tests" to menu.
21435 (GIMPLE Tests): New node.
21436 (RTL Tests): New node.
21438 2017-01-27 Richard Biener <rguenther@suse.de>
21440 PR tree-optimization/79245
21441 * tree-loop-distribution.c (distribute_loop): Apply cost
21442 modeling also to detected patterns.
21444 2017-01-27 Richard Biener <rguenther@suse.de>
21446 PR tree-optimization/71433
21447 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
21448 (compare_assert_loc): New function.
21449 (process_assert_insertions): Sort and optimize assert locations
21450 to remove duplicates and push down identical assertions on
21451 edges to their destination block.
21453 2017-01-27 Richard Biener <rguenther@suse.de>
21455 PR tree-optimization/79244
21456 * tree-vrp.c (remove_range_assertions): Forcefully propagate
21457 out SSA names even if abnormal.
21459 2017-01-27 Jakub Jelinek <jakub@redhat.com>
21461 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
21462 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
21463 instead of MPFR_RNDN.
21465 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
21468 * arm.c (arm_option_override): Don't call build_target_option_node
21469 until after doing all option overrides.
21470 (arm_valid_target_attribute_tree): Likewise.
21472 2017-01-27 Martin Liska <mliska@suse.cz>
21474 * doc/invoke.texi (-fprofile-arcs): Document profiling support
21475 for {cd}tors and C++ {cd}tors.
21477 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
21479 * config/s390/s390.md ("*setmem_long_and")
21480 ("*setmem_long_and_31z"): Use zero_extend instead of and.
21482 2017-01-26 Martin Sebor <msebor@redhat.com>
21484 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
21487 2017-01-26 Martin Sebor <msebor@redhat.com>
21489 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
21490 HAVE_DFmode before using XFmode or DFmode.
21491 (parse_directive): Avoid using the z length modifier to avoid
21492 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
21494 PR middle-end/78703
21495 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
21496 to accept adjustment as an array.
21497 (get_int_range): New function.
21498 (struct directive): Make width and prec arrays.
21499 (directive::set_width, directive::set_precision): Call get_int_range.
21500 (format_integer, format_floating): Handle width and precision ranges.
21501 (format_string, parse_directive): Same.
21503 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21506 * dwarf2out.c (generate_skeleton_bottom_up): For children with
21507 comdat_type_p set, just clone them, but keep the children in the
21511 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
21512 which have direct callers with -fvar-tracking-assignments enabled
21514 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
21515 inside of type units.
21517 2017-01-26 Martin Sebor <msebor@redhat.com>
21519 PR middle-end/78703
21520 * gimple-ssa-sprintf.c (struct result_range): Add likely and
21522 (struct format_result): Replace number_chars, number_chars_min,
21523 and number_chars_max with a single member of struct result_range.
21525 (format_result::operator+=): Adjust.
21526 (struct fmtresult): Remove bounded. Handle likely and unlikely
21528 (fmtresult::adjust_for_width_or_precision): New function.
21529 (fmtresult:type_max_digits): New function.
21530 (bytes_remaining): Handle likely and unlikely counters.
21531 (min_bytes_remaining): Remove.
21532 (format_percent): Simplify.
21533 (format_integer, format_floating): Set likely and unlikely counters.
21534 (get_string_length, format_character, format_string): Same.
21535 (format_plain, should_warn_p): New function.
21536 (maybe_warn): Call should_warn_p. Update diagnostic messages
21537 and handle those for all directives, including plain strings.
21538 (format_directive): Handle likely and unlikely counters.
21539 Remove unnecessary quoting from diagnostics. Add an informational
21541 (add_bytes): Remove.
21542 (pass_sprintf_length::compute_format_length): Simplify.
21543 (try_substitute_return_value): Handle likely and unlikely counters.
21545 2017-01-26 Carl Love <cel@us.ibm.com>
21547 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
21548 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
21550 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
21553 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
21554 endianess for subregs into account.
21555 * lra-constraints.c (lra_constraints): Do risky transformations
21556 always on the first iteration.
21557 * lra-lives.c (check_pseudos_live_through_calls): Add arg
21558 last_call_used_reg_set.
21559 (process_bb_lives): Define and use last_call_used_reg_set.
21560 * lra.c (lra): Always continue after lra_constraints on the first
21563 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
21565 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
21567 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
21569 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21571 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
21572 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
21573 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
21574 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
21575 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
21576 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
21577 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
21578 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
21579 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
21581 2017-01-26 Marek Polacek <polacek@redhat.com>
21584 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
21585 for the third operand.
21587 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21589 PR middle-end/79236
21590 * omp-low.c (struct omp_context): Add simt_stmt field.
21591 (scan_omp_for): Return omp_context *.
21592 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
21593 context to the _simt_ SIMD stmt.
21594 (lower_omp_for): For combined SIMD with sibling _simt_
21595 SIMD, make sure to use the same decls in _looptemp_
21596 clauses as in the sibling.
21598 2017-01-26 David Sherwood <david.sherwood@arm.com>
21600 PR middle-end/79212
21601 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
21604 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21607 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
21608 emit fld b; fld a; if possible.
21610 * brig-builtins.def: Update copyright years.
21611 * config/arm/arm_acle_builtins.def: Update copyright years.
21613 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
21616 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
21617 constraint instead of o for the stxsd instruction.
21619 2017-01-25 Carl Love <cel@us.ibm.com>
21621 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
21622 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
21624 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
21626 * doc/invoke.texi (C++ Dialect Options): Fix typo.
21628 2017-01-25 Richard Biener <rguenther@suse.de>
21630 PR tree-optimization/69264
21631 * target.def (vector_alignment_reachable): Improve documentation.
21632 * doc/tm.texi: Regenerate.
21633 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
21635 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
21636 earlier changes with respect to TYPE_USER_ALIGN.
21637 (vector_alignment_reachable_p): Likewise. Improve dumping.
21639 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21642 * config/arm/arm.md (xordi3): Force constant operand into a register
21645 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21647 * doc/invoke.texi (-fstore-merging): Correct default optimization
21648 levels at which it is enabled.
21649 (-O): Move -fstore-merging from list to...
21652 2017-01-25 Richard Biener <rguenther@suse.de>
21655 * omp-expand.c: Include debug.h.
21656 (expand_omp_taskreg): Make sure to generate early debug before
21657 outlining anything from a function.
21658 (expand_omp_target): Likewise.
21659 (grid_expand_target_grid_body): Likewise.
21661 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
21664 * asan.c (get_translation_unit_decl): New function.
21665 (asan_add_global): Extract modules file name from globals
21666 TRANSLATION_UNIT_DECL name.
21668 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
21671 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
21672 for long calls with APCS frame and VFP.
21674 2017-01-24 David Malcolm <dmalcolm@redhat.com>
21676 * cfg.c (original_copy_tables_initialized_p): New function.
21677 * cfg.h (original_copy_tables_initialized_p): New decl.
21678 * cfgrtl.c (relink_block_chain): Guard the call to
21679 free_original_copy_tables with a call to
21680 original_copy_tables_initialized_p.
21681 * cgraph.h (symtab_node::native_rtl_p): New decl.
21682 * cgraphunit.c (symtab_node::native_rtl_p): New function.
21683 (symtab_node::needed_p): Don't assert for early assembly output
21684 for __RTL functions.
21685 (cgraph_node::finalize_function): Set "force_output" for __RTL
21687 (cgraph_node::analyze): Bail out early for __RTL functions.
21688 (analyze_functions): Update assertion to support __RTL functions.
21689 (cgraph_node::expand): Bail out early for __RTL functions.
21690 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
21692 * function.h (struct function): Update comment for field
21694 * gimple-expr.c: Include "tree-pass.h".
21695 (gimple_has_body_p): Return false for __RTL functions.
21696 * Makefile.in (OBJS): Add run-rtl-passes.o.
21697 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
21699 (gcc::pass_manager::get_clean_slate): New accessor.
21700 * passes.c: Include "insn-addr.h".
21701 (should_skip_pass_p): Add logging. Update logic for running
21702 "expand" to be compatible with both __GIMPLE and __RTL. Guard
21703 property-provider override so it is only done for gimple passes.
21705 (skip_pass): New function.
21706 (execute_one_pass): Call skip_pass when skipping passes.
21707 * read-md.c (md_reader::read_char): Support filtering
21708 the input to a subset of line numbers.
21709 (md_reader::md_reader): Initialize fields
21710 m_first_line and m_last_line.
21711 (md_reader::read_file_fragment): New function.
21712 * read-md.h (md_reader::read_file_fragment): New decl.
21713 (md_reader::m_first_line): New field.
21714 (md_reader::m_last_line): New field.
21715 * read-rtl-function.c (function_reader::create_function): Only
21716 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
21717 curr_properties. Set DECL_INITIAL to a dummy block.
21718 (read_rtl_function_body_from_file_range): New function.
21719 * read-rtl-function.h (read_rtl_function_body_from_file_range):
21721 * run-rtl-passes.c: New file.
21722 * run-rtl-passes.h: New file.
21724 2017-01-24 Jeff Law <law@redhat.com>
21726 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
21729 2017-01-24 Bin Cheng <bin.cheng@arm.com>
21731 PR tree-optimization/79159
21732 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
21733 (record_nonwrapping_iv): Improve boundary using above function if no
21734 value range information.
21736 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
21737 Martin Jambor <mjambor@suse.cz>
21739 * brig-builtins.def: New file.
21740 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
21741 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
21742 (DEF_HSAIL_SAT_BUILTIN): Likewise.
21743 (DEF_HSAIL_INTR_BUILTIN): Likewise.
21744 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
21745 * builtin-types.def (BT_INT8): New.
21746 (BT_INT16): Likewise.
21747 (BT_UINT8): Likewise.
21748 (BT_UINT16): Likewise.
21749 (BT_FN_ULONG): Likewise.
21750 (BT_FN_UINT_INT): Likewise.
21751 (BT_FN_UINT_ULONG): Likewise.
21752 (BT_FN_UINT_LONG): Likewise.
21753 (BT_FN_UINT_PTR): Likewise.
21754 (BT_FN_ULONG_PTR): Likewise.
21755 (BT_FN_INT8_FLOAT): Likewise.
21756 (BT_FN_INT16_FLOAT): Likewise.
21757 (BT_FN_UINT32_FLOAT): Likewise.
21758 (BT_FN_UINT16_FLOAT): Likewise.
21759 (BT_FN_UINT8_FLOAT): Likewise.
21760 (BT_FN_UINT64_FLOAT): Likewise.
21761 (BT_FN_UINT16_UINT32): Likewise.
21762 (BT_FN_UINT32_UINT16): Likewise.
21763 (BT_FN_UINT16_UINT16_UINT16): Likewise.
21764 (BT_FN_INT_PTR_INT): Likewise.
21765 (BT_FN_UINT_PTR_UINT): Likewise.
21766 (BT_FN_LONG_PTR_LONG): Likewise.
21767 (BT_FN_ULONG_PTR_ULONG): Likewise.
21768 (BT_FN_VOID_UINT64_UINT64): Likewise.
21769 (BT_FN_UINT8_UINT8_UINT8): Likewise.
21770 (BT_FN_INT8_INT8_INT8): Likewise.
21771 (BT_FN_INT16_INT16_INT16): Likewise.
21772 (BT_FN_INT_INT_INT): Likewise.
21773 (BT_FN_UINT_FLOAT_UINT): Likewise.
21774 (BT_FN_FLOAT_UINT_UINT): Likewise.
21775 (BT_FN_ULONG_UINT_UINT): Likewise.
21776 (BT_FN_ULONG_UINT_PTR): Likewise.
21777 (BT_FN_ULONG_ULONG_ULONG): Likewise.
21778 (BT_FN_UINT_UINT_UINT): Likewise.
21779 (BT_FN_VOID_UINT_PTR): Likewise.
21780 (BT_FN_UINT_UINT_PTR: Likewise.
21781 (BT_FN_UINT32_UINT64_PTR): Likewise.
21782 (BT_FN_INT_INT_UINT_UINT): Likewise.
21783 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
21784 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
21785 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
21786 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
21787 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
21788 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
21789 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
21790 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
21791 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
21792 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
21793 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
21794 * doc/frontends.texi: List BRIG FE.
21795 * doc/install.texi (Testing): Add BRIG tesring requirements.
21796 * doc/invoke.texi (Overall Options): Mention BRIG.
21797 * doc/standards.texi (Standards): Doucment BRIG HSA version.
21799 2017-01-24 Richard Biener <rguenther@suse.de>
21801 PR translation/79208
21802 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
21804 2017-01-24 Martin Jambor <mjambor@suse.cz>
21807 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
21808 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
21809 and known_contexts.
21811 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
21813 PR middle-end/79123
21814 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
21815 casts from signed to unsigned really don't have a range.
21817 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
21819 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
21820 GMP_RNDx for compatiblity.
21822 2017-01-24 Martin Liska <mliska@suse.cz>
21825 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
21826 that would prevent us to call alloca with -1 as argument.
21828 2017-01-24 Jakub Jelinek <jakub@redhat.com>
21830 * dwarf2out.c (output_compilation_unit_header, output_file_names):
21831 Avoid -Wformat-security warning.
21833 2017-01-23 Andrew Pinski <apinski@cavium.com>
21835 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
21838 2017-01-23 Martin Sebor <msebor@redhat.com>
21840 PR middle-end/78703
21841 * gimple-ssa-sprintf.c (warn_level): New global.
21842 (format_integer): Use it here and throughout the rest of the file.
21843 Use the same switch to compute sign as base.
21844 (maybe_warn): New function.
21845 (format_directive): Factor out warnings into maybe_warn.
21846 Add debugging output. Use warn_level.
21847 (add_bytes): Use warn_level.
21848 (pass_sprintf_length::compute_format_length): Add debugging output.
21849 (try_substitute_return_value): Same.
21850 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
21852 PR middle-end/78703
21853 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
21854 (struct fmtresult, format_integer, format_floating): Adjust.
21855 (fmtresult::fmtresult): Set max correctly in two argument ctor.
21856 (get_string_length, format_string,format_directive): Same.
21857 (pass_sprintf_length::compute_format_length): Same.
21858 (try_substitute_return_value): Simplify slightly.
21860 PR middle-end/78703
21861 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
21862 (fmtresult::operator+=): Outlined.
21863 (struct fmtresult): Add ctors.
21864 (struct conversion_spec): Rename...
21865 (struct directive): ...to this. Add and remove data members.
21866 (directive::set_width, directive::set_precision): New functions.
21867 (format_percent): Use fmtresult ctor.
21868 (get_width_and_precision): Remove.
21869 (format_integer): Make naming changes. Avoid computing width and
21871 (format_floating): Same. Adjust indentation.
21872 (format_character, format_none): New functions.
21873 (format_string): Moved character handling to format_character.
21874 (format_directive): Remove arguments, change return type.
21875 (parse_directive): New function.
21876 (pass_sprintf_length::compute_format_length): Move directive
21877 parsing to parse_directive.
21879 2017-01-23 Jakub Jelinek <jakub@redhat.com>
21881 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
21882 (assign_assembler_name_if_needed): ... this.
21883 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
21884 (assign_assembler_name_if_needed): ... this.
21885 (free_lang_data_in_cgraph): Adjust callers.
21886 * cgraphunit.c (cgraph_node::analyze): Likewise.
21887 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
21890 2017-01-23 Richard Biener <rguenther@suse.de>
21892 PR tree-optimization/79088
21893 PR tree-optimization/79188
21894 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
21895 resetting loop bounds after last path deletion. Reset loop
21896 bounds of the target loop, make code match the comments.
21897 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
21898 Make sure loops need no fixups.
21900 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
21902 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
21903 exponent support with double type for first argument.
21904 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
21905 type returned by __builtin_vec_extract_sig,
21906 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
21907 functions from "vector int" to "vector unsigned int" or from
21908 "vector long long int" to "vector unsigned long long int".
21909 Changed type returned by __builtin_vec_extract_exp,
21910 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
21911 functions from "vector int" to "vector unsigned int" or from
21912 "vector long long int" to "vector unsigned long long int".
21913 Changed return type of __builtin_vec_test_data_class,
21914 __builtin_vec_test_data_class_sp, and
21915 __builtin_vec_test_data_class_dp from "vector int" to
21916 "vector bool int" or from "vector long long int" to "vector bool
21917 long long int" and changed second argument type from "unsigned
21918 int" to "int". Added new overloaded function forms "vector float
21919 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
21920 "vector float __builtin_vec_insert_exp_sp (vector float, vector
21921 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
21922 double, vector unsigned long long int)" and "vector double
21923 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
21924 long int)". Changed return type of
21925 __builtin_scalar_test_data_class and
21926 __builtin_scalar_test_data_class_sp and
21927 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
21928 int" and changed second argument from "unsigned int" to "int".
21929 Changed type returned by __builtin_scalar_test_neg,
21930 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
21931 from "int" to "bool int". Added new overloaded function form
21932 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
21933 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
21934 exponent double-precision with floating point first argument.
21935 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
21936 documentation of scalar_test_data_class, scalar_test_neg,
21937 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
21938 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
21939 vec_test_data_class built-in functions to reflect refinements in
21940 their type signatures.
21942 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
21944 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
21946 (aarch64_elf_asm_destructor): Likewise.
21948 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
21950 PR rtl-optimization/78634
21951 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
21952 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
21953 * ifcvt.c (noce_try_cmove): Add missing cost check.
21955 PR rtl-optimization/71724
21956 * combine.c (if_then_else_cond): Look for situations where it is
21957 beneficial to undo the work of one of the recursive calls.
21959 2017-01-23 Bin Cheng <bin.cheng@arm.com>
21961 PR tree-optimization/70754
21962 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
21963 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
21964 combined stmt before it if not NULL.
21965 (combine_chains): Process refs reversely and compute dominance point
21968 2017-01-23 Martin Liska <mliska@suse.cz>
21970 PR tree-optimization/79196
21971 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
21972 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
21974 (strlen_optimize_stmt): Call the renamed function.
21976 2017-01-23 Michael Matz <matz@suse.de>
21978 PR tree-optimization/78384
21979 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
21981 2017-01-23 Richard Biener <rguenther@suse.de>
21983 PR tree-optimization/79186
21984 * tree-vrp.c (register_new_assert_for): Make sure we've seen
21985 both incoming edges before moving an assert.
21987 2017-01-23 Martin Jambor <mjambor@suse.cz>
21989 * ipa-prop.c (load_from_param_1): Removed.
21990 (load_from_unmodified_param): Bits from load_from_param_1 put back
21992 (load_from_param): Removed.
21993 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
21994 with stmt. Reverted back to use of load_from_unmodified_param.
21996 2017-01-23 Martin Jambor <mjambor@suse.cz>
21999 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
22000 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
22001 field a pointer to garbage collected vector, mark lattices and
22002 ipcp_orig_node with GTY((skip)).
22003 (ipa_get_param_count): Adjust to descriptors being a pointer.
22004 (ipa_get_param): Likewise.
22005 (ipa_get_type): Likewise.
22006 (ipa_get_param_move_cost): Likewise.
22007 (ipa_set_param_used): Likewise.
22008 (ipa_get_controlled_uses): Likewise.
22009 (ipa_set_controlled_uses): Likewise.
22010 (ipa_is_param_used): Likewise.
22011 (ipa_node_params_t): Move into garbage collector. New methods insert
22013 (ipa_node_params_sum): Annotate wth GTY(()).
22014 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
22016 (ipa_load_from_parm_agg): Adjust declaration.
22017 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
22018 * ipa-profile.c (ipa_profile): Likewise.
22019 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
22020 (ipa_populate_param_decls): Make descriptors parameter garbage
22022 (ipa_dump_param): Adjust to descriptors being a pointer.
22023 (ipa_alloc_node_params): Likewise.
22024 (ipa_initialize_node_params): Likewise.
22025 (load_from_param_1): Make descriptors parameter garbage collected.
22026 (load_from_unmodified_param): Likewise.
22027 (load_from_param): Likewise.
22028 (ipa_load_from_parm_agg): Likewise.
22029 (ipa_node_params::~ipa_node_params): Removed.
22030 (ipa_free_all_node_params): Remove call to delete operator.
22031 (ipa_node_params_t::insert): New.
22032 (ipa_node_params_t::remove): Likewise.
22033 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
22034 copy known_csts and known_contexts vectors.
22035 (ipa_read_node_info): Adjust to descriptors being a pointer.
22036 (ipcp_modif_dom_walker): Make m_descriptors field garbage
22038 (ipcp_transform_function): Make descriptors variable garbage
22041 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
22043 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
22044 * config/i386/avx512dqintrin.h: Ditto.
22045 * config/i386/avx512fintrin.h: Ditto.
22046 * config/i386/i386.c: Handle new builtins.
22047 * config/i386/i386-builtin.def: Add new builtins.
22048 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
22049 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
22051 2017-01-23 Jakub Jelinek <jakub@redhat.com>
22052 Martin Liska <mliska@suse.cz>
22054 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
22055 * asan.c (asan_expand_poison_ifn): Support stores and use
22056 appropriate ASAN report function.
22057 * internal-fn.c (expand_ASAN_POISON_USE): New function.
22058 * internal-fn.def (ASAN_POISON_USE): Declare.
22059 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
22060 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
22061 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
22062 ASAN_POISON calls w/o LHS.
22063 * tree-ssa.c (execute_update_addresses_taken): Create clobber
22064 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
22065 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
22066 * gimplify.c (asan_poison_variables): Add attribute
22067 use_after_scope_memory to variables that really needs to live
22069 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
22070 having the attribute.
22072 2017-01-23 Martin Liska <mliska@suse.cz>
22074 * asan.c (create_asan_shadow_var): New function.
22075 (asan_expand_poison_ifn): Likewise.
22076 * asan.h (asan_expand_poison_ifn): New declaration.
22077 * internal-fn.c (expand_ASAN_POISON): Likewise.
22078 * internal-fn.def (ASAN_POISON): New builtin.
22079 * sanopt.c (pass_sanopt::execute): Expand
22080 asan_expand_poison_ifn.
22081 * tree-inline.c (copy_decl_for_dup_finish): Make function
22083 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
22084 * tree-ssa.c (is_asan_mark_p): New function.
22085 (execute_update_addresses_taken): Rewrite local variables
22086 (identified just by use-after-scope as addressable) into SSA.
22088 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
22090 * doc/install.texi (Specific): opensource.apple.com uses https
22091 now. Remove trailing slash.
22093 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
22095 * README.Portability: Remove note on an Irix compatibility issue.
22097 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
22099 * gcov.c (INCLUDE_ALGORITHM): Define.
22100 (INCLUDE_VECTOR): Define.
22101 No longer include <vector> and <algorithm> directly.
22103 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
22105 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
22107 * doc/invoke.texi (Code Gen Options): Ditto.
22109 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
22112 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
22114 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
22116 rtl-optimization/79125
22117 * cprop.c (local_cprop_pass): Handle cases where we make an
22118 unconditional trap.
22120 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
22124 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
22125 read from, for big endian.
22127 2017-01-20 Jiong Wang <jiong.wang@arm.com>
22129 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
22130 register pauth builtins for LP64 only.
22132 2017-01-20 Marek Polacek <polacek@redhat.com>
22135 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
22138 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
22140 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
22142 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
22143 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
22144 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
22146 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22149 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
22150 in big-endian mode when they are not a single duplicated value.
22152 2017-01-20 Richard Biener <rguenther@suse.de>
22154 * BASE-VER: Bump to 7.0.1.
22156 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
22158 * omp-low.c (omplow_simd_context): New struct. Use it...
22159 (lower_rec_simd_input_clauses): ...here and...
22160 (lower_rec_input_clauses): ...here to hold common data. Adjust all
22161 references to idx, lane, max_vf, is_simt.
22163 2017-01-20 Graham Markall <graham.markall@embecosm.com>
22165 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
22168 2017-01-20 Martin Jambor <mjambor@suse.cz>
22170 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
22171 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
22173 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
22174 (GTFILES): Rename hsa.c to hsa-common.c.
22175 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
22176 * hsa-dump.c: Likewise.
22177 * hsa-gen.c: Likewise.
22178 * hsa-regalloc.c: Likewise.
22179 * ipa-hsa.c: Likewise.
22180 * omp-expand.c: Likewise.
22181 * omp-low.c: Likewise.
22182 * toplev.c: Likewise.
22184 2017-01-20 Marek Polacek <polacek@redhat.com>
22187 * doc/invoke.texi: Document -Wduplicated-branches.
22188 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
22189 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
22190 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
22191 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
22192 return 0 only when not OEP_LEXICOGRAPHIC.
22193 (fold_build_cleanup_point_expr): Use the expression
22194 location when building CLEANUP_POINT_EXPR.
22195 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
22196 * tree.c (add_expr): Handle error_mark_node.
22198 2017-01-20 Martin Liska <mliska@suse.cz>
22201 * tree-profile.c (init_ic_make_global_vars): Do not call
22203 (gimple_init_gcov_profiler): Likewise.
22205 2017-01-20 Martin Liska <mliska@suse.cz>
22208 * cgraph.h (maybe_create_reference): Remove argument and
22210 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
22212 * ipa-cp.c (create_specialized_node): Likewise.
22213 * symtab.c (symtab_node::maybe_create_reference): Handle
22214 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
22216 2017-01-20 Martin Liska <mliska@suse.cz>
22218 * read-rtl-function.c (function_reader::create_function): Use
22219 build_decl instread of build_decl_stat.
22221 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
22223 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
22224 * config/i386/avx512dqintrin.h: Ditto.
22225 * config/i386/avx512fintrin.h: Ditto.
22226 * config/i386/i386-builtin-types.def: Add new types.
22227 * config/i386/i386.c: Handle new types.
22228 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
22229 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
22230 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
22231 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
22232 (__builtin_ia32_kshiftridi): New.
22233 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
22235 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
22239 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
22240 define to rs6000_init_stack_protect_guard.
22241 (rs6000_init_stack_protect_guard): New function.
22243 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
22244 Yunqiang Su <yunqiang.su@imgtec.com>
22246 * config.gcc (supported_defaults): Add madd4.
22247 (with_madd4): Add validation.
22248 (all_defaults): Add madd4.
22249 * config/mips/mips.opt (mmadd4): New option.
22250 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
22252 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
22254 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
22255 (ISA_HAS_FUSED_MADD4): Likewise.
22256 * doc/invoke.texi (-mmadd4): Document the new option.
22257 * doc/install.texi (--with-madd4): Document the new option.
22259 2017-01-19 Jiong Wang <jiong.wang@arm.com>
22261 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
22262 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
22263 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
22264 (aarch64_init_pauth_hint_builtins): New.
22265 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
22266 (aarch64_expand_builtin): Expand new builtins.
22268 2017-01-19 Jiong Wang <jiong.wang@arm.com>
22270 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
22271 * combine-stack-adj.c (no_unhandled_cfa): Handle
22272 REG_CFA_TOGGLE_RA_MANGLE.
22273 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
22274 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
22275 info for return address signing.
22276 (aarch64_expand_epilogue): Likewise.
22278 2017-01-19 Jiong Wang <jiong.wang@arm.com>
22280 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
22281 * config/aarch64/aarch64-protos.h
22282 (aarch64_return_address_signing_enabled): New declaration.
22283 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
22285 (aarch64_expand_prologue): Sign return address before it's pushed onto
22287 (aarch64_expand_epilogue): Authenticate return address fetched from
22289 (aarch64_override_options): Sanity check for ILP32 and ISA level.
22290 (aarch64_attributes): New function attributes for "sign-return-address".
22291 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
22292 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
22293 ("*do_return"): Generate combined instructions according to key index.
22294 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
22295 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
22297 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
22298 * config/aarch64/aarch64.opt (msign-return-address=): New.
22299 * doc/extend.texi (AArch64 Function Attributes): Documents
22300 "sign-return-address=".
22301 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
22303 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
22305 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
22306 overall option summary.
22308 2017-01-19 Jiong Wang <jiong.wang@arm.com>
22310 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
22311 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
22312 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
22313 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
22315 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
22317 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
22318 -mpower9-minmax by default for -mcpu=power9.
22319 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
22320 128-bit floating point.
22322 2017-01-20 Alan Modra <amodra@gmail.com>
22324 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
22325 optimizing for size.
22327 2017-01-20 Alan Modra <amodra@gmail.com>
22330 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
22331 for strcmp and strncmp from corresponding builtin decl.
22333 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
22335 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
22336 instead of i386/rtems-64.h.
22337 * config/i386/rtems-64.h: Remove.
22339 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
22343 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
22345 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
22347 2017-01-19 Tamar Christina <tamar.christina@arm.com>
22349 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
22350 Change int to HOST_WIDE_INT.
22351 * config/aarch64/aarch64-protos.h
22352 (aarch64_simd_gen_const_vector_dup): Likewise.
22353 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
22355 2017-01-19 David Malcolm <dmalcolm@redhat.com>
22357 * langhooks-def.h (lhd_type_for_size): New decl.
22358 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
22359 * langhooks.c (lhd_type_for_size): New function, taken from
22362 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
22364 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
22365 define_bypass for CR latency.
22366 (power9-cracked-alu): Update bypass latency and remove power9-branch.
22367 (power9-alu2): Add define_bypass for CR latency.
22369 (power9-mul): Update insn latency.
22370 (power9-mul-compare): Update insn latency, bypass latency and remove
22373 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22375 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
22377 * config/aarch64/aarch64.md
22378 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
22379 aarch64_nopcrelative_literal_loads.
22380 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
22382 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
22384 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
22385 TARGET_LOONGSON_3A.
22386 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
22388 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
22391 * config.gcc (supported_defaults): Add lxc1-sxc1.
22392 (with_lxc1_sxc1): Add validation.
22393 (all_defaults): Add lxc1-sxc1.
22394 * config/mips/mips.opt (mlxc1-sxc1): New option.
22395 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
22397 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
22398 __mips_no_lxc1_sxc1.
22399 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
22400 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
22401 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
22403 2017-01-19 Richard Biener <rguenther@suse.de>
22405 PR tree-optimization/72488
22406 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
22407 sure to restore SSA info.
22408 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
22410 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
22412 PR rtl-optimization/79121
22413 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
22414 of the inner type when shifting an extended value.
22416 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
22419 * symtab.c (symtab_node::equal_address_to): Fix comparing of
22420 interposable aliases.
22422 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
22425 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
22426 Use the evmergelohi instruction.
22427 (mov_si<mode>_e500_subreg4_2_le): Likewise.
22428 (mov_sitf_e500_subreg8_2_be): Likewise.
22429 (mov_sitf_e500_subreg12_2_le): Likewise.
22430 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
22431 (mov_si<mode>_e500_subreg4_2_be): Likewise.
22432 (mov_sitf_e500_subreg8_2_le): Likewise.
22433 (mov_sitf_e500_subreg12_2_be): Likewise.
22435 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22437 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
22438 attribute from vecsimple to vecperm.
22439 (altivec_vbpermq2): Likewise.
22441 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22444 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
22446 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22447 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
22448 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
22449 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
22450 case where N arg is SIZE_MAX.
22451 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
22452 (cmpstrsi): Add pattern.
22454 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22456 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22457 __builtin_vec_revb builtins.
22458 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
22459 built-in functions to support generation of the ISA 3.0 XXBR<x>
22460 vector byte reverse instructions.
22461 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
22462 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
22463 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
22464 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
22465 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
22466 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
22467 (P9V_BUILTIN_VEC_REVB): Likewise.
22468 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
22469 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
22470 (p9_xxbrq_v16qi): Likewise.
22471 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
22472 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
22473 (p9_xxbrh_v8hi): Likewise.
22474 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
22475 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
22476 vec_revb built-in functions.
22478 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
22480 PR rtl-optimization/78952
22481 * config/i386/i386.md (any_extract): New code iterator.
22482 (*insvqi_2): Use any_extract for source operand.
22483 (*insvqi_3): Use any_shiftrt for source operand.
22485 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
22487 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
22489 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
22491 2017-01-18 Matthias Klose <doko@ubuntu.com>
22493 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
22495 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22497 * config/rs6000/altivec.h (vec_bperm): Change #define.
22498 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
22499 (altivec_vbpermq2): New define_insn.
22500 (altivec_vbpermd): Likewise.
22501 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
22502 function interface.
22503 (VBPERMD): Likewise.
22504 (VBPERM): New polymorphic function interface.
22505 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
22506 Add entries for P9V_BUILTIN_VEC_VBPERM.
22507 * doc/extend.texi: Add interfaces for vec_bperm.
22509 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22511 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
22512 first letter of error messages.
22513 (s390_resolve_overloaded_builtin): Likewise.
22514 * config/s390/s390.c (s390_expand_builtin): Likewise.
22515 (s390_invalid_arg_for_unprototyped_fn): Likewise.
22516 (s390_valid_target_attribute_inner_p): Likewise.
22517 * config/s390/s390.md ("tabort"): Likewise.
22519 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
22521 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
22522 (ISA_AVOID_DIV_HILO): New macro.
22523 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
22524 (ISA_HAS_DDIV): Likewise.
22526 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
22528 * doc/invoke.texi (fabi-version): Correct number of occurrences.
22530 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
22532 * doc/invoke.texi (fabi-version): Spelling fix.
22534 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
22537 * doc/invoke.texi (fabi-version): Mention mangling fix for
22540 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
22543 * doc/invoke.texi (fabi-version): Document discriminator mangling.
22545 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
22548 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
22549 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
22551 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
22553 (stack_protect_test): Ditto.
22554 * config/rs6000/rs6000.opt (mstack-protector-guard=,
22555 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
22557 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
22558 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
22559 -mstack-protector-guard-offset=.
22560 (RS/6000 and PowerPC Options): Ditto.
22562 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
22564 * config/i386/i386.h (MASK_CLASS_P): New define.
22565 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
22566 there are no registers from different register sets also when
22567 mask registers are used. Update function comment.
22568 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
22569 to (*k/*r) and (*k/*km) alternatives.
22571 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
22573 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
22574 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
22575 (EH_RETURN_HANDLER_RTX): New define.
22576 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
22577 Force frame pointer in EH return functions.
22578 (aarch64_expand_epilogue): Add barrier for eh_return.
22579 (aarch64_final_eh_return_addr): Remove.
22580 (aarch64_eh_return_handler_rtx): New function.
22581 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
22583 (aarch64_eh_return_handler_rtx): New prototype.
22585 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22587 * config/rs6000/altivec.h (vec_rlmi): New #define.
22588 (vec_vrlnm): Likewise.
22589 (vec_rlnm): Likewise.
22590 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
22591 (UNSPEC_VRLNM): Likewise.
22592 (VIlong): New mode iterator.
22593 (altivec_vrl<VI_char>mi): New define_insn.
22594 (altivec_vrl<VI_char>nm): Likewise.
22595 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
22597 (VRLDNM): Likewise.
22598 (RLNM): New polymorphic function entry.
22599 (VRLWMI): New monomorphic function entry.
22600 (VRLDMI): Likewise.
22601 (RLMI): New polymorphic function entry.
22602 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
22603 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
22604 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
22607 2017-01-17 Jakub Jelinek <jakub@redhat.com>
22610 * dwarf2out.c (field_byte_offset): Restore the
22611 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
22612 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
22613 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
22616 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
22619 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
22621 2017-01-17 Jakub Jelinek <jakub@redhat.com>
22624 * dwarf2out.c (add_data_member_location_attribute): For constant
22625 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
22626 instead of DW_AT_data_member_location, DW_AT_bit_offset and
22627 DW_AT_byte_size attributes.
22629 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
22631 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
22632 after forcing to constant memory when the code model is medium.
22634 2017-01-17 Julia Koval <julia.koval@intel.com>
22637 * config/i386/avx512fintrin.h
22638 (_mm512_i32gather_ps): Change __addr type to void const*.
22639 (_mm512_mask_i32gather_ps): Ditto.
22640 (_mm512_i32gather_pd): Ditto.
22641 (_mm512_mask_i32gather_pd): Ditto.
22642 (_mm512_i64gather_ps): Ditto.
22643 (_mm512_mask_i64gather_ps): Ditto.
22644 (_mm512_i64gather_pd): Ditto.
22645 (_mm512_mask_i64gather_pd): Ditto.
22646 (_mm512_i32gather_epi32): Ditto.
22647 (_mm512_mask_i32gather_epi32): Ditto.
22648 (_mm512_i32gather_epi64): Ditto.
22649 (_mm512_mask_i32gather_epi64): Ditto.
22650 (_mm512_i64gather_epi32): Ditto.
22651 (_mm512_mask_i64gather_epi32): Ditto.
22652 (_mm512_i64gather_epi64): Ditto.
22653 (_mm512_mask_i64gather_epi64): Ditto.
22654 (_mm512_i32scatter_ps): Change __addr type to void*.
22655 (_mm512_mask_i32scatter_ps): Ditto.
22656 (_mm512_i32scatter_pd): Ditto.
22657 (_mm512_mask_i32scatter_pd): Ditto.
22658 (_mm512_i64scatter_ps): Ditto.
22659 (_mm512_mask_i64scatter_ps): Ditto.
22660 (_mm512_i64scatter_pd): Ditto.
22661 (_mm512_mask_i64scatter_pd): Ditto.
22662 (_mm512_i32scatter_epi32): Ditto.
22663 (_mm512_mask_i32scatter_epi32): Ditto.
22664 (_mm512_i32scatter_epi64): Ditto.
22665 (_mm512_mask_i32scatter_epi64): Ditto.
22666 (_mm512_i64scatter_epi32): Ditto.
22667 (_mm512_mask_i64scatter_epi32): Ditto.
22668 (_mm512_i64scatter_epi64): Ditto.
22669 (_mm512_mask_i64scatter_epi64): Ditto.
22670 * config/i386/avx512pfintrin.h
22671 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
22672 (_mm512_mask_prefetch_i32gather_ps): Ditto.
22673 (_mm512_mask_prefetch_i64gather_pd): Ditto.
22674 (_mm512_mask_prefetch_i64gather_ps): Ditto.
22675 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
22676 (_mm512_prefetch_i32scatter_ps): Ditto.
22677 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
22678 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
22679 (_mm512_prefetch_i64scatter_pd): Ditto.
22680 (_mm512_prefetch_i64scatter_ps): Ditto.
22681 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
22682 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
22683 * config/i386/avx512vlintrin.h
22684 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
22685 (_mm_mmask_i32gather_ps): Ditto.
22686 (_mm256_mmask_i32gather_pd): Ditto.
22687 (_mm_mmask_i32gather_pd): Ditto.
22688 (_mm256_mmask_i64gather_ps): Ditto.
22689 (_mm_mmask_i64gather_ps): Ditto.
22690 (_mm256_mmask_i64gather_pd): Ditto.
22691 (_mm_mmask_i64gather_pd): Ditto.
22692 (_mm256_mmask_i32gather_epi32): Ditto.
22693 (_mm_mmask_i32gather_epi32): Ditto.
22694 (_mm256_mmask_i32gather_epi64): Ditto.
22695 (_mm_mmask_i32gather_epi64): Ditto.
22696 (_mm256_mmask_i64gather_epi32): Ditto.
22697 (_mm_mmask_i64gather_epi32): Ditto.
22698 (_mm256_mmask_i64gather_epi64): Ditto.
22699 (_mm_mmask_i64gather_epi64): Ditto.
22700 (_mm256_i32scatter_ps): Change __addr type to void*.
22701 (_mm256_mask_i32scatter_ps): Ditto.
22702 (_mm_i32scatter_ps): Ditto.
22703 (_mm_mask_i32scatter_ps): Ditto.
22704 (_mm256_i32scatter_pd): Ditto.
22705 (_mm256_mask_i32scatter_pd): Ditto.
22706 (_mm_i32scatter_pd): Ditto.
22707 (_mm_mask_i32scatter_pd): Ditto.
22708 (_mm256_i64scatter_ps): Ditto.
22709 (_mm256_mask_i64scatter_ps): Ditto.
22710 (_mm_i64scatter_ps): Ditto.
22711 (_mm_mask_i64scatter_ps): Ditto.
22712 (_mm256_i64scatter_pd): Ditto.
22713 (_mm256_mask_i64scatter_pd): Ditto.
22714 (_mm_i64scatter_pd): Ditto.
22715 (_mm_mask_i64scatter_pd): Ditto.
22716 (_mm256_i32scatter_epi32): Ditto.
22717 (_mm256_mask_i32scatter_epi32): Ditto.
22718 (_mm_i32scatter_epi32): Ditto.
22719 (_mm_mask_i32scatter_epi32): Ditto.
22720 (_mm256_i32scatter_epi64): Ditto.
22721 (_mm256_mask_i32scatter_epi64): Ditto.
22722 (_mm_i32scatter_epi64): Ditto.
22723 (_mm_mask_i32scatter_epi64): Ditto.
22724 (_mm256_i64scatter_epi32): Ditto.
22725 (_mm256_mask_i64scatter_epi32): Ditto.
22726 (_mm_i64scatter_epi32): Ditto.
22727 (_mm_mask_i64scatter_epi32): Ditto.
22728 (_mm256_i64scatter_epi64): Ditto.
22729 (_mm256_mask_i64scatter_epi64): Ditto.
22730 (_mm_i64scatter_epi64): Ditto.
22731 (_mm_mask_i64scatter_epi64): Ditto.
22732 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
22733 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
22734 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
22735 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
22736 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
22737 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
22738 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
22739 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
22740 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
22741 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
22742 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
22743 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
22744 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
22745 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
22746 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
22747 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
22748 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
22749 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
22750 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
22751 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
22752 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
22753 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
22754 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
22755 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
22756 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
22757 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
22758 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
22759 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
22760 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
22761 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
22762 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
22763 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
22764 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
22765 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
22766 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
22767 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
22768 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
22769 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
22770 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
22771 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
22772 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
22773 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
22774 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
22775 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
22776 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
22777 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
22778 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
22779 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
22780 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
22781 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
22782 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
22783 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
22784 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
22785 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
22786 definitions accordingly.
22788 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
22789 Kuan-Lin Chen <kuanlinchentw@gmail.com>
22792 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
22795 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
22798 * ira-conflicts.c (ira_build_conflicts): Update total conflict
22799 hard regs for inner regno.
22801 2017-01-17 Martin Liska <mliska@suse.cz>
22804 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
22805 assumption and add comment.
22807 2017-01-17 Nathan Sidwell <nathan@acm.org>
22809 * ipa-visibility.c (localize_node): New function, broken out of ...
22810 (function_and_variable_visibility): ... here. Call it.
22812 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
22814 PR middle-end/77445
22815 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
22816 correctly set frequency of oudgoing edge.
22817 (duplicate_thread_path): Fix profile updating.
22819 2017-01-17 Jakub Jelinek <jakub@redhat.com>
22822 * configure.ac: Add GCC_BASE_VER.
22823 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
22824 version from BASE-VER file.
22825 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
22826 (gcc.o): Depend on $(BASEVER).
22827 * common.opt (dumpfullversion): New option.
22828 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
22829 * doc/invoke.texi: Document -dumpfullversion.
22830 * doc/install.texi: Document --with-gcc-major-version-only.
22831 * configure: Regenerated.
22833 2017-01-17 Richard Biener <rguenther@suse.de>
22835 PR tree-optimization/71433
22836 * tree-vrp.c (register_new_assert_for): Merge same asserts
22837 on all incoming edges.
22838 (process_assert_insertions_for): Handle insertions at the
22841 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
22843 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
22844 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
22846 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
22849 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
22852 2017-01-17 Alan Modra <amodra@gmail.com>
22855 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
22856 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
22857 symbolic stack limit when pic.
22859 2017-01-16 Martin Sebor <msebor@redhat.com>
22861 PR tree-optimization/78608
22862 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
22864 2017-01-16 Jeff Law <law@redhat.com>
22867 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
22868 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
22869 for several include directories that may be relative to sysroot.
22870 * config/i386/x-mingw32 (gplus_includedir): Define.
22871 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
22872 (native_system_includedir): Likewise.
22873 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
22874 override if TARGET_SYSTEM_ROOT is defined.
22875 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
22877 PR tree-optimization/79090
22878 PR tree-optimization/33562
22879 PR tree-optimization/61912
22880 PR tree-optimization/77485
22881 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
22882 and computed trims into the dump file.
22884 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
22886 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
22888 2017-01-16 Jakub Jelinek <jakub@redhat.com>
22891 * gimplify.c (gimplify_init_constructor): If want_value and
22892 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
22896 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
22897 sequence. Formatting fixes.
22898 (doloop_optimize): Formatting fixes.
22901 * gcc.c (debug_level_greater_than_spec_func): New function.
22902 (static_spec_functions): Add debug-level-gt spec function.
22903 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
22905 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
22906 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
22907 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
22908 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
22909 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
22910 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
22912 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
22914 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
22915 QImode fixups to general and mask registers only.
22917 2017-01-16 Carl Love <cel@us.ibm.com>
22919 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
22920 for built-in functions
22921 vector signed char vec_nabs (vector signed char)
22922 vector signed short vec_nabs (vector signed short)
22923 vector signed int vec_nabs (vector signed int)
22924 vector signed long long vec_nabs (vector signed long long)
22925 vector float vec_nabs (vector float)
22926 vector double vec_nabs (vector double)
22927 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
22929 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
22930 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
22931 * doc/extend.texi: Update the documentation file for the new built-in
22934 2017-01-16 Martin Sebor <msebor@redhat.com>
22936 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
22939 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22941 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
22942 UNSPEC_VSX__XXSPLTD to require special splat handling.
22944 2017-01-16 David Malcolm <dmalcolm@redhat.com>
22947 * system.h: Poison strndup.
22949 2017-01-16 Alan Modra <amodra@gmail.com>
22952 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
22955 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
22957 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
22959 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
22961 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
22962 call recog here. Assert that INSN_CODE (insn) is non-negative.
22964 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
22967 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
22969 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
22970 in the currently scheduled RTL fragment.
22972 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
22974 PR rtl-optimization/78751
22975 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
22978 2017-01-14 Jeff Law <law@redhat.com>
22980 PR tree-optimization/79090
22981 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
22982 variable length stores.
22983 (compute_trims): Delete dead assignment to *trim_tail.
22984 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
22987 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
22989 PR rtl-optimization/78626
22990 PR rtl-optimization/78727
22991 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
22992 of a block, and split such blocks after everything else is finished.
22994 2017-01-14 Alan Modra <amodra@gmail.com>
22997 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
22998 target legitimate_combined_insn.
22999 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
23000 (rs6000_legitimate_combined_insn): New function.
23001 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
23003 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
23004 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
23005 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
23007 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
23009 * doc/frontends.texi (G++ and GCC): Remove references to Java.
23011 2017-01-13 Jeff Law <law@redhat.com>
23013 PR tree-optimization/33562
23014 PR tree-optimization/61912
23015 PR tree-optimization/77485
23016 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
23018 (delete_dead_assignment): Likewise.
23019 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
23020 statement to delete_dead_call and delete_dead_assignment.
23022 2017-01-13 David Malcolm <dmalcolm@redhat.com>
23025 * substring-locations.c (format_warning_va): Strengthen case 1 so
23026 that both endpoints of the substring must be within the format
23027 range for just the substring to be printed.
23029 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
23031 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
23032 * config/i386/i386.c (ix86_target_string): Add missing options
23033 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
23034 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
23035 flags_other and ix86_target_other to flags2_other. Display unknown
23037 (ix86_valid_target_attribute_inner_p): Add missing options and
23038 reorder options by implied ISAs, as in ix86_target_string.
23040 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
23042 * hash-table.h (hash_table::too_empty_p): New function.
23043 (hash_table::expand): Use it.
23044 (hash_table::traverse): Likewise.
23045 (hash_table::empty_slot): Use sizeof (value_type) instead of
23046 sizeof (PTR) to convert bytes to elements. Shrink the table
23047 if the current size is excessive for the current number of
23050 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
23052 * ira-costs.c (record_reg_classes): Break from the inner loop
23053 early once alt_fail is known to be true. Update outer loop
23054 handling accordingly.
23056 2017-01-13 Jeff Law <law@redhat.com>
23058 * tree-ssa-dse.c (decrement_count): New function.
23059 (increment_start_addr, maybe_trim_memstar_call): Likewise.
23060 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
23061 when we know the partially dead statement is a mem* function.
23063 PR tree-optimization/61912
23064 PR tree-optimization/77485
23065 * tree-ssa-dse.c: Include expr.h.
23066 (maybe_trim_constructor_store): New function.
23067 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
23069 PR tree-optimization/33562
23070 PR tree-optimization/61912
23071 PR tree-optimization/77485
23072 * doc/invoke.texi: Document new dse-max-object-size param.
23073 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
23074 * tree-ssa-dse.c: Include params.h.
23075 (dse_store_status): New enum.
23076 (initialize_ao_ref_for_dse): New, partially extracted from
23078 (valid_ao_ref_for_dse, normalize_ref): New.
23079 (setup_live_bytes_from_ref, compute_trims): Likewise.
23080 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
23081 (maybe_trim_partially_dead_store): Likewise.
23082 (maybe_trim_complex_store): Likewise.
23083 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
23084 Track what bytes live from the original store. Return tri-state
23085 for dead, partially dead or live.
23086 (dse_dom_walker): Add constructor, destructor and new private members.
23087 (delete_dead_call, delete_dead_assignment): New extracted from
23089 (dse_optimize_stmt): Make a member of dse_dom_walker.
23090 Use initialize_ao_ref_for_dse.
23092 PR tree-optimization/33562
23093 PR tree-optimization/61912
23094 PR tree-optimization/77485
23095 * sbitmap.h (bitmap_count_bits): Prototype.
23096 (bitmap_clear_range, bitmap_set_range): Likewise.
23097 * sbitmap.c (bitmap_clear_range): New function.
23098 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
23100 2017-01-13 Martin Liska <mliska@suse.cz>
23103 * function.c (set_cfun): Add new argument force.
23104 * function.h (set_cfun): Likewise.
23105 * ipa-inline-transform.c (inline_call): Use the function when
23106 strict alising from is dropped for function we inline to.
23108 2017-01-13 Richard Biener <rguenther@suse.de>
23110 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
23111 for dumping GIMPLE INTEGER_CSTs.
23113 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23115 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
23116 to 201112L since C++17.
23118 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
23121 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
23122 if -fsanitize=kernel-address is present.
23124 2017-01-13 Richard Biener <rguenther@suse.de>
23126 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
23127 as _Literal ( type ) number in case usual suffixes do not
23128 preserve all information.
23130 2017-01-13 Richard Biener <rguenther@suse.de>
23132 PR tree-optimization/77283
23133 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
23134 and ssa-iterators.h.
23135 (is_feasible_trace): Implement a cost model based on joiner
23138 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
23141 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
23142 char or short to __float128/_Float128 directly.
23144 2017-01-12 Martin Sebor <msebor@redhat.com>
23146 to -Wformat-overflow.
23147 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
23148 (min_bytes_remaining): Same.
23149 (get_string_length): Same.
23150 (format_string): Same.
23151 (format_directive): Same.
23153 (pass_sprintf_length::handle_gimple_call): Same.
23155 2017-01-12 Jakub Jelinek <jakub@redhat.com>
23157 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
23158 info.nowrite calls with no lhs that can't throw. Return bool
23159 whether gsi_remove has been called or not.
23160 (pass_sprintf_length::handle_gimple_call): Return bool whether
23161 try_substitute_return_value called gsi_remove. Formatting fix.
23162 (pass_sprintf_length::execute): Don't use gsi_remove if
23163 handle_gimple_call returned true.
23166 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
23167 be removed due to side-effects, don't remove following barrier nor
23168 turn the successor edge into fallthru edge.
23170 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23173 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
23174 element-reversing loads and stores as not swappable.
23176 2017-01-12 Nathan Sidwell <nathan@acm.org>
23177 Nicolai Stange <nicstange@gmail.com>
23179 * combine.c (try_combine): Don't ignore result of overlap checking
23180 loop. Combine overlap & asm check into single loop.
23182 2017-01-12 Richard Biener <rguenther@suse.de>
23184 * tree-pretty-print.c (dump_generic_node): Provide -gimple
23185 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
23187 2017-01-12 Richard Biener <rguenther@suse.de>
23189 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
23190 and TS_TARGET_OPTION directly derive from TS_BASE.
23191 * tree-core.h (tree_optimization_option): Derive from tree_base.
23192 (tree_target_option): Likewise.
23194 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
23196 * config/i386/i386.c (memory_address_length): Increase len
23197 only when rip_relative_addr_p returns false.
23199 2017-01-11 Julia Koval <julia.koval@intel.com>
23201 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
23202 (OPTION_MASK_ISA_SGX_SET): New.
23203 (ix86_handle_option): Handle OPT_msgx.
23204 * config.gcc: Added sgxintrin.h.
23205 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
23206 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
23207 * config/i386/i386.c (ix86_target_string): Add -msgx.
23209 (ix86_option_override_internal): Handle new options.
23210 (ix86_valid_target_attribute_inner_p): Add sgx.
23211 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
23212 * config/i386/i386.opt: Add msgx.
23213 * config/i386/sgxintrin.h: New file.
23214 * config/i386/x86intrin.h: Add sgxintrin.h.
23216 2017-01-11 Jakub Jelinek <jakub@redhat.com>
23219 * fold-const.c (maybe_nonzero_address): Return 1 for function
23221 (tree_single_nonzero_warnv_p): Don't handle function local objects
23225 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
23228 2017-01-11 David Malcolm <dmalcolm@redhat.com>
23231 * opts.c: Include "spellcheck.h"
23232 (struct string_fragment): New struct.
23233 (struct edit_distance_traits<const string_fragment &>): New
23235 (get_closest_sanitizer_option): New function.
23236 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
23238 2017-01-11 Jakub Jelinek <jakub@redhat.com>
23240 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
23242 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
23243 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
23244 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
23245 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
23246 for initial die_offset if dwarf_split_debug_info.
23247 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
23248 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
23250 (output_skeleton_debug_sections): Formatting fix. Use
23251 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
23252 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
23254 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
23256 * config/arm/cortex-a53.md: Add bypasses for
23257 cortex_a53_r2f_cvt.
23258 (cortex_a53_r2f): Only use for transfers.
23259 (cortex_a53_f2r): Likewise.
23260 (cortex_a53_r2f_cvt): Add reservation for conversions.
23261 (cortex_a53_f2r_cvt): Likewise.
23263 2017-01-11 Tamar Christina <tamar.christina@arm.com>
23265 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
23266 to all inlined functions, change static to extern.
23268 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
23271 * config/arm/arm.c (legitimize_pic_address): Handle reference to
23273 (arm_assemble_integer): Likewise.
23275 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
23277 * config.gcc: Use new awk script to check CPU, FPU and architecture
23278 parameters for --with-... options.
23279 * config/arm/parsecpu.awk: New file
23280 * config/arm/arm-cpus.in: New file.
23281 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
23283 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
23285 * config/arm/t-arm: Update dependency rules.
23286 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
23287 of processing .def files.
23288 * config/arm/genopt.sh: Deleted.
23289 * config/arm/gentune.sh: Deleted.
23290 * config/arm/arm-cores.def: Deleted.
23291 * config/arm/arm-arches.def: Deleted.
23292 * config/arm/arm-fpus.def: Deleted.
23293 * config/arm/arm-tune.md: Regenerated.
23294 * config/arm/arm-tables.opt: Regenerated.
23295 * config/arm/arm-cpu.h: New generated file.
23296 * config/arm/arm-cpu-data.h: New generated file.
23297 * config/arm/arm-cpu-cdata.h: New generated file.
23299 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
23302 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
23304 (input_varpool_node): Unpack dynamically_initialized bit.
23306 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
23308 PR rtl-optimization/79032
23309 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
23310 the alignment of the adjusted memory reference against that of MODE,
23311 instead of the alignment of the original memory reference.
23313 2017-01-11 Martin Jambor <mjambor@suse.cz>
23315 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
23317 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
23318 decorated functions.
23320 2017-01-11 Richard Biener <rguenther@suse.de>
23322 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
23323 set range/nonnull info for PHI results. Do not set it on
23324 stmts marked for removal.
23326 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
23328 * expr.c (store_field): In the bitfield case, fetch the return value
23329 from the registers before applying a single big-endian adjustment.
23330 Always do a final load for a BLKmode value not larger than a word.
23332 2017-01-10 David Malcolm <dmalcolm@redhat.com>
23335 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
23336 that we correctly handle column numbers greater than
23337 LINE_MAP_MAX_COLUMN_NUMBER.
23339 2017-01-10 Martin Sebor <msebor@redhat.com>
23341 PR middle-end/78245
23342 * gimple-ssa-sprintf.c (get_destination_size): Call
23343 {init,fini}object_sizes.
23344 * tree-object-size.c (addr_object_size): Adjust.
23345 (pass_through_call): Adjust.
23346 (pass_object_sizes::execute): Adjust.
23347 * tree-object-size.h (fini_object_sizes): Declare.
23349 2017-01-10 Martin Sebor <msebor@redhat.com>
23351 PR tree-optimization/78775
23352 * builtins.c (get_size_range): Move...
23353 * calls.c: ...to here.
23354 (alloc_max_size): Accept zero argument.
23355 (operand_signed_p): Remove.
23356 (maybe_warn_alloc_args_overflow): Call get_size_range.
23357 * calls.h (get_size_range): Declare.
23359 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
23361 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
23362 from TI's devices.csv file as of September 2016.
23363 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
23365 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
23367 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
23368 * doc/invoke.texi: Likewise.
23369 * doc/md.texi: Likewise.
23370 * doc/objc.texi: Likewise.
23372 2017-01-10 Joshua Conner <joshconner@google.com>
23374 * config/arm/fuchsia-elf.h: New file.
23375 * config/fuchsia.h: New file.
23376 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
23377 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
23379 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
23381 2016-01-10 Richard Biener <rguenther@suse.de>
23383 PR tree-optimization/79034
23384 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
23385 Propagate out degenerate PHIs in the joiner.
23387 2017-01-10 Martin Liska <mliska@suse.cz>
23389 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
23390 (sort_congruence_classes_by_decl_uid): Likewise.
23391 (sort_congruence_class_groups_by_decl_uid): Likewise.
23392 (sem_item_optimizer::merge_classes): Sort class, groups in these
23393 classes and members in the groups by DECL_UID of declarations.
23394 This would make merge operations stable.
23396 2017-01-10 Martin Liska <mliska@suse.cz>
23398 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
23399 usage of m_classes_vec.
23400 (sem_item_optimizer::~sem_item_optimizer): Likewise.
23401 (sem_item_optimizer::get_group_by_hash): Likewise.
23402 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
23403 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
23404 (sem_item_optimizer::verify_classes): Likewise.
23405 (sem_item_optimizer::process_cong_reduction): Likewise.
23406 (sem_item_optimizer::dump_cong_classes): Likewise.
23407 (sem_item_optimizer::merge_classes): Likewise.
23408 * ipa-icf.h (congruence_class_hash): Rename from
23409 congruence_class_group_hash. Remove declaration of m_classes_vec.
23411 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
23413 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
23414 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
23415 * config.gcc: Add avx512vpopcntdqintrin.h.
23416 * config/i386/avx512vpopcntdqintrin.h: New.
23417 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
23418 * config/i386/i386-builtin-types.def: Add new types.
23419 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
23420 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
23421 __builtin_ia32_vpopcountq_v8di_mask): New.
23422 * config/i386/i386-c.c (ix86_target_macros_internal): Define
23423 __AVX512VPOPCNTDQ__.
23424 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
23425 (PTA_AVX512VPOPCNTDQ): Define.
23426 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
23427 TARGET_AVX512VPOPCNTDQ_P): Define.
23428 * config/i386/i386.opt: Add mavx512vpopcntdq.
23429 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
23430 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
23432 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23434 PR middle-end/77484
23435 * predict.def (PRED_CALL): Set to 67.
23437 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
23439 * expr.c (store_field): In the bitfield case, if the value comes from
23440 a function call and is of an aggregate type returned in registers, do
23441 not modify the field mode; extract the value in all cases if the mode
23442 is BLKmode and the size is not larger than a word.
23444 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
23447 * config/i386/cpuid.h: Fix undefined behavior.
23449 2017-01-04 Jeff Law <law@redhat.com>
23451 PR tree-optimization/79007
23452 PR tree-optimization/67955
23453 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
23454 conservative for pt.null when flag_non_call_exceptions is on.
23456 2017-01-09 Jakub Jelinek <jakub@redhat.com>
23458 PR translation/79019
23459 PR translation/79020
23460 * params.def (PARAM_INLINE_MIN_SPEEDUP,
23461 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
23462 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
23464 * config/avr/avr.opt (maccumulate-args): Likewise.
23465 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
23466 * common.opt (freport-bug): Likewise.
23467 * cif-code.def (CIF_FINAL_ERROR): Likewise.
23468 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
23469 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
23470 translatable string.
23471 * config/i386/i386.c (function_value_32): Likewise.
23472 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
23473 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
23475 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
23476 * common/config/msp430/msp430-common.c (msp430_handle_option):
23478 * symtab.c (symtab_node::verify_base): Likewise.
23479 * opts.c (set_debug_level): Likewise.
23480 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
23481 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
23482 missing whitespace to translatable strings.
23483 * config/avr/avr.md (bswapsi2): Fix typo in comment.
23484 * config/sh/superh.h: Likewise.
23485 * config/i386/xopintrin.h: Likewise.
23486 * config/i386/znver1.md: Likewise.
23487 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
23488 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
23489 * double-int.h (struct double_int): Likewise.
23490 * double-int.c (div_and_round_double): Likewise.
23491 * wide-int.cc: Likewise.
23492 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
23493 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
23494 * cfgcleanup.c (crossjumps_occured): Renamed to ...
23495 (crossjumps_occurred): ... this.
23496 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
23499 PR tree-optimization/78899
23500 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
23501 returning bool return struct loop *, NULL for failure and the new
23503 (versionable_outer_loop_p): Don't version outer loop if it has
23504 dont_vectorized bit set.
23505 (tree_if_conversion): When versioning outer loop, ensure
23506 tree_if_conversion is performed also on the inner loop of the
23507 non-vectorizable outer loop copy.
23508 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
23509 LOOP_VECTORIZED in inner loop of the scalar outer loop and
23510 prevent vectorization of it.
23511 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
23512 the outer loop vectorization of the non-scalar version is attempted
23513 before vectorization of the inner loop in scalar version. If
23514 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
23515 vectorization of its inner loop.
23516 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
23517 has 2 inner loops, rename also on edges from bb whose single pred
23518 is outer_loop->header. Fix typo in function comment.
23520 2017-01-09 Martin Sebor <msebor@redhat.com>
23523 * asan.c (asan_emit_stack_protection): Increase local buffer size
23524 to avoid snprintf truncation warning.
23526 2017-01-09 Andrew Pinski <apinski@cavium.com>
23528 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
23529 to reference thunderx2t99 for the tuning structure
23530 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
23532 (thunderx2t99_extra_costs): This.
23533 * config/aarch64/aarch64-tune.md: Regenerate.
23534 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
23535 (vulcan_addrcost_table): This.
23536 (vulcan_regmove_cost): Rename to ...
23537 (thunderx2t99_regmove_cost): This.
23538 (vulcan_vector_cost): Rename to ...
23539 (thunderx2t99_vector_cost): this.
23540 (vulcan_branch_cost): Rename to ...
23541 (thunderx2t99_branch_cost): This.
23542 (vulcan_tunings): Rename to ...
23543 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
23544 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
23546 2017-01-09 Martin Jambor <mjambor@suse.cz>
23550 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
23551 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
23552 (propagate_vr_accross_jump_function): Use the above function for all
23553 value range computations for pass-through jump functions and type
23554 converasion from explicit value range values.
23555 (ipcp_propagate_stage): Do not attempt to deduce types of formal
23556 parameters from TYPE_ARG_TYPES.
23557 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
23558 (ipa_write_node_info): Stream type of the actual argument.
23559 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
23561 2017-01-09 Martin Liska <mliska@suse.cz>
23564 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
23565 (lookup_compiler): Do not show error message with have_E.
23567 2017-01-09 Jakub Jelinek <jakub@redhat.com>
23569 PR tree-optimization/78938
23570 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
23571 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
23572 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
23573 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
23576 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23578 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
23581 2017-01-09 Richard Biener <rguenther@suse.de>
23583 PR tree-optimization/78997
23584 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
23585 name condition properly.
23587 2017-01-09 Richard Biener <rguenther@suse.de>
23590 * dwarf2out.c (is_cxx): New overload with context.
23591 (is_naming_typedef_decl): Use it.
23593 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
23595 * invoke.texi (Option Summary): Correct spacing in option lists
23596 and add line breaks to fix over-long lines.
23598 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
23600 PR middle-end/17660
23602 * extend.texi (Common Variable Attributes): Add xref to GCC
23603 Internals manual to explain mode attribute keywords.
23605 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
23608 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
23609 and Preprocessor Options.
23610 (Options for Linking): Document -pthread here....
23611 (RS/6000 and PowerPC Options): ...not here.
23612 (Solaris 2 Options): ...or here.
23613 * doc/cppopts.texi: Document -pthread.
23615 2017-01-08 Martin Sebor <msebor@redhat.com>
23617 PR middle-end/77708
23618 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
23619 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
23620 New member functions.
23621 (format_directive): Used them.
23623 (pass_sprintf_length::handle_gimple_call): Same.
23624 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
23625 to avoid truncation for any argument.
23626 (extract_affine_mul): Same.
23627 * tree.c (get_file_function_name): Same.
23629 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23631 PR middle-end/77484
23632 * predict.def (PRED_INDIR_CALL): Set to 86.
23634 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
23636 PR preprocessor/54124
23637 * doc/cppopts.texi: Reformat -d subtable to list the full name
23638 of the options. Add cross-reference to the docs for the general
23639 compiler -d options.
23640 * doc/invoke.texi (Developer Options): Add cross-reference to the
23641 preprocessor-specific -d option documentation.
23643 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
23645 PR preprocessor/13498
23646 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
23647 redudant material, and reflect new command-line options.
23648 (System Headers): Likewise.
23650 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
23652 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
23653 -isystem, and -idirafter. Copy-edit.
23654 * doc/cppopts.texi: Copy-edit. Remove contradiction about
23655 default for -ftrack-macro-expansion. Delete obsolete and
23656 badly-formatted implementation details about -fdebug-cpp output.
23657 * doc/cppwarnopts.texi: Copy-edit.
23659 2017-01-07 David Malcolm <dmalcolm@redhat.com>
23662 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
23663 that the transition from a max line width >= 1<<10 to narrower
23664 lines works correctly.
23666 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
23668 * doc/options.texi (PerFunction): New.
23669 * opt-functions.awk (switch_flags): Map both Optimization and
23670 PerFunction to CL_OPTIMIZATION.
23671 * opth-gen.awk: Test for PerFunction flag along with
23673 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
23674 it only when the latter is present. Skip those that don't in
23675 the hash function generator.
23676 * common.opt (fvar-tracking): Mark as PerFunction instead of
23678 (fvar-tracking-assignments): Likewise.
23679 (fvar-tracking-assignments-toggle): Likewise.
23680 (fvar-tracking-uninit): Likewise.
23682 2017-01-07 Jakub Jelinek <jakub@redhat.com>
23684 PR translation/79018
23685 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
23688 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
23691 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
23692 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
23693 TARGET_LONG_JUMP_TABLE_OFFSETS.
23694 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
23695 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
23696 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
23697 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
23698 * config/m68k/m68k.md (tablejump expander): Likewise.
23699 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
23700 TARGET_LONG_JUMP_TABLE_OFFSETS.
23701 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
23702 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
23704 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
23705 David Holsgrove <david.holsgrove@xilinx.com>
23707 * common/config/microblaze/microblaze-common.c
23708 (TARGET_EXCEPT_UNWIND_INFO): Remove.
23709 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
23711 * config/microblaze/microblaze.c (microblaze_must_save_register)
23712 (microblaze_expand_epilogue, microblaze_return_addr): Handle
23714 (microblaze_eh_return): New function.
23715 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
23716 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
23717 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
23718 * config/microblaze/microblaze.md (eh_return): New pattern.
23720 2017-01-06 Jakub Jelinek <jakub@redhat.com>
23722 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
23723 GCC_DIAGNOSTIC_STRINGIFY): Define.
23725 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
23727 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23729 * config/arm/arm.md (<mcrr>): New.
23731 * config/arm/arm.c (arm_arch5te): New.
23732 (arm_option_override): Set arm_arch5te.
23733 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
23735 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
23736 (arm_mcrr_qualifiers): ... this. New.
23737 (MRRC_QUALIFIERS): Define to...
23738 (arm_mrrc_qualifiers): ... this. New.
23739 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
23741 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
23742 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
23743 (MRRCI, mrrc, MRRC): New.
23744 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
23745 VUNSPEC_MRRC2): New.
23747 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23749 * config/arm/arm.md (<mcr>): New.
23751 * config/arm/arm.c (arm_coproc_builtin_available): Add
23752 support for mcr, mrc, mcr2 and mrc2.
23753 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
23754 (arm_mcr_qualifiers): ... this. New.
23755 (MRC_QUALIFIERS): Define to ...
23756 (arm_mrc_qualifiers): ... this. New.
23757 (MCR_QUALIFIERS): Define to ...
23758 (arm_mcr_qualifiers): ... this. New.
23759 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
23761 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
23762 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
23763 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
23764 VUNSPEC_MRC2): New.
23766 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23768 * config/arm/arm.md (*ldc): New.
23772 * config/arm/arm.c (arm_coproc_builtin_available): Add
23773 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
23774 (arm_coproc_ldc_stc_legitimate_address): New.
23775 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
23776 'qualifier_const_pointer'.
23777 (LDC_QUALIFIERS): Define to...
23778 (arm_ldc_qualifiers): ... this. New.
23779 (STC_QUALIFIERS): Define to...
23780 (arm_stc_qualifiers): ... this. New.
23781 * config/arm/arm-protos.h
23782 (arm_coproc_ldc_stc_legitimate_address): New.
23783 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
23784 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
23785 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
23786 stc2, stcl, stc2l): New.
23787 * config/arm/constraints.md (Uz): New.
23788 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
23789 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
23790 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
23791 VUNSPEC_STC2L): New.
23793 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23795 * config/arm/arm.md (<cdp>): New.
23796 * config/arm/arm.c (neon_const_bounds): Rename this ...
23797 (arm_const_bounds): ... this.
23798 (arm_coproc_builtin_available): New.
23799 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
23800 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
23801 (CDP_QUALIFIERS): Define to...
23802 (arm_cdp_qualifiers): ... this. New.
23804 (arm_expand_builtin_args): Add case for 6 arguments.
23805 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
23806 (arm_const_bounds): ... this.
23807 (arm_coproc_builtin_available): New.
23808 * config/arm/arm_acle.h (__arm_cdp): New.
23810 * config/arm/arm_acle_builtins.def (cdp): New.
23812 * config/arm/iterators.md (CDPI,CDP,cdp): New.
23813 * config/arm/neon.md: Rename all 'neon_const_bounds' to
23814 'arm_const_bounds'.
23815 * config/arm/types.md (coproc): New.
23816 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
23817 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
23818 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
23819 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
23821 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23823 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
23824 (UBINOP_QUALIFIERS): New.
23826 (acle_builtin_data): New. Change comment.
23827 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
23828 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
23829 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
23830 arm_acle_builtins.def.
23831 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
23832 (arm_init_acle_builtins): New.
23833 (CRC32_BUILTIN): Remove.
23834 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
23835 crc32cb, crc32ch and crc32cw.
23836 (arm_init_crc32_builtins): Remove.
23837 (arm_init_builtins): Use arm_init_acle_builtins rather
23838 than arm_init_crc32_builtins.
23839 (arm_expand_acle_builtin): New.
23840 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
23841 * config/arm/arm_acle_builtins.def: New.
23843 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23845 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
23846 (arm_builtin_datum): ... this.
23847 (arm_init_neon_builtin): Rename to ...
23848 (arm_init_builtin): ... this. Add a new parameters PREFIX
23849 and USE_SIG_IN_NAME.
23850 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
23851 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
23852 'arm_builtin_datum'.
23853 (arm_init_vfp_builtins): Likewise.
23854 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
23855 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
23856 (arm_expand_neon_args): Rename to ...
23857 (arm_expand_builtin_args): ... this. Rename builtin_arg
23858 enum values and differentiate between ARG_BUILTIN_MEMORY
23859 and ARG_BUILTIN_NEON_MEMORY.
23860 (arm_expand_neon_builtin_1): Rename to ...
23861 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
23862 values, arm_expand_builtin_args and add bool parameter NEON.
23863 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
23864 (arm_expand_vfp_builtin): Likewise.
23865 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
23867 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23869 PR middle-end/77484
23870 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
23871 * predict.c (tree_estimate_probability_bb): Reverse direction of
23872 polymorphic call predictor.
23874 2017-01-06 David Malcolm <dmalcolm@redhat.com>
23876 * passes.c (execute_one_pass): Split out pass-skipping logic into...
23877 (determine_pass_name_match): ...this new function and...
23878 (should_skip_pass_p): ...this new function.
23880 2017-01-06 Nathan Sidwell <nathan@acm.org>
23882 * ipa-visibility.c (function_and_variable_visibility): Reformat
23883 comments and long lines. Remove extrneous if.
23884 * symtab.c (symtab_node::make_decl_local): Fix code format.
23885 (symtab_node::set_section_for_node): Fix comment typo.
23887 2017-01-06 Martin Liska <mliska@suse.cz>
23890 * lra-constraints.c: Rename invariant to lra_invariant.
23891 * predict.c (set_even_probabilities): Initialize e to NULL.
23893 2017-01-05 Martin Sebor <msebor@redhat.com>
23895 PR tree-optimization/78910
23896 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
23897 (format_integer): Correct off-by-one error in the handling
23898 of precision with negative numbers in signed conversions..
23900 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
23902 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
23904 2017-01-05 Jakub Jelinek <jakub@redhat.com>
23906 PR tree-optimization/71016
23907 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
23908 factor_out_conditional_conversion. Formatting fix.
23909 (factor_out_conditional_conversion): Add cond_stmt argument.
23910 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
23911 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
23914 2017-01-05 David Malcolm <dmalcolm@redhat.com>
23916 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
23917 read-rtl-function.o, and selftest-rtl.o.
23918 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
23919 (selftest::aarch64_test_loading_full_dump): New function.
23920 (selftest::aarch64_run_selftests): New function.
23921 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
23922 selftest::aarch64_run_selftests.
23923 * config/i386/i386.c
23924 (selftest::ix86_test_loading_dump_fragment_1): New function.
23925 (selftest::ix86_test_loading_call_insn): New function.
23926 (selftest::ix86_test_loading_full_dump): New function.
23927 (selftest::ix86_test_loading_unspec): New function.
23928 (selftest::ix86_run_selftests): Call the new functions.
23929 * emit-rtl.c (maybe_set_max_label_num): New function.
23930 * emit-rtl.h (maybe_set_max_label_num): New decl.
23931 * function.c (instantiate_decls): Guard call to
23932 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
23933 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
23935 * gensupport.c (gen_reader::gen_reader): Pass "false"
23936 for new "compact" param of rtx_reader.
23937 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
23938 rather than an empty string for NULL strings.
23939 * read-md.c: Potentially include config.h rather than bconfig.h.
23940 Wrap include of errors.h with #ifdef GENERATOR_FILE.
23941 (have_error): New global, copied from errors.c.
23942 (md_reader::read_name): Rename to...
23943 (md_reader::read_name_1): ...this, adding "out_loc" param,
23944 and converting "missing name or number" to returning false, rather
23946 (md_reader::read_name): Reimplement in terms of read_name_1.
23947 (md_reader::read_name_or_nil): New function.
23948 (md_reader::read_string): Handle "(nil)" by returning NULL.
23949 (md_reader::md_reader): Add new param "compact".
23950 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
23951 (md_reader::read_file): New method.
23952 * read-md.h (md_reader::md_reader): Add new param "compact".
23953 (md_reader::read_file): New method.
23954 (md_reader::is_compact): New accessor.
23955 (md_reader::read_name): Convert return type from void to file_location.
23956 (md_reader::read_name_or_nil): New decl.
23957 (md_reader::read_name_1): New decl.
23958 (md_reader::m_compact): New field.
23959 (noop_reader::noop_reader): Pass "false" for new "compact" param
23961 (rtx_reader::rtx_reader): Add new "compact" param.
23962 (rtx_reader::read_rtx_operand): Make virtual and convert return
23963 type from void to rtx.
23964 (rtx_reader::read_until): New decl.
23965 (rtx_reader::handle_any_trailing_information): New virtual function.
23966 (rtx_reader::postprocess): New virtual function.
23967 (rtx_reader::finalize_string): New virtual function.
23968 (rtx_reader::m_in_call_function_usage): New field.
23969 (rtx_reader::m_reuse_rtx_by_id): New field.
23970 * read-rtl-function.c: New file.
23971 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
23972 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
23973 (selftest::verify_three_block_rtl_cfg): New decl.
23974 * read-rtl-function.h: New file.
23975 * read-rtl.c: Potentially include config.h rather than bconfig.h.
23976 For host, include function.h, memmodel.h, and emit-rtl.h.
23977 (one_time_initialization): New function.
23978 (struct compact_insn_name): New struct.
23979 (compact_insn_names): New array.
23980 (find_code): Handle insn codes in compact dumps.
23981 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
23982 (bind_subst_iter_and_attr): Likewise.
23983 (add_condition_to_string): Likewise.
23984 (add_condition_to_rtx): Likewise.
23985 (apply_attribute_uses): Likewise.
23986 (add_current_iterators): Likewise.
23987 (apply_iterators): Likewise.
23988 (initialize_iterators): Guard usage of apply_subst_iterator with
23989 #ifdef GENERATOR_FILE.
23990 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
23991 (md_reader::read_mapping): Likewise.
23992 (add_define_attr_for_define_subst): Likewise.
23993 (add_define_subst_attr): Likewise.
23994 (read_subst_mapping): Likewise.
23995 (check_code_iterator): Likewise.
23996 (rtx_reader::read_rtx): Likewise. Move one-time initialization
23998 (one_time_initialization): New function.
23999 (rtx_reader::read_until): New method.
24000 (read_flags): New function.
24001 (parse_reg_note_name): New function.
24002 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
24003 Handle reuse_rtx ids.
24004 Wrap iterator lookup within #ifdef GENERATOR_FILE.
24005 Add parsing support for RTL dumps, mirroring the special-cases in
24006 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
24007 values, and calling handle_any_trailing_information.
24008 (rtx_reader::read_rtx_operand): Convert return type from void
24009 to rtx, returning return_rtx. Handle case 'e'. Call
24010 finalize_string on XSTR and XTMPL fields.
24011 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
24012 "(nil)" values were omitted. Call the postprocess vfunc on the
24014 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
24015 class ctor. Initialize m_in_call_function_usage. Call
24016 one_time_initialization.
24017 * rtl-tests.c (selftest::test_uncond_jump): Call
24018 set_new_first_and_last_insn.
24019 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
24020 * selftest-rtl.c: New file.
24021 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
24022 (selftest::get_insn_by_uid): New decl.
24023 * selftest-run-tests.c (selftest::run_tests): Call
24024 read_rtl_function_c_tests.
24025 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
24026 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
24029 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
24031 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
24032 operands in a special way. Assert that pos+len <= mode precision.
24034 2017-01-05 Jakub Jelinek <jakub@redhat.com>
24036 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
24037 3 argument Alias with unlimited for the negative form.
24038 (fno-vect-cost-model): Removed.
24040 2017-01-05 Martin Liska <mliska@suse.cz>
24042 * hsa-gen.c (gen_hsa_divmod): New function.
24043 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
24045 2017-01-05 Martin Liska <mliska@suse.cz>
24048 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
24051 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24053 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
24054 small constant length operands.
24056 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24058 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
24059 between loop iterations.
24061 2017-01-05 Martin Liska <mliska@suse.cz>
24064 * gimplify.c (gimplify_decl_expr): Compare to
24065 asan_poisoned_variables instread of checking flags.
24066 (gimplify_target_expr): Likewise.
24067 (gimplify_expr): Likewise.
24068 (gimplify_function_tree): Conditionally initialize
24069 asan_poisoned_variables.
24071 2017-01-04 Jeff Law <law@redhat.com>
24073 PR tree-optimizatin/78812
24074 * rtl.h (contains_mem_rtx_p): Prototype.
24075 * ifcvt.c (containts_mem_rtx_p): Move from here to...
24076 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
24077 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
24078 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
24079 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
24081 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
24083 * input.c (assert_char_at_range): Default-initialize actual_range.
24085 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
24087 * df-scan.c (df_ref_create_structure): Make regno unsigned,
24088 to match the caller.
24090 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
24092 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
24093 insns after final jump in test to emit dummy move.
24095 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
24097 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
24098 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
24100 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
24102 * multiple_target.c (create_dispatcher_calls): Init e_next.
24103 * tree-ssa-loop-split.c (split_loop): Init border.
24104 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
24107 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
24112 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
24113 (altivec_register_operand): Do not return true if the operand
24114 contains a SUBREG mixing SImode and SFmode.
24115 (vsx_register_operand): Likewise.
24116 (vsx_reg_sfsubreg_ok): New predicate.
24117 (vfloat_operand): Do not return true if the operand contains a
24118 SUBREG mixing SImode and SFmode.
24119 (vint_operand): Likewise.
24120 (vlogical_operand): Likewise.
24121 (gpc_reg_operand): Likewise.
24122 (int_reg_operand): Likewise.
24123 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
24124 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
24125 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
24127 (rs6000_emit_move_si_sf_subreg): New helper function.
24128 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
24129 fixup SUBREGs involving SImode and SFmode.
24130 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
24131 numbers for the new peephole2 optimization.
24132 (peephole2 for SFmode unions): New peephole2 to optimize cases in
24133 the GLIBC math library that do AND/IOR/XOR operations on single
24134 precision floating point.
24135 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
24136 target macros to say whether we need to avoid SUBREGs mixing
24138 (TARGET_ALLOW_SF_SUBREG): Likewise.
24139 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
24140 (UNSPEC_SI_FROM_SF): Likewise.
24141 (iorxor): Change spacing.
24142 (and_ior_xor): New iterator for AND, IOR, and XOR.
24143 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
24144 (movdi_from_sf_zero_ext): Likewise.
24145 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
24146 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
24147 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
24148 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
24149 (fms<mode>4): Likewise.
24150 (fnma<mode>4): Likewise.
24151 (fnms<mode>4): Likewise.
24152 (nfma<mode>4): Likewise.
24153 (nfms<mode>4): Likewise.
24155 2017-01-04 Marek Polacek <polacek@redhat.com>
24158 * doc/invoke.texi: Document -Wpointer-compare.
24160 2017-01-04 Jakub Jelinek <jakub@redhat.com>
24162 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
24165 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
24166 descriptions for -gdwarf-5 and emit them as uleb128 instead of
24169 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
24172 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
24173 documentation of the powerpc_popcntb_ok attribute.
24174 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
24175 code to issue warning messages if a requested CPU configuration is
24176 not supported by the binary (assembler and loader) toolchain.
24177 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
24178 made to define a built-in function that has been disabled.
24179 (paired_init_builtins): Add assertion to prevent ICE if attempt is
24180 made to define a built-in function that has been disabled.
24181 (altivec_init_builtins): Add comment explaining why definition
24182 of the DST built-in functions is not preceded by an assertion
24183 check. Add assertions to prevent ICE if attempts are made to
24184 define an altivec predicate or an abs* built-in function that has
24186 (htm_init_builtins): Add comment explaining why definition of the
24187 htm built-in functions is not preceded by an assertion check.
24189 2017-01-04 Jeff Law <law@redhat.com>
24191 PR tree-optimizatin/67955
24192 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
24193 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
24194 the points-to solution does not include pt_null. Use DECL_PT_UID
24197 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
24199 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
24200 Use gen_int_mode instead of gen_lopwart for const_int operands.
24202 2017-01-04 Jakub Jelinek <jakub@redhat.com>
24204 PR tree-optimization/71563
24205 * match.pd: Simplify X << Y into X if Y is known to be 0 or
24206 out of range value - has low bits known to be zero.
24208 2017-01-04 Alan Modra <amodra@gmail.com>
24210 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
24211 * configure: Regenerate.
24212 * config.in: Regenerate.
24214 2017-01-04 Jakub Jelinek <jakub@redhat.com>
24217 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
24218 a substring of the message, but strcmp with the whole message. Ifdef
24219 ENABLE_NLS, translate the message first using dgettext.
24221 2017-01-03 Jeff Law <law@redhat.com>
24223 PR tree-optimizatin/78856
24224 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
24225 (mark_threaded_blocks): Remove code to truncate thread paths that
24226 cross multiple loop headers. Instead invalidate the cached loop
24227 iteration information and handle case of a thread path walking
24228 into an irreducible region.
24230 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
24233 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
24234 assertions. Add support for doing the signbit if the IEEE 128-bit
24235 floating point value is in a GPR.
24236 * config/rs6000/rs6000.md (Fsignbit): Delete.
24237 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
24238 Update the length attribute if the value is in a GPR.
24239 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
24240 the sign or zero extension instruction, since the value is always 0/1.
24241 (signbit<mode>2_dm2): Delete using <Fsignbit>.
24244 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
24245 extracting SImode to a GPR register so that we can generate a
24246 store, limit the vector to be in a traditional Altivec register
24247 for the vextuwrx instruction.
24249 2017-01-03 Ian Lance Taylor <iant@google.com>
24251 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
24253 2017-01-03 Martin Sebor <msebor@redhat.com>
24255 PR tree-optimization/78696
24256 * gimple-ssa-sprintf.c (format_floating): Correct handling of
24257 precision. Use MPFR for %f for greater fidelity. Correct handling
24259 (pass_sprintf_length::compute_format_length): Set width and precision
24260 specified by asrerisk to void_node for vararg functions.
24261 (try_substitute_return_value): Adjust dump output.
24263 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
24265 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
24267 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
24269 * doc/invoke.texi (SPARC options): Document -mlra as the default.
24270 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
24271 -mlra/-mno-lra was passed to the compiler.
24273 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
24275 PR rtl-optimization/65618
24276 * emit-rtl.c (try_split): Move initialization of "before" and
24277 "after" to just before the call to emit_insn_after_setloc.
24279 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
24281 * doc/md.texi (Standard Names): Remove reference to Java frontend.
24283 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
24285 * dwarf2out.c (gen_enumeration_type_die): When
24286 -gno-strict-dwarf, add a DW_AT_encoding attribute.
24288 2017-01-03 Jakub Jelinek <jakub@redhat.com>
24290 PR tree-optimization/78965
24291 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
24292 Change first argument from const call_info & to call_info &. For %n
24293 set info.nowrite to false.
24295 PR middle-end/78901
24296 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
24297 possibly throwing calls.
24299 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
24300 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
24301 and fns handling, rather than in a separate case SSA_NAME.
24303 2017-01-02 Jeff Law <law@redhat.com>
24305 * config/darwin-driver.c (darwin_driver_init): Const-correctness
24306 fixes for first_period and second_period variables.
24308 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
24311 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
24312 (*insvqi_1): New insn pattern.
24313 (*insvqi_1_mem_rex64): Ditto.
24314 (*insvqi_2): Ditto.
24315 (*insvqi_3): Rename from *insvqi.
24317 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
24319 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
24321 * doc/cfg.texi (Edges): Remove reference to Java.
24322 (Maintaining the CFG): Ditto.
24324 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24326 PR middle-end/77674
24327 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
24328 transparent aliases.
24330 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24332 PR middle-end/77484
24333 * predict.def (PRED_CALL): Update hitrate.
24334 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
24335 * predict.c (tree_estimate_probability_bb): Split CALL predictor
24336 into direct/indirect/polymorphic variants.
24338 2017-01-01 Jakub Jelinek <jakub@redhat.com>
24340 Update copyright years.
24342 * gcc.c (process_command): Update copyright notice dates.
24343 * gcov-dump.c (print_version): Ditto.
24344 * gcov.c (print_version): Ditto.
24345 * gcov-tool.c (print_version): Ditto.
24346 * gengtype.c (create_file): Ditto.
24347 * doc/cpp.texi: Bump @copying's copyright year.
24348 * doc/cppinternals.texi: Ditto.
24349 * doc/gcc.texi: Ditto.
24350 * doc/gccint.texi: Ditto.
24351 * doc/gcov.texi: Ditto.
24352 * doc/install.texi: Ditto.
24353 * doc/invoke.texi: Ditto.
24355 Copyright (C) 2017 Free Software Foundation, Inc.
24357 Copying and distribution of this file, with or without modification,
24358 are permitted in any medium without royalty provided the copyright
24359 notice and this notice are preserved.